You finally left home after taking a wrong turn, and from that day on, you had a dream of your own—a dream for which you were willing to pay with your life.

Du Hongwei

The title of this essay is shamelessly stolen from the lyrics of the song “Big Brother, How Are You?” (Dage Ni Hao Ma?). I owe its songwriter and composer, Chen Xiaoqi, an apology. Having already adopted Cao Cao and Yuan Shikai as my elder brothers along the way, I see no harm in acquiring one more brother for the sake of stealing a lyric. Brother Chen, please don’t sue me. I’m already dangerously close to begging in the streets.

A few days ago, at the Temple of Yue Fei, at 80 Beishan Road in Hangzhou’s West Lake District, I deposited several pounds of saliva on the kneeling statues of Qin Hui and his wife. Today, accompanied by my good friend Zhou Shifeng, the former director of Beijing’s Fengrui Law Firm, and his wife, I arrived at another Temple of Yue Fei, at 86 Yuemiao Street in Tangyin County, Anyang, Henan Province. My intention was to perform the full ceremonial homage of three kneelings and nine kowtows. Zhou stopped me.

I asked him why he wouldn’t let me bow before Brother Fei, the national hero who had devoted himself so faithfully to serving his country.

“Are you trying to become a rebel?” he asked.

I nearly lost my balance and fell at the temple gate.

Zhou explained that Yue Fei may well have harbored, deep down, thoughts of defying Emperor Gaozong of Song, Zhao Gou—or at least that was how Zhao Gou might have seen him. From the emperor’s perspective, killing Yue Fei to safeguard his own throne was therefore not entirely incomprehensible. Later generations of rulers, however, had their own political reasons for turning Yue Fei into an immaculate hero of resistance against the Jurchen Jin dynasty.

Otherwise, Zhou asked, why was Yue Fei so determined to bring back the captive emperors Huizong and Qinzong? Why was he so reluctant to withdraw even after Zhao Gou reportedly sent twelve urgent orders recalling him? Why was his army known as the “Yue Family Army”? And why was a military commander so concerned with whether the emperor should bring his elder brother home?

I listened carefully. There seemed to be something in his argument. Then I remembered a visit some years earlier to the former residence of Wang Chuanshan—Wang Fuzhi—in Liulan, on the western bank of the Xiang River in Hengyang County, Hunan. I seemed to recall encountering another interpretation of the relationship between Yue Fei and Emperor Gaozong.

Wang Chuanshan’s assessment of Yue Fei in his Commentaries on the Song Dynasty (Song Lun) was, in fact, quite complicated. He acknowledged Yue Fei’s military abilities, moral character, and achievements, while also recognizing the political tension between Yue Fei and Emperor Gaozong. Yue wanted to recover the Central Plains; Zhao was perpetually anxious about preserving the throne that had so unexpectedly fallen into his lap. One wanted to march north and recover the lost territories; the other wanted to secure his throne in the south. Their political purposes could hardly have been further apart.

To suggest that Yue Fei intended to destroy the Jin dynasty and then turn his armies against the Southern Song, however, goes much too far. The historical record provides no evidence of such a plan. But that does not mean Zhao Gou never suspected one. Emperors are peculiar creatures. Those who have narrowly escaped death and unexpectedly acquired a throne are especially gifted at imagining what other people might be thinking—even when those thoughts have never crossed the other person’s mind.

The history of the Five Dynasties and Ten Kingdoms certainly gave Zhao Gou plenty of reasons to be haunted by such possibilities. The succession of the Later Liang, Later Tang, Later Jin, Later Han, and Later Zhou dynasties was intimately connected with powerful military commanders seizing the throne. Generals accumulated armies, emperors were forced aside, and dynasties changed as though in a revolving-door farce. Hadn’t the Zhao family’s own imperial fortunes begun with the mutiny at Chenqiao, which brought Zhao Kuangyin to power? Sitting on the Southern Song throne and looking back over the previous century or two, how could Zhao Gou not have been frightened?

Seen in this light, the notorious phrase mo xu you, traditionally associated with the dubious charges against Yue Fei, may have concealed something very real in Zhao Gou’s mind, even if there was no evidence of treason. That something was fear. Fear needs no evidence. Imagination will do. Whether Yue Fei had ever contemplated rebellion could, in the emperor’s mind, become almost beside the point. I once wrote an essay exploring this very argument, titled “The Clear Wind Would Not Let Me Stay” (Qing Feng You Yi Nan Liu Wo).

To understand why Zhao Gou came to fear Yue Fei, and how that fear eventually destroyed their relationship, we must first turn our attention to the emperor himself.

Emperor Gaozong of Song, Zhao Gou (1107–1187), courtesy name Deji, was the founding emperor of the Southern Song dynasty. He was the ninth son of Emperor Huizong and the younger brother of Emperor Qinzong. Originally titled Prince of Kang, he later received the temple name Gaozong. He lived to eighty-one by traditional Chinese age reckoning, an exceptionally long life for a Chinese emperor. He reigned for thirty-five years and spent another twenty-five as retired emperor.

During the Jingkang catastrophe, Zhao Gou was among the few important members of the Song imperial family fortunate enough to escape capture by the Jin forces. When the Jin army besieged Kaifeng in the first year of Jingkang, Zhao had already undertaken a mission to the Jin camp. Later, ordered on another diplomatic mission, he reached Cizhou, where Zong Ze and others persuaded him not to continue north. That decision left him with the most valuable political asset he would ever possess: he was still alive, and he was both Huizong’s son and Qinzong’s younger brother.

Huizong and Qinzong were captured. The Northern Song collapsed.

In 1127, Zhao Gou ascended the throne at Yingtian Prefecture, present-day Shangqiu in Henan, reestablishing the Song dynasty in what became known as the Southern Song. His reign began under the era name Jianyan. What followed was an extended flight for survival, including a period at sea. As the Jin armies pursued him in a campaign remembered by the phrase “search the mountains and scour the seas for Zhao Gou,” he fled from Shangqiu to Yangzhou, then moved through Jiankang and Hangzhou. At the height of the crisis, he escaped aboard a ship and took refuge at sea. He eventually established his imperial residence at Lin’an, present-day Hangzhou, which functioned as the Southern Song’s capital.

Having established his government, Zhao Gou pursued two policies at once. He appointed the generals who would later be celebrated for the dynasty’s revival, while simultaneously contemplating peace with the Jin and a secure existence south of the Yangtze. He knew the Song dynasty’s chronic weaknesses all too well. Ever since Zhao Kuangyin had reclaimed military power from his generals over a cup of wine, the dynasty’s military strength had steadily deteriorated. Zhao Gou also understood that a direct confrontation with the Jin commander Wuzhu might not end in victory. In the early years, therefore, he had little choice but to rely on Yue Fei, Han Shizhong, Zhang Jun, Liu Guangshi, and other generals to resist the Jin armies and secure the south.

In 1133, the third year of Shaoxing, Zhao Gou personally wrote the four characters Jing Zhong Yue Fei—“Yue Fei, Utterly Loyal”—and had them made into a banner for the general. It was a public expression of imperial approval. The familiar story of Yue Fei’s mother tattooing the words Jing Zhong Bao Guo, “Serve the Country with Utmost Loyalty,” onto her son’s back comes primarily from later theatrical adaptations, popular romances, and folklore. The History of Song records that Yue Fei bore the characters “Jin Zhong Bao Guo” (“Serve the Country with Utmost Loyalty”) on his back, but no reliable contemporary evidence establishes that his mother herself took a needle to his skin.

Unless, of course, Yue Fei’s mother was some kind of sadistic lunatic who decided to carve four large, black characters—“Serve the Country with Utmost Loyalty”—into her young son’s tender back, proclaiming a patriotic obligation that had precisely nothing to do with the daily life of an ordinary family. The modern internet parody of the story, in which the mother tattoos “Don’t Go Back to China” instead, sounds rather more consistent with the experience of certain people today.

When Zhao Gou wrote those four characters praising Yue Fei’s loyalty, we cannot assume that he had already decided the general would eventually have to die. Yet the inscription contained a marvelous political paradox: the more an emperor demands absolute loyalty from his subjects, the more determined he must be to ensure that their loyalty serves only objectives he himself has approved.

As the saying goes, people are forever advertising the things they lack. Only a society suffering from disharmony needs to launch a campaign to build a “harmonious society.” If society were genuinely harmonious, why the hell would anyone need to keep preaching harmony? By the same logic, the larger an emperor writes the word loyalty, the less certain we can be that he actually trusts his officials. Perhaps he needs them to prove their loyalty again and again. Perhaps he has already begun to suspect that something sinister lies beneath their declarations of devotion.

In the old story, the bandit chief Zuoshandiao insists that “Old Nine must not leave.” But Zhao Gou’s own identity as the ninth son had originally made the imperial throne an exceedingly remote prospect. Since the gradual consolidation of the patriarchal succession system in the Zhou period, succession through the eldest son of the principal wife had become a major principle of dynastic politics. The classical formula was that among sons of the principal wife, seniority took precedence over merit; in choosing among other sons, the mother’s status took precedence over age. Although imperial succession did not always follow these rules, they remained an important foundation of hereditary political order.

Zhao Gou was neither the son of the principal empress nor the eldest son. He ranked ninth. Had everything proceeded normally in the Northern Song imperial family, he would probably have spent his entire life as Prince of Kang. But the Jingkang catastrophe smashed the established succession order to pieces, giving him an opportunity that otherwise would scarcely have existed. His first debt of gratitude was to Zong Ze for persuading him not to proceed into danger. His second was to his own extraordinary luck in avoiding capture. It was as though he had won two different lotteries on the same ticket.

After finally becoming Emperor Gaozong, Zhao Gou must have felt that, if the folk legend of the clay horse carrying Prince Kang across the river were true, he ought to spend every New Year’s Day kowtowing before that miraculous animal in the ancestral temple.

For a ruler whose throne had been acquired under such extraordinary circumstances, one of the most unwelcome possibilities was the return of the captive emperors Huizong and Qinzong. The two former rulers were living as prisoners in the north. Their circumstances could hardly be described as a pleasant opportunity to contemplate their surroundings and sing patriotic songs. But as long as they remained there, Zhao Gou’s throne was spared at least one immediate complication.

In Zhao Gou’s imagined private calculations, why shouldn’t the two former emperors stay up north, sip a little Longjing tea, and swap stories at the bottom of their well? Why on earth was Yue Fei so determined to bring them back to Hangzhou? Did he think the Longjing tea Zhao Gou had sent them was not authentic tea from the five celebrated growing districts—Shi, Long, Yun, Hu, and Mei? And what was all that arrogant shouting about marching straight to Huanglong Prefecture? Huanglong means “Yellow Dragon,” and who was the dragon now if not His Imperial Majesty? Was Yue Fei secretly threatening to smash the emperor himself?

Huanglong Prefecture, of course, had nothing to do with Zhao Gou personally. This entire passage is my invention, an exercise in imagining the emperor’s thoughts. Still, a ruler perpetually anxious about his throne might well hear hidden meanings in even the most heroic military declarations.

In his earlier years, Yue Fei had indeed repeatedly expressed a desire to bring back the two captive sovereigns. Huizong subsequently died in the north, but Qinzong remained alive. Recovering the two emperors was originally a political objective shared by many advocates of resistance to the Jin. It represented the desire to restore lost territory and avenge the humiliation of Jingkang. For Zhao Gou, however, the slogan carried a special political sensitivity.

What would happen to the throne if Qinzong returned?

Qinzong might not necessarily have demanded restoration to power. But his very existence could become a political variable in the Southern Song court. More troubling still, Yue Fei commanded an army and enjoyed considerable prestige. What if someone invoked Qinzong’s name to challenge Zhao Gou’s legitimacy? What was the reigning emperor supposed to do?

And who could guarantee that Yue Fei would not use Qinzong to force Zhao Gou from the throne, install the former emperor as a puppet, and eventually perform the familiar trick of imperial abdication once practiced by Cao Pi and Sima Yan? Did Yue Fei really imagine that his little schemes could escape Zhao Gou’s all-seeing eyes? The lessons of history were hardly distant. The Five Dynasties and Ten Kingdoms were right there in the historical record. Did anyone think Zhao Gou did not understand how ambitious generals played the game?

Again, this is an imagined reconstruction of Zhao Gou’s suspicions. There is no historical evidence that Yue Fei actually devised such a plan. But the most frightening feature of political fear is precisely that it does not require its supposed object to act. Once an emperor believes that you might do something, and that you possess the means to do it, you are already in danger. Zhu Yuanzhang’s executions of Li Shanchang, Hu Weiyong, and Lan Yu during the Ming dynasty can also be understood in terms of this kind of political psychology. Sorry, old friend. For the sake of my descendants, you’ll just have to endure a little unpleasantness.

What makes the story even more absurd is that Yue Fei did not cheerfully withdraw upon receiving the famous twelve gold-plaque orders. According to the History of Song (Song Shi), in its biography of Yue Fei, when the general received an order to retreat, he submitted a memorial protesting the decision. The Jin forces, he argued, were preparing to abandon their baggage and flee rapidly across the river. Meanwhile, influential men throughout the region were rallying to the Song cause, and the soldiers were fighting with determination. The balance of strength had become clear. Victory was within reach, the opportunity might never come again, and such a moment could not lightly be thrown away.

In plain language: the enemy was about to run, the people of the Central Plains were responding to the Song advance, and a once-in-a-lifetime opportunity was at hand. Do not withdraw now.

Zhao Gou would not be persuaded. The History of Song records that Yue Fei received twelve gold-lettered dispatches in a single day, ordering him to withdraw. The general was devastated, but he obeyed. More consequentially, the court undermined Yue Fei’s strategic position by recalling the armies of Zhang Jun, Han Shizhong, Liu Qi, and other commanders. The Yue Family Army was suddenly isolated. Without allied forces protecting its flanks, it risked encirclement and destruction if the Jin concentrated their strength against it. With the other Song armies withdrawing, Yue Fei’s force was left advancing alone into Henan. Even if it defeated the Jin, an open break with the Southern Song court could leave it fighting both the Jin and the Song, with enemies before and behind. The danger would be immense.

There was also the problem of supplies. Despite its popular name, the “Yue Family Army” was not Yue Fei’s private military property. Its provisions depended on transportation from the rear. If Yue defied imperial orders, the court could cut off food and military funds. An army said to number one hundred thousand could hardly sustain itself in the Central Plains indefinitely without supplies.

So, however unwillingly, Yue Fei ultimately obeyed the order to withdraw. The story of the twelve gold plaques, far from proving that the emperor had lost control of his general, demonstrated the emperor’s supreme authority to recall a commander from the front—and his practical capacity to enforce that authority. Yet from the emperor’s perspective, even a general who eventually obeyed might remain unforgivably dangerous.

And this is the point: Yue Fei submitted to the imperial command, but Zhao Gou still did not spare him.

If obedience itself could no longer establish loyalty, what else was Yue Fei supposed to do to put the emperor at ease? Surrender his military command? Keep his mouth shut? Dig the idea of recovering the Central Plains out of his own skull? Or perhaps dig himself a grave, climb inside, and save the emperor the trouble?

The most important question raised by Yue Fei’s death, therefore, is not simply whether he intended to rebel. It is why Zhao Gou came to regard a general against whom there was no conclusive evidence of rebellion, and who ultimately obeyed the order to withdraw, as a political threat that had to be eliminated.

Let us now consider another question. Why did Emperor Xiaozong of Song, Zhao Shen, dare to rehabilitate Yue Fei and restore his reputation while Zhao Gou was still alive?

By then Zhao Gou had abdicated and become retired emperor. Because he had no surviving biological son to inherit the throne, Zhao Shen had been designated imperial heir and eventually succeeded him. Zhao Shen descended from Zhao Kuangyin, Emperor Taizu, while Zhao Gou belonged to the line of Zhao Guangyi, Emperor Taizong. Their blood relationship was distant enough to fall outside the ordinary close kinship categories recognized in traditional Chinese society. Politically and dynastically, however, their relationship was clearly that of adoptive father and son.

This distinction mattered enormously. Zhao Shen owed the legitimacy of his throne to Zhao Gou’s selection and abdication. He could not simply become emperor and immediately denounce his adoptive father as an incompetent tyrant. More important, Xiaozong had a strong desire to recover the Central Plains. If he intended to revive a policy of military resistance, Yue Fei was a historical figure he could scarcely avoid.

Historical records show that Zhao Gou abdicated in June 1162. The following month, Xiaozong issued an edict restoring Yue Fei’s former official titles, arranging a proper reburial, and ordering officials to locate his descendants and consider them for appointment. This was the decisive first step in Yue Fei’s political rehabilitation.

One fact is crucial: Zhao Gou was still alive. He was the retired emperor, and he would live until 1187.

To preserve his adoptive father’s dignity, Xiaozong used exceptionally careful language. The restoration of Yue Fei’s honors was presented as an act undertaken in accordance with the retired emperor’s wishes. The edict did not bluntly declare that Zhao Gou had wrongfully executed a loyal general. It was an exercise in political compromise.

Translated into contemporary language, Xiaozong’s public explanation was something like this: Yes, I am the one carrying out this decision, but my father wanted it all along. I am merely fulfilling the old gentleman’s wishes.

The arrangement was ingenious. The earlier decision would not be publicly judged; the new decision would simply honor a loyal servant of the dynasty. As for why that loyal servant had been executed in the first place, it would be best if nobody asked too many questions. The wording was politically astute. By presenting the rehabilitation as the fulfillment of the retired emperor’s wishes, Xiaozong restored Yue Fei’s honor without openly repudiating the ruler who had condemned him. Whether Zhao Gou welcomed the decision is another question.

Now that was political language at its finest. Xiaozong restored Yue Fei’s reputation without delivering a public slap across Zhao Gou’s face.

Had Xiaozong simply issued an edict declaring, “Gaozong was deceived by treacherous ministers and unjustly executed the loyal Yue Fei,” he would have openly condemned Zhao Gou. The retired emperor could then have intervened, expressed his displeasure, or used his continuing influence to exert pressure. The two men had barely completed the transfer of power. Were they now going to tear the Southern Song court apart over a general who had been dead for twenty years? How was anyone supposed to run the country? And when would the gentlemen of the palace find time to enjoy the beautiful women of the imperial household?

Yue Fei’s later posthumous title, Wumu, belongs to a separate stage in this history. The title was formally settled during Xiaozong’s reign, around 1178–1179, more than a decade after the initial rehabilitation in 1162. Court officials had originally proposed the posthumous title Zhongmin, combining loyalty with an element of grief and pity. The character min could suggest suffering and injustice, thereby drawing attention to the wrongful conviction. To avoid offending Zhao Gou, the court ultimately settled on Wumu. That is how Yue Fei became known as Yue Wumu. Xiaozong’s public formula remained the same: the decision was made in accordance with the retired emperor’s wishes.

This does not alter the larger point. Yue Fei’s official historical standing was not transformed overnight. Restoring his titles, establishing temples and commemorative rites, and formally conferring a posthumous title were separate steps. At each stage, the court adjusted its account of the past in response to the political circumstances of the present.

Several additional factors deserve consideration.

The Shaoxing peace settlement had brought the Southern Song a period of relative stability, but it had not eliminated the conflict between Song and Jin. In 1161, the Jin ruler Wanyan Liang, known as the Prince of Hailing, launched a major invasion of the south. His campaign made clear that the peace settlement could not permanently remove the military threat. After ascending the throne, Xiaozong hoped to restore military vigor and recover the Central Plains. Yue Fei’s image naturally possessed considerable value as a symbol of political mobilization.

Sympathy for Yue Fei had not entirely disappeared among ordinary people or the scholar-official elite. His military achievements, his wrongful imprisonment, and Qin Hui’s role in negotiating peace remained subjects of historical and political controversy. Restoring Yue Fei’s reputation allowed Xiaozong both to respond to a strand of public opinion and to strengthen military morale.

Most important, Xiaozong himself favored resistance to the Jin. He aspired to recover the Central Plains and needed exemplary loyal officials to inspire the court and the army. A celebrated general executed under the previous administration was ideally suited to become a symbol of the new emperor’s political direction.

Thus, Yue Fei’s execution in 1142 and his rehabilitation in 1162 were not two judgments made under the same political conditions. The first occurred when Zhao Gou sought to preserve the peace settlement and control military power. The second came when Xiaozong was preparing to revise policy and revive the cause of recovering the Central Plains.

The resulting situation was almost unbearably ironic. One emperor had Yue Fei killed for his political purposes. Another restored Yue Fei’s reputation for his own political purposes. Yue Fei himself had been dead for twenty years, yet he was once again an active participant in the politics of the imperial court. The life and death of a subject seemed to depend entirely on the political needs of the emperor.

Whether Yue Fei was wrongfully convicted is, of course, a question history must continue to examine.

But for an emperor, the handling of an injustice depends first on who happens to be sitting on the throne that day. The emperor’s needs come first. As for whether a subject was wronged, the emperor may not give a rat’s ass.

There is another aspect of Yue Fei’s death that deserves particular attention: who bears political responsibility?

Later narratives of Yue Fei’s wrongful conviction usually cast Qin Hui as the great villain. Qin Hui pursued negotiations with the Jin, participated in the political campaign against Yue Fei, and helped bring about his imprisonment through political and judicial means. He unquestionably bears substantial responsibility. There is no reason to whitewash his conduct.

But however influential Qin Hui may have been, he was still the emperor’s chief minister—a servant of the throne. Could a chancellor simply dispose of a senior general commanding substantial forces and enjoying enormous public prestige, entirely without the emperor’s knowledge and against his express wishes?

If Zhao Gou had firmly opposed Yue Fei’s execution, was Qin Hui going to steal the imperial seal, forge an edict, and personally order the death of one of the dynasty’s leading generals?

Qin Hui helped engineer the prosecution and bore substantial responsibility for Yue Fei’s death. But he was still the emperor’s minister. Could a chancellor have brought down one of the dynasty’s most prominent generals in open defiance of the sovereign’s wishes? The familiar story of the wicked minister and the innocent emperor is politically convenient. It assigns responsibility to the man who carried out the decision while allowing the man who held ultimate authority to escape scrutiny.

This is one of the most fascinating features of traditional Chinese political narratives: the emperor is always wise, and every disaster is the work of treacherous ministers. When the emperor wants someone killed, a villainous minister performs the deed. When later generations seek rehabilitation, the same minister bears the blame. At worst, the emperor was misled, momentarily confused, or guilty of one unfortunate lapse in judgment—as though a little water had somehow leaked into the imperial brain.

The later installation of kneeling statues of Qin Hui at Yue Fei’s temple, and the popular custom of spitting at them, had the effect of diminishing Zhao Gou’s responsibility in the familiar morality tale of loyal ministers and treacherous officials. Qin Hui must answer for his own actions. But to place the entire blame for Yue Fei’s death on him, while allowing Zhao Gou to stand aside as an innocent spectator, is to let the Southern Song emperor off remarkably lightly.

Qin Hui was hardly an innocent man caught in the crossfire. Even so, he and his wife have spent centuries taking hundreds of billions of tons of saliva in the face on Zhao Gou’s behalf. That much is beyond dispute.

Consider, too, Zhao Gou’s state of mind in old age.

As a young ruler, Zhao Gou had fled before the Jin armies all the way to the sea. His fear of the Jin was profound, and his preference for negotiated peace and a secure southern regime remained central to his political choices. By the time he became retired emperor, however, two decades had passed since the Shaoxing settlement. Both the external situation and his personal political circumstances had changed.

He no longer had to make the ultimate decisions on military and state affairs. He could retire to Deshou Palace, enjoy his honors, drink wine from women’s shoes, and amuse himself with dice games. His position in old age and his posthumous reputation mattered increasingly to him. He might not have wished to provoke an open confrontation with his newly enthroned adopted son over a general who had been dead for many years, particularly if doing so risked disorder at court.

Indeed, had the retired emperor intervened forcefully to prevent Yue Fei’s rehabilitation, he might have generated criticism among officials and the wider public, damaging his own reputation.

The balance of power also mattered. Although Xiaozong was an adopted son, his legitimacy derived from Zhao Gou, and he continued to treat the retired emperor with marked deference. Decisions on military and state affairs had to take account of the attitude at Deshou Palace. Xiaozong understood the need for restraint and consistently maintained a posture of respect toward Zhao Gou. The retired emperor, for his part, accepted some of the new emperor’s political arrangements. The two maintained a relatively stable relationship between retired sovereign and reigning monarch. Yue Fei’s rehabilitation did not produce an open rupture.

There is also an important chronological detail.

By the time Yue Fei was rehabilitated in 1162, Emperor Qinzong, Zhao Huan, was already dead.

Qinzong died in Jin territory in 1156. The Southern Song court did not formally learn of his death until 1161, when it conferred the appropriate posthumous honors. Yue Fei’s rehabilitation took place in July 1162. By then, both in fact and in the understanding of the Southern Song court, Qinzong was no longer alive.

The timing is revealing.

When Yue Fei was executed in 1142, Qinzong was still alive. We cannot establish that Zhao Gou ordered Yue Fei’s death primarily to prevent Qinzong’s return. Nevertheless, the existence of a living former emperor could reasonably have figured among Zhao Gou’s concerns about the security of his throne.

By the time Xiaozong rehabilitated Yue Fei in 1162, the succession problem posed by the possible return of Qinzong had disappeared. On that particular point, at least, Yue Fei’s historical reputation no longer directly implicated the possible restoration of a living former sovereign.

This does not mean that Xiaozong dared to rehabilitate Yue Fei only because Qinzong had died. The new political direction, military circumstances, public opinion, and balance of authority between adopted father and son provide more immediate explanations.

The same chronology helps explain why the political language of Xiaozong’s later Longxing Northern Expedition differed from that of Yue Fei’s campaigns.

With both Huizong and Qinzong dead, the call to bring back the two captive sovereigns no longer had any practical meaning. Xiaozong’s political objectives increasingly centered on recovering lost territory, avenging the humiliation of Jingkang, and restoring the prestige of the Song dynasty.

The emphasis shifted from rescuing two captive emperors to recovering the Central Plains and completing the unfinished work of the dynasty’s ancestors. Xiaozong could thus raise the banner of a northern expedition while avoiding an old slogan that might reopen awkward questions about Zhao Gou’s imperial legitimacy.

In the end, Yue Fei remained Yue Fei. The four characters associated with his patriotic ideal—Huan Wo He Shan, “Restore My Rivers and Mountains”—still referred to the same rivers and mountains, unchanged by so much as an inch. What had changed was the man sitting on the throne, and what that man needed his subjects to believe.

On the basis of this analysis, Zhou Shifeng’s suggestion that Yue Fei was a rebel is not entirely devoid of political logic worth examining. To claim that Yue Fei actually plotted rebellion, however, goes well beyond the available historical evidence. To suggest that Zhao Gou might have perceived Yue Fei as a potential rebel is a different matter, and one deserving serious study.

We must distinguish two questions: whether Yue Fei actually wanted to overthrow Zhao Gou, and whether Zhao Gou believed that Yue Fei possessed the capacity to threaten his throne.

The first is a question of evidence. The second concerns the political psychology of an emperor. An emperor can kill because he is afraid. But the fact that he has killed someone cannot then be used to prove that his fear was justified.

Seen through Zhao Gou’s suspicious eyes, even the celebrated lyric Man Jiang Hong (“The River Runs Red”), traditionally attributed to Yue Fei, could acquire an entirely different meaning. Consider its famous lines:

“Thirty years of glory—dust and earth;
Eight thousand miles of travel—cloud and moon.
Do not idle away your youth,
Lest your hair turn white and regret come too late.”

In Zhao Gou’s imagined interpretation, Yue Fei was saying that the honors of a powerful military commander meant no more to him than dust and earth. What he really wanted was to become emperor. Worse still, he was in a terrible hurry to realize that ambition, afraid that youth would vanish and his hair turn white before he could seize the throne.

And so the famous cry, “When shall the subjects’ hatred be extinguished?” might, in Zhao Gou’s reading, have become something altogether different: “When shall the subjects’ ambitions be extinguished?”

This is the most terrifying implication of mo xu you.

Whether a person has committed a crime ought to be determined by evidence. Under imperial autocracy, however, whether a person is considered dangerous may depend on how the emperor imagines that person’s future.

From this perspective, whether Yue Fei was a loyal minister or a rebel did not always depend on what Yue Fei himself had done. It could depend on where the emperor stood and the standards by which he chose to judge him.

When resistance to the Jin was needed, Yue Fei was the loyal general who served his country with unwavering devotion. When peace was desired, he became an inconvenient obstacle. When a new emperor prepared to renew the northern campaigns, Yue Fei once again became a loyal servant worthy of commemoration.

Yue Fei died at only thirty-nine. How could he have imagined how much work would still be waiting for him after death? One day he had to serve as proof of imperial loyalty; the next, as evidence of suspected treason. Twenty years later, he was required to demonstrate loyalty all over again for a new emperor. During his lifetime he had worked himself to exhaustion, like some miserable, useless pubic hair forever stranded at the threshold, never allowed through the door. After his death, he still had to go on working for the Zhao family.

Perhaps, then, the most exhausting job in Chinese history was not necessarily being an emperor. It may have been serving as a loyal minister whom successive emperors could put to work again and again.

While visiting the Temple of Yue Fei, I encountered a retired official who offered an observation:

“The day the kneeling statue of Qin Hui outside the Yue Fei Temple in Hangzhou is replaced by a statue of Zhao Gou, the Chinese people will have made progress.”

He was, in fact, only partly right.

Replacing Qin Hui with Zhao Gou would at least suggest that people had begun to question the responsibility of the supreme ruler, rather than contenting themselves with blaming everything on treacherous ministers. But if the result were merely to exchange Qin Hui for Zhao Gou, while continuing to kneel before another emperor—or to choose a different historical figure as the new target of public spitting—it would amount to little more than rearranging the ranking of villains. Without questioning the system that made such an injustice possible, replacing one villain with another would amount to little more than changing the names on a monument. The stupidity would remain intact.

Replacing Qin Hui with Zhao Gou might represent a tiny advance in the Chinese public’s intellectual understanding. But only a tiny one. It would still leave us well within the territory of stupidity.

For the question that truly matters is not whether Qin Hui or Zhao Gou should be kneeling outside the temple. It is why an emperor could determine a subject’s life or death on the basis of political suspicion; why a man against whom there was no conclusive evidence of treason could be executed; why, twenty years later, another emperor could simply declare that same man a loyal servant; and why historical justice must so often wait for power to change hands before an old account can be reopened.

If these questions remain unanswered, replacing Qin Hui with Zhao Gou, then replacing Zhao Gou with one of his ancestors, and eventually casting bronze statues of eighteen generations of the Zhao family to line them up outside the temple, would accomplish very little.

The bronze foundries of Hangzhou might receive a few additional orders. The temple administrators would have to consider improving the drainage system to prevent visitors’ saliva from flooding the entrance.

That would be about it.

As I was leaving, I passed through the Shrine of Filial Devotion and saw an elderly woman kneeling before the statue of Yue Yinping, bowing repeatedly and murmuring prayers. She appeared to be asking the young woman to bless her with a long life. I nearly burst out laughing.

According to popular legend, Yue Yinping died at the age of thirteen. Unable to obtain justice for her father, she threw herself into a well while holding a silver vase. Later generations honored her as the Filial Daughter, or Lady Yinping. Whether Yue Yinping actually existed is not established by the historical record. Her story belongs primarily to folklore and subsequent religious devotion.

But the old lady was plainly unconcerned with such distinctions. She simply kept bowing, with extraordinary sincerity.

Of all the people to whom she might have prayed for longevity, why choose a girl who had lived only thirteen years? Was the old woman worried that she herself had already lived too long?

If she was determined to bang her head against the ground as though pounding garlic, she might at least have directed her devotion toward Peng Zu, the legendary sage said to have lived eight hundred years.

Then, having pounded all that garlic before noon, she could have put it to good use in a bowl of one of Henan’s favorite dishes: noodles tossed with garlic sauce.

Ah, the Chinese! Even when they kowtow, they can’t seem to find the right thing to bang their heads against.

Subscribe now

In January 2025, the National Academies of Sciences, Engineering, and Medicine (NASEM) published a report that deserves serious attention from educators in both the United States and China: Transforming Undergraduate STEM Education: Supporting Equitable and Effective Teaching. Drawing on decades of educational research, the 388-page report reaches a consequential conclusion: although American universities devote substantial resources to educating students in science, technology, engineering, and mathematics, teaching practices shown by research to be effective still reach only a relatively small proportion of students. The problem extends beyond the competence of individual instructors to curriculum design, university governance, faculty incentives, and the institutional conditions that shape students’ educational opportunities.

What distinguishes the report is its refusal to attribute the difficulties of American STEM education simply to students who do not work hard enough or teachers who are not sufficiently capable. Nor does it equate better science education with building more laboratories, acquiring sophisticated equipment, or increasing enrollment. Instead, it asks a more fundamental question: Why, after decades of research have established so much about how people learn, have American universities struggled to put that knowledge into everyday practice?

China has equally compelling reasons to consider this question. It has an enormous population of science and engineering students, produces engineering graduates on a vast scale, and has established comprehensive systems of education and research across numerous technological fields. Yet to what extent have expanding enrollment, improved examination performance, and rising numbers of scientific publications been accompanied by gains in students’ capacity for original discovery, independent judgment, and scientific inquiry?

America’s problem is that a university system long committed to individual choice and academic autonomy has not consistently provided effective science education for all its students. China’s challenge may be different: its strengths in institutional coordination and standardized training can encourage an excessive concentration on measurable outcomes—examination scores, university admissions, and academic output.

The two countries differ profoundly in their educational traditions and institutional arrangements. Yet they face a common challenge: when mastering established knowledge is no longer enough, can their schools cultivate people capable of discovering problems, creating knowledge, and exercising independent judgment?

The timing of the report is especially significant. It appeared as generative artificial intelligence was rapidly transforming how knowledge is produced and disseminated. As machines become increasingly capable of explaining concepts, solving problems, writing computer programs, and assisting scientific research, the traditional educational model built around the transmission of knowledge is coming under pressure. American universities must confront this change. Chinese schools cannot avoid it either.

Why Scientific Excellence Has Not Produced Consistently Excellent Teaching

The United States is home to some of the world’s leading research universities. Its institutions have produced numerous Nobel laureates, and the country possesses mature systems of scientific research, university autonomy, and technological entrepreneurship. It is tempting to conclude that American university science education must be equally advanced.

The NASEM report reveals another side of the picture. Research excellence does not automatically translate into effective undergraduate teaching. A distinguished physicist may be an outstanding researcher without necessarily knowing how to teach physics to students with different levels of preparation. A university may attract enormous research grants without ensuring that its first-year students receive an effective introduction to fundamental scientific concepts.

Part of this contradiction arises from the incentive structures of American research universities. Publications, research funding, and academic reputation are closely tied to faculty advancement. Teaching quality, by contrast, is harder to measure and often receives less recognition. Universities consequently place considerable institutional value on discovering knowledge without always giving comparable attention to how that knowledge is understood by the next generation.

The report finds that approaches relying heavily on lectures, reading, memorizing formulas, and repetitive calculations are generally less effective than those that actively engage students in disciplinary practices and encourage them to apply knowledge to real problems. Introductory courses are especially consequential: their design and delivery may lead students to abandon STEM studies prematurely, even when they possess considerable potential for further development.

This points to an educational paradox. Introductory courses are supposed to help students enter the world of science, yet some function as mechanisms of selection and exclusion. When students encounter calculus, physics, or programming for the first time and struggle with established teaching methods, they may conclude that they are simply unsuited to scientific work.

That conclusion may be mistaken. Their difficulties may reflect inadequate preparation, but they may also reveal shortcomings in instruction, curriculum design, or academic support.

Scientific education must, of course, maintain rigorous standards. Universities should not lower professional requirements merely to increase course completion rates. The important distinction is between students who have not yet acquired essential knowledge and students whose capacity to acquire it has been unnecessarily constrained by the way they are taught.

Science education must assess professional competence without allowing its teaching methods to become avoidable barriers to learning. This principle matters in the United States, and it matters just as much in China.

The Report’s Central Insight: Teaching Reform Requires Institutional Reform

The NASEM report identifies seven principles of equitable and effective STEM teaching: actively engaging students in learning; recognizing the knowledge and experiences they bring to the classroom; understanding the emotional and social dimensions of learning; attending to identity and belonging; using multiple forms of evidence to improve instruction; adapting teaching to specific contexts; and providing equitable opportunities through clear objectives and transparent assessment.

The report also distinguishes among these principles according to the strength of the supporting evidence. The first five have a relatively substantial foundation in learning-science research, while the evidence supporting the final two is less extensive. This caution is worth emphasizing. Proposing a plausible educational principle is one thing; demonstrating that it actually improves learning outcomes is another.

Taken together, the principles address three interconnected questions: how students learn, what conditions enable different students to learn effectively, and how institutions can sustain improvements in teaching.

The first concerns active participation in knowledge formation. Learning science means more than remembering scientific conclusions. It requires understanding how those conclusions were reached, why they are considered valid, and under what circumstances they might be overturned. A physics student must be able to apply Newton’s laws accurately, but should also understand how those laws emerged and how experimental evidence has tested their validity.

The second concerns the learner. Students’ experiences, interests, identities, emotional states, and social relationships influence their education. Schools must therefore look beyond the number of questions students answer correctly. They must consider whether students understand their learning objectives, feel able to raise questions, and can persist after failure. Attention to belonging should help capable students overcome avoidable barriers, while leaving academic standards intact.

The third concerns the institution. Even an excellent teacher may struggle to sustain effective instruction if the surrounding university fails to support it. The report therefore calls for collective reviews of curricula and prerequisites, continuing faculty development, and greater attention to teaching competence in graduate and postdoctoral training.

Its larger implication is that educational quality should not depend on a handful of dedicated instructors. It must become a continuing institutional responsibility.

This insight suggests a broader theory of educational institutions: effective learning depends not only on students’ abilities, but also on an organization’s capacity to identify and correct failures in its own practices.

When students repeatedly struggle with a particular concept, teachers need the flexibility to revise instruction. When large numbers fail the same introductory course, departments should examine its structure. When persistent disparities appear in learning outcomes, universities should investigate their causes rather than treating them as inevitable.

Evidence from these experiences must inform decisions about courses, faculty development, and institutional priorities. Educational reform then becomes more than a collection of individual teaching innovations. It becomes part of the university’s ability to learn.

This may be the report’s most consequential lesson for both countries.

What America Must Change: Bringing Teaching Back to the Center of the University

For American higher education, the central challenge is to bring institutional priorities into closer alignment with undergraduate learning.

The research freedom and scientific creativity of university faculty are among America’s greatest educational strengths. They should not be weakened in the name of teaching reform. Yet universities need to reconsider an implicit hierarchy of values in which generating new knowledge receives greater rewards than helping students master it.

Research universities could give more weight in faculty promotion to peer-reviewed curriculum reform, demonstrated improvements in conceptual understanding, scholarship on teaching and learning, and interdisciplinary educational contributions. Such measures would recognize teaching as intellectual work rather than merely a service obligation.

But teaching quality should not be reduced to student satisfaction scores or course completion rates. Students may prefer easier courses without learning more, just as a demanding course does not automatically indicate effective instruction. A credible evaluation system must examine what students understand and can actually do.

Introductory courses provide a particularly useful place to begin. Calculus, general chemistry, physics, and programming establish the foundations for advanced study, but they also frequently determine whether students continue in STEM. Better diagnostic assessment, targeted academic support, tutoring, and coordination between courses could reduce avoidable failure while preserving essential standards.

The goal is not to turn every student into a scientist or engineer. Students should be free to pursue other interests. But those decisions should emerge from informed judgments about their abilities and aspirations, not from a premature loss of confidence after an initial academic setback.

Educational equity adds another dimension. American debates often concentrate on admissions standards, racial and ethnic disparities, and university selection policies. Yet access to a university does not guarantee meaningful access to learning. Students from low-income families, rural communities, or families without a history of higher education may encounter financial pressures, gaps in preparation, and unfamiliar academic expectations even after admission.

The report directs attention toward what happens after students enter the classroom. Fairness cannot be judged solely by who is admitted. It must also consider whether students have a reasonable opportunity to meet demanding academic standards.

This understanding need not collapse into disputes over identity classifications or imply that every difference in outcomes reflects discrimination. It calls for careful attention to individual circumstances, institutional practices, and the evidence explaining why students succeed or struggle.

The final American challenge is continuity. Teaching improvement requires years of experimentation and accumulated experience, but public funding, political priorities, and university leadership can change quickly. Durable reform therefore needs stable professional standards, credible evidence, and institutional commitments capable of surviving short-term policy shifts.

For a country whose scientific achievements depend so heavily on its universities, undergraduate teaching deserves a more secure place in the academic mission.

What China Must Change: From Mastering Answers to Discovering Questions

China approaches the same problem from a different institutional position. Its educational system has considerable strengths in systematic study, sustained academic training, and large-scale implementation of common curricula. These have contributed to the development of a substantial scientific and technological workforce.

But mastery of established knowledge and the capacity for original discovery are not identical. Greater speed in solving problems does not necessarily produce deeper scientific creativity. Many important discoveries in the history of science began when researchers questioned the assumptions behind an accepted problem.

An educational system that consistently treats the correct answer as the destination of learning may encourage students to search for established solutions before considering whether the question itself has been properly formulated.

China’s most important educational challenge is therefore to preserve rigorous foundational training while making the ability to formulate questions an explicit educational objective.

A science class can ask students to solve problems while also identifying their hidden assumptions. An experiment can verify established findings while allowing students to design alternative tests. A university thesis can demonstrate command of the literature while proposing a genuinely worthwhile question for further investigation.

Such changes have implications for examinations and academic evaluation. If achievement is measured almost exclusively through easily quantifiable outcomes, exploratory work whose value may take years to emerge will struggle to find institutional support.

Yet reform should not move toward the opposite extreme. Critical thinking cannot substitute for subject knowledge, and open discussion cannot replace rigorous training. A student without a strong mathematical foundation will struggle to understand advanced physics. A graduate student unfamiliar with the relevant literature will find it difficult to make a meaningful original contribution.

China needs an educational progression in which foundational knowledge equips students to enter a discipline, open-ended inquiry teaches them to recognize the limits of existing explanations, and research experience develops their ability to investigate the unknown.

This reform must begin before university.

In 2025, China’s Ministry of Education issued the Guidelines for Science Education in Primary and Secondary Schools, calling for the cultivation of curiosity, imagination, inquiry, critical thinking, and innovation. Reforms promoting hands-on science learning in 2026 likewise identified long-standing weaknesses: excessive emphasis on established conclusions over investigation, and classroom instruction over practical experimentation.

The policy direction has begun to change. The difficulty is implementation.

A school cannot meaningfully encourage experimentation while treating inquiry as an extracurricular display unrelated to its educational priorities. Nor can students develop confidence in questioning if classroom schedules leave little room for discussion and investigation.

China’s particular challenge is to connect its stated educational ambitions with the actual organization of teaching and assessment. Its capacity for coordinated reform could become a major advantage if it is used to expand opportunities for inquiry rather than merely to enforce uniform outcomes.

Beyond Investment: Making Educational Resources Produce Learning

China faces another challenge highlighted indirectly by the NASEM report: the tendency to treat educational improvement primarily as a matter of resource investment.

School construction, laboratories, digital platforms, qualified teachers, and university research facilities are essential. In less-developed regions, adequate infrastructure and staffing remain prerequisites for educational quality.

But owning a laboratory does not mean students can conduct experiments independently. Providing an intelligent learning platform does not ensure that students become active learners. Nor does expensive research equipment necessarily give undergraduates more opportunities to pursue their own questions.

Educational resources create value when they are converted into meaningful learning experiences.

This distinction has consequences for how progress is measured. Schools should examine not only the laboratories they have built or the competitions they have entered, but also whether students can design experiments, interpret data, identify mistakes, and apply knowledge in unfamiliar circumstances.

Science competitions are valuable for identifying and nurturing exceptional talent. Yet a rising number of medals and award-winning projects cannot demonstrate that scientific understanding has improved across the student population.

The issue is especially important in a country marked by substantial differences among regions, schools, and family backgrounds. Educational opportunity is shaped not only by whether students attend school, but also by what kinds of intellectual experiences are available to them.

A student in an ordinary rural school should have opportunities to investigate scientific questions, just as a student in an elite urban school does. Their resources may differ, but scientific curiosity and the capacity to learn are not confined to institutions with the largest budgets.

Here the American report’s concern with equity intersects with China’s challenge of educational quality. Expanding opportunity means ensuring that the benefits of effective science education reach ordinary students, not only those selected for elite programs.

Artificial Intelligence Changes the Question: What Must Students Still Learn to Do Themselves?

The NASEM report focuses primarily on long-standing problems of undergraduate STEM education. Although it recognizes the changing technological environment, it does not establish generative AI as a proven solution to those problems. Its principles nevertheless provide a useful starting point for considering education in the AI era.

Traditional science education assumes that students develop competence through repeated study, memorization, and problem-solving. That remains true. What is changing is the relationship between completing a task and acquiring the ability to perform it.

Artificial intelligence can now solve many mathematical problems, explain physical concepts, write computer programs, and assist researchers in developing hypotheses. If schools continue to judge learning primarily by the accuracy of submitted work, students may produce increasingly impressive assignments without achieving comparable gains in understanding.

This raises a question more profound than conventional academic dishonesty: Does successfully completing a task with AI mean that a student has learned how to perform it?

The answer depends on how AI is used.

An AI-generated explanation may help a student overcome a conceptual difficulty. It may also allow the student to bypass the effort required to understand the concept. The educational value lies in whether the student can explain, question, verify, and apply what has been produced.

Science education should therefore develop two complementary capabilities.

The first is independent cognitive capacity: foundational knowledge, logical reasoning, scientific understanding, and the ability to evaluate evidence without relying entirely on technological assistance.

The second is human–AI cognitive collaboration: the ability to use AI to explore hypotheses, analyze data, conduct simulations, and expand the scope of inquiry while recognizing possible errors and the limits of computational results.

These capacities reinforce each other. Students lacking independent judgment risk becoming passive consumers of machine-generated answers. Those unable to collaborate effectively with AI may fail to benefit from powerful new tools.

Assessment should reflect this distinction. Students may still need to demonstrate essential knowledge without assistance, but they should also be evaluated on how intelligently they use AI.

Consider an engineering design assignment. After obtaining several proposals from an AI system, can a student identify their physical constraints, safety risks, and cost differences? When AI produces contradictory explanations, can the student use scientific evidence to determine which is more credible?

Such skills may be more relevant to future scientific work than speed alone.

Nevertheless, technological optimism requires restraint. Personalized AI explanations cannot automatically replace teachers, collaborative learning, or physical experiments. Whether AI-assisted instruction produces lasting improvements in independent competence and knowledge transfer must be tested across disciplines, ages, and educational settings.

The opportunities and risks will also differ between countries. In America, AI may provide students with weaker academic preparation more timely assistance, but unequal access and differences in digital literacy could widen existing disparities. In China, it may extend educational support to underserved regions while also making it easier for students to obtain standardized answers without fully understanding them.

Both countries must distinguish between technologies that help students complete more tasks and those that genuinely expand their ability to think.

Toward a New Philosophy of Education: Cognitive Revisability

The NASEM report is a study of undergraduate science education, but its findings point toward a larger philosophical question: What should count as success in education?

The system of mass schooling developed during industrialization emphasized standardized curricula, predictable academic progression, uniform assessment, and preparation for relatively stable occupations. It expanded educational opportunity on an enormous scale and helped create the workforce on which modern scientific and economic development depended.

Those achievements remain valuable. Yet a world in which knowledge changes rapidly, occupations evolve, and AI increasingly participates in research requires more than the transmission of established knowledge.

Education must cultivate the ability to keep learning when circumstances change—to reconsider judgments, reorganize knowledge, and act intelligently in unfamiliar situations.

This capacity might be called cognitive revisability in education.

A student who possesses cognitive revisability can form independent judgments without regarding them as permanently settled. Such a student can remain committed to scientific principles while changing an explanation when new evidence appears, and can act on existing knowledge while recognizing its limits.

The same principle applies to educational institutions. A school capable of cognitive revisability should be able to reconsider its curriculum when evidence shows that students are not learning effectively, adopt improved methods when research warrants change, and reexamine its objectives as technology and society evolve.

This idea shifts the emphasis from particular teaching techniques to the intellectual character of the educational system itself.

Educational reform has often proceeded from governments or administrators to teachers and students. Such coordination can make large-scale change possible, but it risks becoming a one-way process when those responsible for implementation have little opportunity to influence policy through their experience.

A more adaptive system requires information to move in both directions. Students’ learning experiences, teachers’ classroom observations, and findings from educational research must all help shape decisions.

America and China possess different resources for achieving this. The United States benefits from a diverse university landscape and considerable space for experimentation, but often struggles to spread successful practices. China has substantial capacity for coordination and implementation, but must ensure that uniform objectives leave sufficient room for local initiative and carefully evaluated alternatives.

Neither model offers a complete answer. Both can improve by making institutional adaptation a central feature of educational governance.

The purpose of comparison, then, is not to determine which national system is superior. It is to ask how each might become more capable of learning from its own experience.

The Next Educational Competition Will Be About Discovery—and the Capacity to Learn

The NASEM report examines American undergraduate STEM education, but it exposes a dilemma shared by modern school systems: We know a great deal about effective learning. Why do educational institutions still struggle to put that knowledge to use?

Artificial intelligence makes the question more urgent.

The United States will continue to need outstanding scientists and engineers. China will likewise need to expand its scientific and technological talent. But a nation’s capacity for discovery cannot be adequately measured through university rankings, research funding, graduate numbers, or academic publications alone.

We should also ask how many students discover an enduring interest in science, how many learn to formulate original questions, and how many acquire an understanding of evidence and independent judgment—even if they never pursue scientific careers.

For America, the challenge is to ensure that the intellectual achievements of its universities are matched by the quality of the education they provide. For China, it is to turn the strengths of systematic training and large-scale education into greater opportunities for independent inquiry and original discovery.

Both must confront an additional question: As artificial intelligence becomes increasingly capable of finding answers, can schools continue to cultivate people who know how to discover questions, evaluate evidence, and revise their own understanding?

The next phase of international educational competition will depend not only on how efficiently countries transmit knowledge, but also on how successfully they develop the capacity for discovery—and how readily their educational institutions learn from their mistakes.

American and Chinese education have different strengths and different limitations. The lesson of the NASEM report is that educational excellence requires more than talented teachers, diligent students, or abundant resources. It depends on institutions willing and able to examine their assumptions, test their practices, and change when the evidence demands it.

Only an educational system that continues to learn can prepare students for a world in which knowledge itself never stands still.

References and Sources

1. National Academies of Sciences, Engineering, and Medicine. (2025). Transforming Undergraduate STEM Education: Supporting Equitable and Effective Teaching. National Academies Press. DOI: 10.17226/28268. Full report.

2. National Academies of Sciences, Engineering, and Medicine. Consensus Study Highlights, January 2025. Official summary and seven principles of teaching.

3. Ministry of Education of the People’s Republic of China. Guidelines for Science Education in Primary and Secondary Schools, January 2025.

4. Central Committee of the Communist Party of China and State Council. Outline of the Plan for Building China into a Leading Country in Education (2024–2035), published in 2025.

5. Ministry of Education of the People’s Republic of China. Deepening Learning-by-Doing in Science Education to Strengthen the Foundations of a Leading Nation in Education, Science, Technology, and Talent, August 2026.

Author’s Note: This article draws on the original NASEM report and relevant educational policy documents. The report examines undergraduate STEM education in the United States, not the entire American K–12 educational system. The discussion of Chinese education, AI-assisted learning, and future international educational competition represents a comparative analysis and independent interpretation developed from the report’s findings.

Subscribe now

For more than two decades, one thing sat near the center of China’s economic model: land. Local governments acquired it, built industrial parks and roads, sold residential parcels, attracted developers and turned rising land values into revenue and borrowing power. Land fed the property boom, property supported local government finance, and local finance in turn paid for infrastructure, investment promotion and urban expansion. When people spoke of China’s formidable state capacity, this was often what they meant: the government could assemble land, credit and political authority quickly enough to turn farmland into factories, rail stations, highways and entire new cities.

That model has not disappeared, but its best years are over. China’s long property downturn has weakened land-based finance, while debt accumulated by local governments and their financing vehicles has made the old system harder to sustain. It is tempting to conclude that the decline of the property model must also mean a decline in Chinese state capacity. That may be wrong. What appears to be happening instead is a change in the kind of state capacity China is building.

The older model was organized around land, real estate, investment promotion, infrastructure and G.D.P. The emerging model is increasingly organized around electricity, manufacturing, technology, supply chains, data and national security. The state is not simply becoming larger or smaller. It is changing what it is built to do.

From Building Cities to Building Systems

Susan H. Whiting’s recent work on Chinese land and property rights helps explain the old model. Her argument is that Chinese law did more than protect ownership. It also helped the state reassign property rights, move land from one use to another and channel the disputes that followed into courts, petitioning systems, mediation and other institutions. One of the central capacities of the land era, then, was the ability to reorganize property.

That helps explain something about China’s extraordinary speed of physical transformation. Building a highway or an industrial park looks like an engineering task, but in practice it requires much more: land has to be acquired, rights altered, financing arranged, local agencies coordinated, companies brought in and sometimes residents moved. China was unusually good at combining these things quickly. Land was central because it performed several jobs at once. It was an asset, a source of government revenue, a tool for attracting investment and collateral for borrowing. It was also the physical space on which urbanization took place.

The result was a form of state capacity particularly suited to changing the physical landscape. Call it spatial state capacity. Its central questions were concrete: Where should the highway go? Where should the new district be built? How should land be converted? How can financing be assembled? How can a project be completed quickly? That kind of capacity was enormously useful during rapid industrialization and urbanization. But cities cannot expand forever, property prices cannot rise forever, and land eventually reaches economic and political limits.

The new challenge is different. Electric vehicles are not highways. Semiconductors are not new towns. Artificial intelligence, advanced batteries, electrical grids, critical minerals and industrial software cannot be created simply by clearing land and financing construction. They are systems. An electric vehicle depends on batteries, minerals, power electronics, software, charging networks, electricity and sophisticated manufacturing. Solar power connects polysilicon, wafers, cells, modules, storage and the grid. Semiconductor production depends on equipment, materials, design software, engineers, energy and global supply chains. No single piece works by itself.

The state capacity required for this world is therefore different from the capacity required to build roads and cities. The important verbs are no longer simply acquire, finance and build. They are increasingly coordinate, standardize, scale, secure and adapt. This is a more demanding form of capacity. It might be called systemic state capacity.

A Different Kind of State

A second change concerns who sets the goals. For much of the reform era, one of China’s most powerful economic engines was competition among local governments. Cities and provinces competed for factories, capital and development projects. They built industrial zones, sold land and expanded infrastructure. Beijing set broad priorities, but local officials were given strong incentives to pursue growth. That system generated remarkable energy, but it also produced a predictable problem: almost every locality wanted its own development zone, its own signature industries and its own route to higher G.D.P.

Recent work by Yi Feng and Fulong Wu describes a shift toward what they call strategic statization. The basic idea is that local development is becoming more closely tied to national strategic priorities. Local governments still compete and market tools still matter, but technology, supply-chain resilience, energy security and national security increasingly shape which sectors receive support.

The important change is not simply that the state is intervening more. The state is pursuing a different objective. The old question was often: Will this project bring investment and growth? The newer questions are increasingly: Is this technology strategically important? Is the supply chain secure? Can China produce the critical components itself? Does the country have enough electricity, computing power and industrial capacity to support it?

That is a move from something like a local growth state toward a strategic industrial state. An industry can now matter even when its short-term returns are modest, if its strategic value is considered high. A highly profitable sector can also lose favor if it no longer fits national priorities. That is a much deeper change than simply saying that Beijing is “back in the economy.”

It also raises an unavoidable question: Who pays for the new state? The land era had a powerful financing mechanism. Urbanization raised land values, higher land values increased local government revenue and borrowing capacity, and that money financed more urbanization. It was a self-reinforcing loop: urbanization created land value, and land value financed more urbanization.

When the property market turns down, that loop becomes much harder to sustain. Local governments lose land revenue, debt remains and existing infrastructure still has to be maintained. At the same time, the new model requires enormous investment in advanced manufacturing, energy, power grids, research, industrial equipment and strategic infrastructure. China therefore faces an awkward transition: it has to finance the new system while still paying for the old one. This is not merely a debate over whether to rescue the property market. It is a balance-sheet problem on a national scale, and it will be expensive.

Two Chinas, Two Curves

If this transition continues, China’s economy may increasingly be defined by two opposing trends: real estate will become less central, while electricity, advanced manufacturing and industrial technology become more important. That does not mean housing ceases to matter. Property still represents a large share of household wealth, and banks and local governments remain heavily exposed to it. But the center of gravity may be moving.

In the past, the most visible symbols of Chinese state capacity were airports, expressways, high-speed rail and new cities. In the future, they may be giant power grids, electric-vehicle supply chains, batteries, robotics, industrial machinery, data centers and control over key inputs. The underlying economic logic is changing as well. The old model looked roughly like this: land → credit → real estate → infrastructure → urbanization. The emerging model looks more like: electricity → technology → manufacturing → scale → supply chains → reinvestment.

This is why it has become increasingly misleading to judge China only by the condition of its property market. Look at real estate, and China can appear to be an economy in decline. Look at electricity, batteries, electric vehicles, industrial equipment and some parts of advanced manufacturing, and a very different picture appears. Both pictures are true. The more useful question is not whether China is rising or falling, but whether the new system can grow faster than the old one deteriorates. That may be the central economic question of the next decade.

The Danger of Being Good at the Wrong Thing

Strong state capacity has obvious advantages, but it also has a less discussed danger: it can scale mistakes. A weak government may fail to build a bad project; a strong government may build the same bad project everywhere.

The new model therefore carries at least three risks. The first is overinvestment. When national priorities identify a sector as strategic, many local governments may rush toward the same target. That can produce enormous capacity — and enormous excess capacity. The second is uncertainty. Infrastructure is comparatively easy to plan. A city can often tell whether it needs another highway, airport or power line. Technology is different. No government can know in advance which battery chemistry will dominate, which artificial-intelligence architecture will prove most valuable or which robotics company will become globally competitive. This is the classic information problem of industrial policy, and it becomes more serious as an economy moves closer to the technological frontier. The third risk is that fiscal problems do not disappear; they merely change form. Under the old model, local governments borrowed to build roads and new districts. Under the new model, they can borrow to build semiconductor plants, battery factories, data centers or government-backed investment funds. The sector changes. Bad capital allocation does not necessarily disappear with it.

The transition therefore cannot be reduced to a simple story in which manufacturing replaces property. The harder challenge is whether the new form of Chinese state capacity can become better at coping with uncertainty than the old one.

The Hardest Shift: From Mobilization to Learning

This may be China’s most important long-term problem. The state that built modern China was extraordinarily good at mobilization. Decide to build a railway, then secure the land, arrange the financing, coordinate the agencies and build the railway. That capacity remains enormously valuable, but it is not enough at the technological frontier, because the defining feature of frontier innovation is that nobody knows the answer beforehand.

That changes what a capable state must do. It must allow different companies to try different things, tolerate failure, abandon technologies that do not work, leave space for private investors to exercise judgment and for scientists to challenge official assumptions, preserve some room for local experimentation and allow market prices to transmit bad news as well as good news.

In other words, China may eventually need a second shift in state capacity: from a mobilizing state to a learning state. A mobilizing state is good at concentrating resources; a learning state is good at discovering when those resources are being used badly. The first was essential to catch-up industrialization. The second becomes more important as an economy moves toward innovation.

This is why the familiar argument over “more state” or “more market” is not very useful here. It treats state capacity as though it were one thing. It is not. Land-based fiscal capacity can weaken while technological coordination grows stronger. Local investment autonomy can shrink while national industrial coordination expands. Digital administrative capacity can rise even as private investment confidence falls.

The real question is not whether the Chinese state is becoming stronger. It is which parts of the state are becoming stronger, and what they are becoming stronger at doing.

That is what the phrase “shift in state capacity” is meant to capture. Over the past four decades, China built one of the largest industrial systems in history by becoming exceptionally good at turning land, credit and infrastructure into urbanization and manufacturing. The next problem is harder: Can the same political system become equally good at producing sustained innovation?

If it can, the result will be more than another growth cycle. China will have created something harder to reproduce than the old property model: an ecosystem linking electricity, manufacturing, technology, data and supply chains at enormous scale. If it cannot, the country may end up carrying the debts of its old development model while adding the excess investment of the new one.

The most important change in China today, then, is not that “the state has returned.” The state never really left. The change is that the state is trying to do something different. For years, its signature skill was turning land into cities. Now it is trying to turn electricity, technology and manufacturing into national power.

What It Means for the United States

This shift also changes the problem China presents to the United States. Twenty years ago, the American anxiety was largely about the “China factory.” China manufactured shoes, furniture, toys and textiles at enormous scale, often at prices American companies struggled to match. Today, the more consequential challenge may be the China system.

Batteries, electric vehicles, solar equipment, critical minerals, power-grid technology and advanced manufacturing are not simply products. They are parts of the industrial infrastructure on which other industries depend. The question is therefore no longer only whether China sells too much. It is whether other countries can operate key sectors without Chinese supply chains.

The United States has already begun answering that question with its own vocabulary: resilience, reshoring, supply-chain security, industrial capacity, national security. These terms sound different from China’s language of industrial strategy, but they point toward the same underlying concern: Can a country still organize the industries on which its power depends?

U.S.-China competition is therefore becoming harder to understand simply as Tesla versus BYD, Apple versus Huawei or one company against another. Behind the companies stand increasingly complex systems. Who has enough electricity? Who has the grid? Who controls the minerals? Who can make the machines that make the machines? Who has the engineers? Who can take a laboratory breakthrough and manufacture it at scale? Those questions once belonged mainly to industrial economics. Now they increasingly belong to national security.

The competition between the United States and China may therefore come to depend less on headline G.D.P. than on something less visible: how quickly each country can turn resources into strategic capability. That does not mean war is inevitable. It means that economic policy is beginning to take on some of the logic of preparedness. Batteries are both consumer products and strategic storage technologies. Semiconductors are both commercial goods and the foundation of artificial intelligence and military systems. Power grids serve households, but they also determine whether factories and data centers can function. Ports, minerals, cloud computing, satellites and industrial software increasingly sit on both sides of the old line between economics and security.That line is fading.

This is why the American challenge is larger than Chinese subsidies. Tariffs can make Chinese goods more expensive, but they cannot by themselves create a battery ecosystem, an engineering base or a network of suppliers. Industrial ecosystems are built slowly, and once they exist, they reinforce themselves. A battery manufacturer attracts materials companies, materials companies attract equipment suppliers, equipment suppliers attract engineers, engineers attract research, and scale lowers costs. The final product is not merely a successful company. It is a system.

And systems are much harder to compete with than products.

That may be the deeper significance of China’s shift in state capacity. China is trying to build a new kind of economic state. In responding to it, the United States may find that it has to rebuild one of its own.

Subscribe now

Modern people have acquired a deep faith in the repairability of relationships. If we are honest enough, rational enough, generous enough, almost any relationship, we are told, can be saved. Friends can clear up misunderstandings; couples can learn to communicate; families can work through old injuries. Therapy culture, popular psychology, and contemporary media all reinforce the same vocabulary: express your feelings, set boundaries, communicate better, repair the bond. There is nothing inherently wrong with this ideal, but it rests on an assumption that is rarely stated: both people must be willing to confront reality and take responsibility for the part they played in creating the problem. Without that condition, communication may cease to be a means of repair and become one more mechanism of exhaustion.

This is what makes certain relationships so difficult. After a conflict, one person refuses to discuss what they did and focuses instead on why you became angry; they ignore the aggression and scrutinize your tone; they bypass the original injury and concentrate on the moment when you finally lost control. The causal order is quietly reversed. Before long, the person who was hurt begins explaining themselves: Am I too sensitive? Did I overreact? Was my way of speaking the real problem? Did I, in fact, hurt the other person? The original issue disappears. What makes this structure so effective is not simply the denial of wrongdoing. It is the redefinition of the dispute: the problem is no longer my behavior but your reaction to it. One person’s conduct drops out of view, while the other person’s emotional response becomes the entire subject. Over time, one party is asked not only to absorb the injury but also to assume responsibility for having been injured by it. This is blame-shifting in one of its more subtle forms.

I. The Hardest Relationships Are Not Those in Which People Make Mistakes, but Those in Which No One Can Own One

Popular psychology is fond of labels: “toxic person,” “narcissist,” “emotional vampire,” “victim personality.” Such terms can occasionally help people make sense of their experience, but they can just as easily reduce a complicated relationship to a personality diagnosis. Selfishness does not automatically amount to pathology, and a person who shifts blame during an argument cannot be defined by a few bad conflicts. More revealing than any label is a simpler question: Can this person acknowledge that they, too, helped create the problem?

Healthy relationships do not require people to be incapable of wrongdoing. In any enduring relationship, people lose their temper, misread one another, say cruel things, and behave badly. What determines whether a relationship can be repaired is whether, after the conflict, each person can turn back and see their own part in it. “What I said did hurt you.” “I bear some responsibility for this.” “I do not entirely agree with your reaction, but I understand why you reacted that way.” These are ordinary sentences, yet they require a difficult psychological capacity: the ability to step temporarily outside the identities of “the victim” and “the person who is right” and look at oneself from somewhere else.

Some people find this almost impossible. For them, admitting that a particular action was wrong can feel perilously close to admitting, “I am a bad person.” To protect the self, responsibility has to be pushed outward. A stable cycle emerges: I act; you are hurt; you react; I seize on your reaction; I prove that I am the real victim. Once this pattern hardens, reasoned discussion rarely gets very far, because the two people are no longer arguing about the same thing. One is trying to establish what happened; the other is trying to preserve the conviction that I cannot be the one who was wrong.

II. Victimhood Begins as an Experience and Can Become an Identity

Any discussion of “victim narratives” requires care. Real victimization exists. Deception, violence, humiliation, coercion, and unfair treatment cause real harm. The fashionable instruction not to “play the victim” should never become another way of silencing people who have genuinely been injured. The problem lies elsewhere: victimhood may begin as an experience and gradually become an identity through which the world is interpreted.

When someone comes to understand themselves primarily as “the person whom others always hurt,” reality begins to assume a stable causal structure. My pain comes from other people. My failures come from circumstances. My emotions are caused by other people’s provocations. My relationships collapse because others fail to understand me. Wherever suffering appears, an external party must be responsible. This interpretation has an obvious short-term function: it protects self-esteem and postpones the need for self-examination. But the cost is equally clear. It weakens one of the most important human capacities: agency.

If all my suffering comes from other people, then change must also wait for other people. They must apologize. They must change. They must understand me. The world must become acceptable to me before I can become calm. A person may gain the moral high ground while losing practical control over their own life. That is the deeper danger of a permanent victim identity: not that it makes someone appear weak, but that it can hand control of one’s life to the very people one most resents.

III. Why These Relationships Are So Exhausting

Conflict already consumes psychological resources. We have to sort through facts, infer motives, regulate emotion, revise our language, and consider the other person’s feelings. If both people are doing this work, a relationship may be difficult but can still move forward. If only one person is doing it, the structure changes. One person keeps asking, “What did I do wrong?” while the other waits for, “When will you admit you were wrong?” One person adjusts boundaries; the other looks for ways around them. One person tries to understand the problem both have created; the other needs only to prove that the problem lies with you. Over time, the more reflective person may become the one most inclined to doubt themselves.

These relationships do not necessarily produce spectacular fights. More often, the damage takes the form of chronic low-level erosion. Before speaking, you calculate whether the other person will become angry. Before expressing a need, you wonder whether you are being selfish. After succeeding at something, you hesitate to show too much pleasure. Even happiness begins to carry a trace of apology. Gradually, you learn to make yourself smaller. This is where the deeper power of relationships becomes visible: we are not only injured by relationships; we are formed by them. People tend to internalize the rules of the environments in which they live. If a relationship repeatedly tells you not to stand out, not to be too sensitive, not to need too much, not to make the other person uncomfortable, then eventually no external pressure is required. You begin to police yourself. At that point, what has been damaged is no longer merely the relationship. It is your relationship with yourself.

IV. How “Toxic” Became a Way of Talking About Intimacy

The contemporary obsession with “toxic relationships” deserves to be taken seriously, but not quite at face value. Over the past decade, terms such as “toxic,” “boundaries,” “emotional labor,” “emotional value,” “draining,” and “protect your peace” have migrated from therapeutic language into ordinary speech. Their popularity is usually treated as evidence that people have become more psychologically literate. That is partly true. But the vocabulary also reflects a larger historical change in the structure of relationships themselves.

In traditional societies, many relationships were difficult to leave. Family, clan, marriage, neighborhood, church, workplace, and local community imposed forms of belonging that were only partly chosen. A person could be unhappy inside a relationship and still have very few credible ways out. Under those conditions, endurance became a moral virtue. One was expected to restrain oneself, tolerate disappointment, preserve appearances, honor obligations, and remain attached even when attachment was costly. Modern life has weakened many of those constraints. Urban mobility, women’s economic independence, divorce law, anonymous cities, digital communication, and the decline of lifelong institutions have all made relationships more voluntary. One can leave a job, leave a city, end a marriage, block a number, exit a group chat, unfollow an account, or simply allow a friendship to disappear. The cost of exit has fallen.

This is, in many respects, an achievement. It has allowed people to leave arrangements that earlier generations were expected to endure indefinitely. The language of “toxicity” has also given people a way to describe forms of injury that older moral vocabularies often normalized: manipulation, humiliation, chronic disrespect, coercive dependence. But once relationships become more voluntary, they also become easier to evaluate through the logic of choice, and choice in modern societies is rarely innocent of the market.

Does this person meet my emotional needs? Is this relationship adding value to my life? Am I getting enough back for what I put in? Is this person “good for my energy”? Should I cut my losses? The language sounds psychological, but its grammar is increasingly economic. Relationships begin to resemble portfolios. Intimacy begins to resemble a service. Emotional life becomes something to optimize. A friendship that disappoints can be “outgrown.” A partner who fails to respond correctly can be labeled “toxic.” A difficult relative can be “cut off.” The same society that has taught us not to remain trapped in damaging relationships has also made it easier to treat any demanding relationship as a bad investment.

This is the paradox. Earlier generations often suffered because relationships were too difficult to leave. Ours may increasingly suffer because leaving has become one of the first available solutions to relational difficulty. The problem is no longer simply how to escape bad relationships. It is how to distinguish a relationship that is genuinely destructive from one that is merely difficult, inconvenient, disappointing, or resistant to optimization. That distinction is harder than the language of “toxicity” makes it sound.

V. Boundaries Protect Not Comfort but Subjectivity

This is why the concept of “boundaries” needs to be handled more carefully. Today boundaries are often treated as a way of saying, “Do not make me uncomfortable.” But every real relationship produces discomfort. Friends point out our faults. Partners ask us to take responsibility. Children disrupt our routines. Parents force us to confront histories we would sometimes rather avoid. Intimacy itself contains friction.

What boundaries protect, then, is not comfort but subjectivity: my continuing right to interpret my own experience. You may disagree with my emotions, but you cannot simply declare that I had no right to be hurt. You may explain your intentions, but “I did not mean it” cannot erase the consequences of what you did. You may have needs of your own, but you cannot require me to sacrifice my integrity indefinitely in order to satisfy them. Mature boundaries do not keep other people outside. They preserve a place inside the relationship from which one can still say: I remain myself.

The deepest harm in many draining relationships occurs precisely here. It does not always take the form of overt control. Instead, one person gradually begins to understand themselves through another person’s reactions. If the other person is pleased, I must have done the right thing. If they are hurt, I immediately doubt myself. If they accuse me, I immediately begin explaining. Eventually, the other person becomes the mirror through which I interpret myself. In this sense, leaving certain relationships is about more than reducing conflict. It is about recovering the authority to interpret one’s own life.

VI. Not Every Conflict Deserves to Be Resolved

Modern people also possess a powerful conviction that problems must be resolved. As long as some misunderstanding remains unexplained, some person still fails to understand us, or some relationship has not received a proper ending, we feel that life has not fully moved on. Yet many differences between people cannot be completely resolved. Values differ. Temperaments differ. Conceptions of responsibility differ. Needs for intimacy differ. Even memories of the same event can diverge radically.

Sometimes more communication merely produces more communication. Explanations generate further explanations; arguments produce further arguments; two people spend enormous amounts of life trying to force the other to admit: I was right. Maturity may involve discovering that many things do not require the other person’s agreement in order to end. You do not need someone who hurt you to finally say, “Yes, you really were hurt,” before you are entitled to leave. Nor do you need to prove that you have “done everything possible” before ending a relationship.

For some relationships, the most mature response is not repair but recognition: We are no longer capable of relating to each other in a healthy way. That does not require a final verdict about who was entirely right and who was entirely wrong. It means only that the cost of continuing has come to exceed what the relationship can still sustain.

VII. Leaving Is a Capacity, but It Should Not Become a Habit

When people are young, they often regard persistence as a virtue: preserve the relationship, preserve the friendship, preserve the family role, keep talking until everything is clear, keep trying until the other person understands. With age, another question becomes increasingly important: How much more of my life is this worth?

Time is never an abstract resource. An evening spent arguing cannot also be spent reading. An afternoon spent explaining yourself cannot simultaneously be spent with someone else you love. Ten years devoted to repairing a relationship that cannot be repaired are also ten years unavailable for building another life. In this sense, leaving can be understood as a way of allocating the finite resources of a life. It does not necessarily mean, “You do not matter.” It may simply mean: My life matters too.

A mature understanding of relationships is therefore neither endless endurance nor immediate departure whenever discomfort appears. It is the capacity to distinguish among relationships: some deserve preservation, some repair, some reduced expectations, some distance, and some an ending. Judgment consists precisely in making these distinctions. Yet we should resist another modern illusion: turning “cut your losses” into a universal principle of life. Relationships are not stocks, and people are not products. Real intimacy inevitably contains cost, inconvenience, misunderstanding, compromise, and waiting. If every relationship that makes us uncomfortable is classified as “draining,” the result may not be freedom but a life that is highly purified, highly controllable, and profoundly lonely.

VIII. Between Obligation and Exit

The older moral world of relationships was organized around obligation. Stay because this is your family. Stay because this is your marriage. Stay because elders must be respected. Stay because loyalty matters. Stay because departure would shame the group. This ethic produced forms of stability that modern people often underestimate. It also trapped people inside relationships that could be cruel, humiliating, or dead.

The contemporary moral world has moved in almost the opposite direction. Its key words are autonomy, boundaries, authenticity, self-care, emotional health, and exit. This change has been liberating. But every liberation creates its own temptations. When obligation governed relationships too strongly, people were asked to sacrifice themselves to preserve the bond. When autonomy governs too strongly, the bond itself risks becoming valuable only insofar as it continuously affirms the self.

The first error says: You must remain, whatever it costs you. The second says: You should leave whenever the relationship costs you too much. Neither is sufficient. A durable relationship always asks something of us. It imposes inconvenience, limits options, and requires memory, forgiveness, patience, and occasionally the willingness to remain present when presence is unrewarding. No serious account of intimacy can eliminate sacrifice. But sacrifice has a limit.

The question is not whether a relationship costs us something. Every meaningful relationship does. The question is what kind of cost it asks us to bear. There is a difference between the discomfort of being challenged and the erosion of being chronically invalidated; between compromise and self-erasure; between taking responsibility and carrying all responsibility; between forgiving imperfection and repeatedly submitting to the same harm.

A mature ethics of relationship therefore has to resist both older demands for unconditional endurance and newer fantasies of frictionless intimacy. Its standard might sound something like this: I am willing to bear the costs of relationship, but not the destruction of my subjectivity. I accept that intimacy creates friction, but not a structure in which responsibility permanently flows in one direction. I am willing to repair what can be repaired, but I will not make another person’s refusal of self-examination the central project of my life. I accept that people I love will sometimes make me uncomfortable, but I will not remain indefinitely in a relationship that steadily requires me to become less myself.

This standard is harder to apply than asking whether someone is “toxic.” That is precisely why it is more useful.

IX. Growing Up Means Deciding Again What Deserves Your Life

The real value of the advice “do not try to fix everyone” is therefore not that it offers a philosophy of indifference. It reminds us of something more basic: psychological resources are finite. Attention is finite. Patience is finite. Emotional energy is finite. Time is finite. Life is finite. Maturity does not mean acquiring the ability to solve every relationship. It means gradually learning which problems deserve to be solved, which relationships deserve repair, which misunderstandings can be released, which people do not need to be convinced, and when it is time to leave.

When we are young, winning and losing can seem terribly important. We want to prove who was right, make the other person understand, establish the truth once and for all. Later we begin to see that many wars between people produce no real victor, only two increasingly exhausted participants. Leaving a relationship, then, does not necessarily mean “I won,” nor does it mean “you lost.” Sometimes it means only this: I am no longer willing to spend the rest of my life fighting this war.

A mature departure does not require revenge or a theatrical goodbye. It simply redirects attention away from the question “How can I change this person?” and toward a more consequential one: To whom, to what kinds of relationships, and to what kind of life do I want to give the finite portion of life that remains to me?

On September 30, 2026, Britain’s MI5 issued its first standalone “espionage alert,” urging British universities to re-examine certain research collaborations involving China. British authorities said that an organization they linked to China’s state-security system had used research funding to establish contact with scholars working in artificial intelligence, cybersecurity, covert communications, and related fields. More than 100 British researchers had participated in relevant projects, many of them possibly unaware of the relationships behind the funding. The Chinese Embassy in Britain rejected the allegations.

At almost exactly the same moment, Cambridge University Press was preparing to publish Anna L. Ahlers’s China in Global Science, a forthcoming contribution to its Cambridge Elements series Global China. Cambridge describes China as having become a scientific powerhouse at remarkable speed: a leading producer of high-ranking publications, a site of frontier research, and an initiator of scientific alliances beyond the traditional OECD world. More important, the publisher says that China’s scientific ambitions are no longer confined to building domestic technological capacity; they increasingly concern global leadership.

The juxtaposition is almost too perfect. One story asks whether scientific openness to China has become a security risk. The other asks what happens when China is no longer merely entering global science but helping to shape it. Together they reveal a central paradox of twenty-first-century science: China became a scientific power through global science; now that China has become a scientific power, global science is beginning to rediscover borders.

That paradox is also the intellectual center of Ahlers’s book. Cambridge presents the Element as an effort to trace China’s integration into science understood not simply as a collection of national research systems but as a world institution. Ahlers explicitly resists treating Chinese science either as an exotic exception or as a monolithic enterprise wholly directed by the party-state. Her interest lies instead in the interaction between two forces: China has adopted organizational forms and norms from global scientific knowledge production, while its growing scale and power are beginning to reshape those same institutions.

That is a more consequential argument than another inventory of Chinese papers, patents, laboratories, or research spending. It shifts the question from How strong is Chinese science? to What kind of global science emerges once China becomes one of its central actors? That shift is what makes China in Global Science worth reading.

A note of caution is necessary. Cambridge still lists the Element as forthcoming, with online publication expected in October 2026 and print publication on October 31. The discussion here therefore evaluates the argument Cambridge attributes to the book, together with Ahlers’s established research program, rather than pretending to assess chapter-by-chapter evidence that is not yet publicly available.

China Is Not Outside Global Science

Ahlers’s first important move is methodological. Much writing about Chinese science still begins with a binary: on one side stands an open, international, peer-governed “Western” scientific system; on the other stands a centralized, mission-driven Chinese system shaped by the party-state. The attraction of this contrast is obvious. China does possess strong national research strategies, large state programs, politically defined technological priorities, and increasingly explicit security objectives. But the binary obscures as much as it reveals.

China’s scientific rise did not occur outside the international institutions of modern science. Chinese universities adopted organizational forms that were already global. Chinese scientists entered international peer-review systems, published in international journals, trained abroad, participated in transnational research communities, and worked within disciplines whose standards were not invented in Beijing. This is precisely why Ahlers’s formulation matters. Cambridge says that the book studies how China became integrated into global science “as a world institution,” while retaining distinctive national arrangements of governance and mission.

The argument can be reduced to a simple formula: Global scientific institutions × Chinese state capacity → the contemporary Chinese scientific system. Global science supplied knowledge, disciplinary standards, publication systems, professional identities, research networks, and institutional templates. China absorbed these elements into its own universities, research programs, funding structures, and industrial strategies. But Ahlers’s claim does not stop with absorption. China is beginning to shape the institutions it entered.

That second movement is the more interesting one. Imitation means that A copies B. Integration means that A enters B. Transformation begins when A’s entry becomes large enough to alter B itself. This is the point at which China in Global Science becomes more than a book about China. It becomes a book about the changing organization of science.

What Ahlers Changes in the China-Science Debate

The conceptual strength of Ahlers’s approach is that it refuses two explanations that are each partly true and therefore especially tempting. The first says that China’s scientific rise demonstrates the success of a distinctive Chinese model. The second says that China simply became successful by adopting a Western scientific system.

Neither is sufficient. The first underestimates how thoroughly Chinese science has depended on global institutions. The second underestimates what happens when globally circulating institutions are reorganized inside a political system with China’s scale, state capacity, funding power, and strategic ambitions. Ahlers instead places the interaction itself at the center.

This is consistent with the larger intellectual project of Cambridge’s Global China series, which seeks to move China studies beyond China’s territorial boundaries and study how China’s global engagements change both China and the world around it. In that sense, China in Global Science fits the series unusually well. Its subject is not really “Chinese science” as a bounded national object. Its subject is the relationship between China and a global scientific order in which the distinction between “inside” and “outside” China is becoming harder to sustain.

That is the book’s first major contribution.

From Catching Up to Defining the Field

Ahlers’s second important intervention follows from the first. For years, the most common measurements of Chinese scientific power were catch-up indicators: publications, citations, patents, university rankings, research spending, overseas-trained scientists, and the number of laboratories approaching the global frontier. These remain useful, but they belong to a particular historical stage.

A catch-up indicator asks: How far is China from the leader? A different kind of indicator becomes necessary once the question changes to: Is China beginning to define what leadership itself looks like?

These are different questions. In the first stage, China entered a competition that others had already organized. The leading journals, laboratories, disciplinary centers, and research standards largely already existed. Chinese scientists had to enter these systems and perform increasingly well within them. A scientific power, however, does more than publish more papers. It builds research facilities that others need to use, organizes international networks that foreign scientists want to join, sets technical standards, establishes journals with global influence, attracts foreign researchers, and makes scientists elsewhere dependent on its data, instruments, platforms, and collaborative networks.

The next stage of Chinese science therefore cannot be judged solely by output indicators. It requires another set of measures, which I would call rule-producing indicators: who builds the major facilities that foreign scientists need, who creates databases that other researchers cannot easily replace, who defines technical standards, who convenes international research networks, who attracts foreign scientists for sustained periods, who establishes journals and institutions that begin to determine what questions receive attention, and who is able to make the scientific development of other countries partly dependent on its infrastructure.

A country that does these things has crossed an important threshold. It is no longer merely competing effectively inside a scientific order designed elsewhere. It is helping organize that order. This is why the phrase global leadership matters more than another statistic showing that China publishes large numbers of papers.

The “Science Silk Road” and the Geography of Knowledge

Ahlers’s broader research program makes this point even sharper. Her current work examines what she and her collaborators call “China’s Science Silk Road and the New Geopolitics of Knowledge Production.” That framing pushes the analysis beyond laboratories and publication counts toward the geography of scientific networks.

The usual image of the Belt and Road Initiative is physical infrastructure: ports, railways, power plants, roads. But scientific power also depends on another kind of infrastructure: knowledge infrastructure. Universities are infrastructure. Laboratories are infrastructure. Databases are infrastructure. Satellites and observatories are infrastructure. Research funds are infrastructure. So are systems for training scientists and engineers. Their political effects can be less visible than those of a railway, but they may last longer.

If an African country uses a railway built by China, it acquires a logistical connection to China. If its engineers train in Chinese institutions, its laboratories use Chinese instruments, its researchers participate in Chinese-funded projects, and its scientific data flow through platforms organized by Chinese institutions, something more consequential has happened: a network of knowledge has formed. The difference matters because networks create habits, dependencies, standards, careers, and intellectual reference points.

Ahlers’s research therefore opens a question much larger than whether China is “cooperating” with the Global South: Is China beginning to alter the geography through which scientific knowledge itself is produced? For much of the postwar era, global science had a recognizable hierarchy. The United States sat near its center, joined by Western Europe, Japan, and a small number of other advanced economies. Scientists from elsewhere often entered global science by entering networks organized around those centers. China first rose by entering that system. It may now be building additional centers.

If so, “Science Silk Road” is not simply another name for science diplomacy. It may describe the formation of alternative routes through which researchers, institutions, instruments, data, and research agendas circulate. That possibility deserves more attention than the phrase “Belt and Road” usually receives.

The Book’s Strongest Idea: Nationalized and Globalized at Once

The most valuable argument in China in Global Science, at least as Cambridge presents it, may be that Chinese science cannot be adequately described as either globalized or nationalized. It is both.

This sounds simple, but its implications are not. If we emphasize nationalization, we see national laboratories, government research programs, strategic technologies, military-civilian applications, security controls, and politically defined missions. If we emphasize globalization, we see international journals, foreign-trained scientists, multinational collaborations, imported institutional norms, shared disciplinary standards, and dense networks of knowledge exchange. Both descriptions are accurate. The mistake is to assume that one must cancel out the other.

Ahlers’s framework suggests something more interesting: globalization can itself become a source of national scientific capacity. That reverses a common assumption. We often speak as if national power increases when globalization decreases, as if a country becomes more sovereign simply by becoming less dependent. But China’s scientific rise suggests another route. A state may increase its power precisely by becoming more deeply embedded in global networks, provided it can convert those connections into domestic capabilities.

This is where Ahlers’s argument connects to one of the most important questions in contemporary political economy: what does “autonomy” mean in a world built on interdependence?

From Self-Sufficiency to Network Autonomy

China’s language of technological self-reliance is often interpreted as a project of reducing foreign dependence. Some reduction of vulnerability is clearly part of it. But self-sufficiency is not the same as autonomy.

Modern science makes pure self-sufficiency extraordinarily difficult. Advanced semiconductors depend on international networks of equipment, software, materials, fabrication expertise, and markets. Fusion research requires multinational engineering capabilities. Artificial intelligence depends on international papers, open-source software, chips, researchers, and infrastructure. Basic research itself rests on knowledge accumulated across borders.

The more useful concept may therefore be network autonomy. Network autonomy does not mean leaving a network. It means retaining meaningful freedom of action while remaining inside one because your own position has become difficult to replace. There is a fundamental difference between dependence and interdependence. If Country A requires a crucial technology from Country B while B requires almost nothing from A, A is vulnerable. If A still depends on B for some technologies but B and many others depend on A for batteries, minerals processing, manufacturing capacity, scientific facilities, or critical data, the structure changes. Neither side is independent, but dependence has become reciprocal.

Power comes not from escaping the network but from occupying a strategically important place inside it. Seen this way, China’s long-term scientific goal may not simply be to reduce dependence on the world. It may be to make the world more dependent on China. That distinction deserves to be brought into the discussion of Ahlers’s book because it follows logically from her decision to analyze China in global science, rather than Chinese science as a separate system.

Where the Argument Needs More Pressure

This is also where Ahlers’s framework deserves its hardest test. An emphasis on integration risks underestimating asymmetry. Science may be globally connected without all participants enjoying the same freedom to publish, collaborate, investigate, or criticize. A system can adopt the formal institutions of global science — journals, universities, peer review, laboratories, international collaboration — while embedding them in a political order that sets different boundaries on inquiry.

The question is therefore not merely whether China has adopted global scientific norms. It is which norms travel successfully, which are modified, and which encounter political limits. Peer review can be imported. Laboratory organization can be imported. Publication metrics can be imported. But what about academic freedom, institutional independence, the freedom to pursue politically inconvenient questions, the ability to share certain forms of data, or the right to collaborate without strategic screening?

These are not incidental details. They help determine what kind of “global science” is actually emerging.

So the most difficult question for Ahlers’s thesis is not whether China belongs to global science. Clearly it does. The more demanding question is: What happens to the norms of global science when one of its largest centers operates under a different political settlement from the one in which many of those norms historically developed? That is where the book’s argument could become genuinely consequential.

From the Globalization of Chinese Science to the Securitization of Global Science

The problem becomes even harder because the political environment around science is changing at the same time. One of the governing ideals of global science is the circulation of knowledge. One of the governing principles of the modern state is that strategically important capabilities cannot be allowed to circulate without limit. For a long time those two principles could coexist with relatively little friction in many scientific fields.

AI, quantum technology, semiconductors, biotechnology, communications, and advanced materials make the separation harder. In these fields, the distance between scientific knowledge and strategic capability can be very short. The MI5 warning illustrates the difficulty. If researchers themselves are participating in what they believe to be legitimate scientific cooperation, at what point does research collaboration become a national-security problem?

That is no longer a marginal policy question. It is becoming part of the institutional environment in which global science operates. The effect may be profound. Research-security boundaries used to be imagined as lines around institutions or countries. Increasingly, they may run through individual laboratories and even individual projects. The same university may collaborate in one area and face restrictions in another. The same researcher may publish one dataset openly while another is controlled. The same foreign partner may be welcomed in climate research and treated cautiously in quantum communications.

The future borders of science may therefore no longer be drawn primarily on maps. They may be drawn inside knowledge itself.

Why Ahlers’s Timing Matters

This is what makes the timing of China in Global Science unusually good. The book appears just as two historical processes collide. China’s integration into global science has become too successful to treat the country as an outsider, while geopolitical rivalry is making states less comfortable with the idea that scientific openness can remain politically neutral.

Ahlers’s emphasis on integration therefore arrives precisely when integration is being politically renegotiated. This gives the book an importance larger than its compact Cambridge Elements format might suggest. The value of the book is unlikely to lie in producing an encyclopedic history of Chinese science. Its promise lies in changing the frame through which that history is interpreted.

Instead of asking whether China has a scientific system comparable to “ours,” it asks what happens when the distinction between “Chinese science” and “global science” itself becomes unstable. That is a much larger question.

Openness Will Not Simply Disappear

Nor should the present security turn be mistaken for inevitable scientific decoupling. Some scientific problems remain so large that international cooperation is not merely desirable but structurally useful. Fusion research is an obvious example. Large scientific facilities, long engineering timelines, huge costs, and distributed expertise all push in the direction of collaboration, while other fields carry more obvious dual-use risks.

The likely future is therefore neither universal openness nor complete closure. It is managed openness. Some knowledge will remain widely shared, some data will be restricted, some collaborations will require additional review, some infrastructures will remain multinational, and some fields will be treated almost as extensions of national security. Global science may remain global while becoming much more internally segmented.

That possibility strengthens Ahlers’s central insight rather than weakening it. If China is now one of the actors helping reshape the institutions of global science, then China’s own choices about openness, security, collaboration, and scientific governance matter well beyond its borders.

The Future of Global Science

One of the great ideals of modern science is that truth has no nationality. A formula that holds in Beijing should hold in Boston. An experiment does not become false because the researcher who conducts it carries the wrong passport. That ideal remains indispensable.

But scientific truth may be universal while scientific production remains political. Scientists work in institutions. Institutions need money. Laboratories need equipment. Databases have owners. Researchers need visas. Universities operate under law. Governments decide which technologies can move across borders.

Ahlers’s book matters because it places China precisely at the intersection of these two realities. China did not rise scientifically by escaping global science. It rose by entering it. And yet China’s rise is now one of the forces compelling that global system to reconsider the terms of openness on which its expansion rested.

This is the paradox at the center of China in Global Science: Global science helped make China a scientific power; China’s emergence as a scientific power is now helping transform what “global science” means.

The strongest part of Ahlers’s intervention is that it allows us to see both developments at once. China is neither an external force standing outside global science nor simply another participant whose rise changes nothing. It is a product of the global scientific order and increasingly one of the forces reshaping that order.

That makes “network autonomy” a useful extension of the book’s argument. The most powerful scientific state in the next era may not be the one that produces everything domestically and depends on no one. It may be the one that remains deeply embedded in global networks while occupying enough indispensable nodes that others cannot easily bypass it. For China, that would mean moving beyond the defensive ambition of reducing dependence toward the more consequential ambition of creating mutual dependence.

For global science, it means something larger. During the Cold War, the borders of science were drawn largely by the Iron Curtain. In the age of AI, those borders may be drawn inside every laboratory.

Subscribe now

The emotion a society should fear most is not necessarily anger. Anger, at least, suggests that people still believe something can change. They complain because they still expect a response; they protest because they still think their actions might produce consequences; they argue with institutions, workplaces, families, even fate itself because they still see themselves as participants in the game. A deeper crisis begins when people stop arguing and stop expecting. They continue going to work and living their lives, outwardly following the rules, while gradually withdrawing their investment in the future. Entrepreneurs invest less. Young people postpone marriage and children. Employees stop chasing promotion. Students lower their ambition to “just don’t make mistakes.” Civil servants become less proactive. The middle class saves more and spends less. Intellectuals retreat from public expression. Each of these choices, taken separately, may have an entirely reasonable personal explanation. But when many of them begin moving in the same direction—less investment, less risk, smaller ambitions—they cease to be merely private choices. They begin to form a social psychology.

Calling this condition “decadence” is not quite right. The word suggests laziness, indulgence, or moral decline. What deserves closer attention today is something more like defensive withdrawal. People have not become less rational. If anything, they are calculating risk more carefully, and it is precisely this calculation that leads them to invest a little less, expect a little less, and expose themselves a little less. The deeper question, then, is not simply why people no longer work as hard. It is this: under what conditions does a society make doing less begin to look like the most rational choice?

I. What Changes People Is Not Hardship, but the Relationship Between Effort and Outcome

Poverty by itself does not necessarily produce resignation. Many societies in history endured conditions far harsher than those of today while retaining a powerful drive for advancement. What matters more is how people understand the relationship between action and outcome. If a young person believes that studying hard will probably lead to a better job, that working hard will probably lead to higher income, and that higher income will improve housing, marriage, and family life, then even a difficult present can still justify continued effort. What psychology calls a “sense of control” does not mean that people actually control everything. It means they believe their actions can meaningfully affect the outcome.

Once that relationship becomes uncertain, the psychological structure changes. Working three extra hours may not improve one’s chances of promotion. Getting into a good university may no longer guarantee a career commensurate with the investment. A family may save for years and still find that the pressures of housing, education, and retirement have barely eased. A small-business owner considering another round of investment may think first about future uncertainty rather than future profit. People do not immediately rebel under such conditions. More often, they first adjust their expectations. The social mindset shifts from “How do I win?” to “How do I avoid losing?”

The question once was: “What opportunity should I seize?” It becomes: “How do I keep from falling?” When growing numbers of people move from an opportunity mindset to a risk mindset, behavior changes systematically. Consumption weakens and saving rises. Fewer people start businesses, while more seek civil-service jobs. Job changes become more cautious and career choices tilt toward stability. Marriage and childbearing are recalculated. Parents may gradually shift from asking, “What kind of person do you want to become?” to warning, “Whatever you do, don’t fall behind.” This is more than pessimism. It is a form of psychological contraction in a risk-conscious society.

II. “Lying Flat” Is Less About Rejecting Success Than Losing Faith in the Traditional Path to It

In recent years, phrases such as tang ping—“lying flat”—along with “giving up” and “Buddha-like” detachment have become common descriptions of younger people in China. It is easy to interpret these as signs of changing values: young people are afraid of hardship, unwilling to strive, unwilling to endure. That explanation is too shallow. Many of those said to be “lying flat” have not lost their desires. They have simply recalculated the old model of success built on high investment, intense competition, and delayed gratification.

Suppose the social contract a person grows up with runs like this: study hard, get into a good university, find a good organization, work hard, buy a home, marry, have children, and gradually move upward. For this sequence to remain credible, each step must have a reasonably stable relationship to the next. When returns to education, employment security, housing costs, child-rearing costs, workplace competition, and social mobility all begin to shift at once, younger people naturally reassess the path. Refusing marriage, delaying children, declining to buy a home, avoiding entrepreneurship, or losing interest in promotion may in many cases reflect not a disappearance of desire but a decision to exit a game whose perceived return on investment has fallen.

Sociology has a useful term here: anomie. It does not simply mean that society has no rules. It can also describe a condition in which the rules remain, but the reward structure that once made them persuasive begins to lose credibility. People still hear the old story of success, but find it harder to verify through their own lives. A peculiar psychological contradiction follows: society continues to demand upward mobility, while individuals become less certain that a viable path upward still exists. The safest response is to lower one’s desires.

Psychologically, this is a familiar form of self-protection. If I no longer expect promotion, I cannot be hurt by failing to receive it. If I do not marry, I avoid the risk of a failed marriage. If I do not have children, I avoid the pressures of education and parenting. If I do not start a business, I cannot lose my capital. If I do not speak, I do not bear the cost of speaking. Every step backward buys a little security. But when an entire society begins to move this way, the macroeconomic and social consequences are very different.

III. The Real Danger Is Not “Lying Flat,” but a Low-Investment Society

A society can absorb a certain number of disengaged people. Every era has pessimists, failures, and people who choose to exit. The deeper danger begins when low investment becomes a rational strategy.

Economics usually discusses investment in terms of capital. Social psychology must also examine another kind of investment: whether people are willing to commit their time, emotions, trust, and creativity to the future. An entrepreneur deciding whether to expand production is investing. A young person spending ten years mastering a profession is investing. Two people deciding to marry are making an emotional investment. A family choosing to have a child is investing in society twenty years from now. A civil servant who voluntarily does more than the minimum is making an organizational investment. An intellectual willing to take risks in expressing judgment is making a public investment. All of these actions rest on the same premise: I believe the future is worth what I am giving up today.

For this reason, a society’s deepest confidence is not revealed by how often people say they are confident. It appears in long-term behavior: whether people buy long-term assets, whether they have children, whether they dare to start companies, whether they are willing to study something that may take ten years to pay off, whether they trust strangers, and whether they are willing to assume responsibility beyond the minimum required by an organization. If all of these forms of long-term investment begin to decline together, a society gradually moves from a developmental psychology toward a preservation psychology.

This runs deeper than anger. Angry people still want to change their environment. Preservation-minded people first want to protect themselves. Their guiding principles become: do not make mistakes, do not take on debt, do not step forward, do not assume extra responsibility, do not be the last person to sign the document, do not tie your fate to any long-term commitment. Inside organizations, this produces a familiar bureaucratic logic: the less you do, the fewer mistakes you make; if you do nothing, you make none. In business, it becomes: survive, but do not expand. In family life: wait before having children. In public life: I know, but I will not say. When everyone is reducing exposure, society may even look unusually calm. But that calm is not necessarily stability. It can look more like a system running at a permanently lower temperature.

IV. Why the Middle Class and the Young Are Especially Vulnerable to a Psychology of Withdrawal

Social psychology is never evenly distributed. The people most vulnerable to withdrawal are often neither the poorest nor the richest, but those with relatively high expectations, substantial education, and a growing sense of uncertainty about the future. This is why the middle class and highly educated young people matter so much.

The poor have less to lose to begin with. The wealthy have stronger asset cushions. The defining psychological feature of the middle class is that it has built a way of life that must be continuously maintained: housing, education, careers, retirement, health care, children, and social status. A serious shock to any one of these can create a palpable fear of downward mobility. What the middle class fears most may therefore not be poverty itself, but losing the life it has already secured.

That produces intense status anxiety. When a society’s sense of opportunity declines, the middle class often does not become radical; it becomes conservative. It saves more, reduces leverage, starts fewer businesses, seeks institutional security, pushes children toward more stable careers, and retreats from public expression. Younger people face the opposite problem. They do not yet possess stable assets or social status, but they can already see the cost of acquiring them. The logic of withdrawal therefore differs across generations: middle-aged people fear losing what they already have, while younger people wonder whether the entire package is worth entering in the first place. The result is the same: a lower tolerance for risk.

This helps explain a seemingly strange phenomenon: individuals may become better educated while society as a whole becomes more cautious. Education improves people’s capacity for calculation, but it also improves their capacity to identify risk. Seeing more possible paths to failure does not automatically make a person braver. Sometimes it makes them more careful.

If there is a form of social resignation in China today that deserves serious study, its core is not that young people have suddenly become lazy, nor that society has somehow lost its work ethic. The deeper change is that more people are reassessing the relationship among effort, risk, and reward.

This cannot be solved by exhortation. If experience has taught people that additional effort does not produce commensurate results, telling them to “work harder” will usually deepen their alienation. What social psychology needs restored are three things.

The first is predictability. People do not demand a future without risk. They need to believe that the rules are broadly stable, and that long-term decisions made today will not be rendered meaningless tomorrow by changes they could not reasonably anticipate. The second is fairness. People can accept that others are more successful than they are. What is much harder to sustain is the feeling that outcomes depend less on ability, effort, and rules than on forces they cannot understand or influence. Once fairness declines, cooperation and investment decline with it. The third, and most important, is efficacy. People need to believe again that if they do a little more, it can make some difference; that learning increases ability; that work can improve income and career prospects; that entrepreneurship has at least a reasonable chance of success; that expression can enter normal public discussion; and that raising children still means believing those children will have room to grow.

This is a kind of social efficacy. If a society can restore it, people will begin investing again on their own. They do not need to be endlessly told to strive, because durable effort has never been manufactured by slogans. It grows out of a simpler conviction: between today and tomorrow, there is a road I can still help build.

There is a proposition worth taking seriously: a society’s deepest crisis may appear not when more people openly oppose it, but when more people quietly reduce their investment in it. Yet the argument can be pushed one step further. The real danger is not that people have physically left, nor that they have stopped speaking. It is that they remain in their workplaces, families, cities, and society, while psychologically removing the future from their own life plans.

They still work. They still consume. They still follow the rules. But they make fewer long-term decisions, take fewer risks, and believe less in the possibility of change. At that point, a society can begin to experience a form of decline that conventional statistics are poorly designed to detect. GDP may still be growing. The cities may still be brightly lit. Offices may still be full. Yet more and more people begin to as：Why bother?

Those two words are what a society should fear most. Anger can still lead to action. But when “Why bother?” becomes the philosophy of a generation, what society loses is not merely a source of growth. It loses one of its most valuable forms of psychological capital: the willingness to invest in the future.

Subscribe now

If intellectual history were a fair contest, we might believe a reassuring story: good ideas are eventually recognized, great works find their audience, and useful technologies make their way into the world. Civilization may take detours, but given enough time, gold will shine. David Byrne’s new book, Sleeping Beauties: Why Good Ideas Go Dormant and How They Wake Up, is about why this story is so often false.

Byrne, one of the founders of Talking Heads, has always been difficult to describe simply as a musician. He has worked in film, theater, visual art, and record publishing, and is also the author of How Music Works. His new book, published in the United States by Penguin Press on October 6, 2026, begins with a deceptively simple question: why do some discoveries, works of art, and technologies now considered important attract little notice when they first appear, sometimes even ridicule, only to return decades or centuries later?

“Sleeping Beauty” was already a term in scientometrics before Byrne adopted it. He was inspired by a 2004 study by Anthony van Raan of Leiden University. The history of science contains a curious pattern: some papers receive almost no citations for years, then suddenly enter a new research network and begin to attract attention. Knowledge does not spread at a constant speed. A discovery can exist without yet entering the field of vision of its age. Byrne’s insight was to extend this pattern beyond scientific papers. Art, literature, architecture, medicine, energy, infrastructure, even entire ways of life can suffer a similar fate.

He examines more than thirty cases: continental drift and dark matter, handwashing and solar power, Bach, Bruegel, Vermeer, Melville, Gaudí, literary works, and ancient water-management systems in southern India. They differ widely, but they point toward one uncomfortable conclusion: great ideas do not automatically win. Whether an idea is right and whether an age is able to see it are two different questions.

I

The most interesting thing about Byrne’s book is that it turns what we usually treat as a problem of intellectual quality into a problem of intellectual ecology. We normally ask whether an idea is good. Byrne pushes us to ask who can hear it, who can understand it, whether the age has a language for describing it, whether it threatens established interests, and whether publishers, universities, laboratories, museums, markets, or media can carry it to a wider audience. Is there someone willing to spend years arguing for it? More important, has the age itself produced a problem that suddenly makes the idea necessary? Remove enough of these conditions, and even a powerful idea may remain asleep.

Gaudí offers an easy example. It is hard for anyone visiting Barcelona today to imagine that, for a long time after his death, he was not the “modern architectural genius” we now take for granted. His buildings were once dismissed as grotesque, even ugly. Later, artists such as Salvador Dalí became advocates; aesthetic standards changed; photography, tourism, and cultural institutions expanded his visibility; UNESCO eventually incorporated his work into the World Heritage system. Gaudí did not rise from the grave to redesign the Sagrada Família. What changed was the world looking at him. Sometimes the idea has not changed. The world around it has.

Byrne offers an even more powerful example from southern India. Under the Chola dynasty, sophisticated systems of water management developed around tanks, wells, reservoirs, and forms of local communal governance suited to monsoon conditions. Over time, much of this knowledge and institutional practice receded from the mainstream. Today, as water shortages and flooding have become urgent again, some of these systems and practices are being restored. Byrne has discussed this case in recent interviews, and it led him toward a larger realization: knowledge is never as “global” as we imagine. The fact that something is known in one place does not mean it is known elsewhere. Nor does being written down mean that knowledge remains alive.

Modern people often imagine history as an ever-expanding database: humanity knows more and more, so later generations must naturally know more than earlier ones. Byrne reminds us that the record is messier. Knowledge accumulates, but it is also lost. It advances, but it can also recede. A technology may be mature and still vanish from social life because economic conditions, institutions, or power relations have changed. History is not an endlessly expanding hard drive. It is closer to a forest: things grow and die; seeds can remain underground for years and return only when temperature, rain, and soil have changed.

Seen this way, Chinese intellectual history becomes especially interesting, because China may contain one of the world’s largest archives of “Sleeping Beauty” ideas. The clearest example is Mohism.

During the Warring States period, Mohism was a major intellectual force. It argued for impartial concern, opposition to aggressive warfare, and the elevation of the worthy, while also developing sophisticated ideas about argument, language, knowledge, and reasoning. Yet after the Qin and Han dynasties, Mohism gradually disappeared as an organized school, and large portions of the Mozi went unread for long periods. Some texts nearly vanished altogether. The fact that we can still read later Mohist writings on logic and knowledge is itself partly the result of historical accident: the complete text survived within the Daoist canon before returning to scholarly attention. During the Qing, scholars such as Bi Yuan, Sun Xingyan, and Sun Yirang repeatedly collated and edited the text. With the publication of Sun’s Mozi Jiangu in 1894, Mohism increasingly reentered modern Chinese intellectual history as a tradition available for serious reconsideration. Modern scholarship has also shown that, as late-Qing intellectuals encountered Western science and logic, they began rereading Mohism through a new set of questions.

What happened cannot simply be called the “discovery of Mozi.” Mozi had never disappeared. What disappeared was a capacity and a need to read him. In a society preoccupied with classical orthodoxy, the examination essay, and the established canon, parts of the Mo Jing could seem strange or marginal. By the nineteenth century, when Chinese intellectuals were suddenly forced to confront science, logic, state competition, and the question of why the West had become powerful, the same texts began to mean something different. The text had not changed. The questions had. Once the questions changed, certain parts of the text lit up. Intellectual revival, then, is rarely just excavation. It is often the present reinventing the past.

Wang Fuzhi followed a similar “Sleeping Beauty” trajectory. Living through the Ming-Qing transition, he left behind a vast body of philosophical and historical writing, but for a long time his works did not circulate on a scale commensurate with his later reputation. An edition of the Chuanshan Yishu appeared in 1839, and in 1865 a major Jinling edition sponsored by Zeng Guofan greatly expanded their circulation. Why then? To say merely that Wang Fuzhi was a great thinker and was finally discovered explains almost nothing. As late-Qing China entered an age of crisis, intellectuals began searching again for resources with which to think about the state, history, civilizational boundaries, institutions, rise and decline. Printing and publication provided material conditions, local elites and political figures served as intermediaries, and national crisis created a new need to read. A seventeenth-century thinker was reread by the nineteenth. This was not simple transmission. It was a new pairing: old ideas meeting new problems. That is when the Sleeping Beauty wakes.

II

For an idea to wake, I think at least five conditions are necessary.

The first is preservation. An idea must survive long enough to reach the future. If the Mozi had disappeared entirely, everything that followed would have been impossible. A book, a manuscript collection, a building, a water system, even an oral tradition needs some form of carrier. Intellectual history remembers authors and often forgets copyists, collectors, collators, printers, and translators. Yet they decide which ideas retain the possibility of a second life.

The second is a problem. Ideas do not wake by themselves. Usually reality creates the problem first. Solar technologies existed long before changes in energy prices, environmental pressure, and technical conditions gave them a new position. Local water-management practices returned to attention not because modern people suddenly became nostalgic but because water once again became a pressing problem. Changes in economic and social conditions can make possible what once had no place to stand.

The third is an interpreter. Beside every Sleeping Beauty there is often someone who wakes her: an editor, scholar, artist, collector, or stubborn amateur. Kafka had Max Brod; Gaudí was later reassessed by admirers such as Dalí. Byrne himself belongs to this category. Early in the book he writes about the Brazilian experimental musician Tom Zé. Through Luaka Bop, the label Byrne helped establish, he helped bring a musician who had largely fallen from international view to a new audience. One of Byrne’s most interesting suggestions is to pay attention to the young, the obsessive, and the amateur. Professional systems contain a paradox: they are often very good at judging what already fits established standards, and therefore sometimes very bad at recognizing what falls outside them.

The fourth is a network. One person can discover an idea, but one person rarely makes it live again. An idea has to enter publishing houses, universities, schools, museums, media, scholarly communities, markets, or public debate. It has to be cited, translated, taught, disputed, and rewritten. Without those networks, “rediscovery” remains a private event. This is why Byrne’s book, though ostensibly about creativity, is at a deeper level about institutions. We like talking about genius but much less about the infrastructure that makes genius visible. Ideas have infrastructure too: libraries, universities, publishers, translators, prizes, journals, search engines, and today’s algorithmic platforms. They seem merely to distribute ideas, but in practice they decide what enters public vision and what remains asleep.

The fifth condition is reinterpretation. Once a Sleeping Beauty wakes, she is rarely the same Sleeping Beauty. Mohism could not reenter twentieth-century China exactly as it had existed in the Warring States period. Modern readers became interested in Mohist logic, utilitarian tendencies, pacifism, and supposed scientific spirit; those questions themselves came from the modern world. Wang Fuzhi’s reentry into late-Qing thought likewise placed him within new concepts of nation, state, and historical consciousness. Intellectual revival is therefore not A → A. It is closer to this: A enters a new world, B, and becomes A′. What can be restored unchanged is usually an artifact. An idea capable of participating in the present will almost always pass through translation, selection, and reinterpretation. Civilizational continuity, seen this way, is not simply preservation. It is a capacity for continued generation.

This is also where Sleeping Beauties invites a further question. Byrne begins from a seductive premise: many things later recognized as important were once wrongly ignored. But the proposition cannot simply be reversed. To have been ignored is not the same as to have been important.

The history of forgotten ideas is vastly larger than the history of ideas later revived. Most probably deserved to be forgotten. Looking back at figures such as Gaudí, Kafka, or Mendel, who eventually achieved recognition, we are easily drawn into survivorship bias. Because they now matter so much, we assume those who failed to recognize them must have been foolish. But historical actors did not know the ending. Every age produces thousands of new ideas. Who can know which one is the Gaudí of a century hence and which one is simply a bad architect?

This helps explain why, although Publishers Weekly reviewed the book favorably, it found Byrne’s later attempt to generalize the conditions under which “sleeping beauties” reawaken less compelling than the case studies that precede it. The richer the examples, the harder it becomes to derive a universal formula. There is also a more difficult problem: ideas that wake are not necessarily good ideas. As The Guardian observed, rediscovered ideas can be progressive, but they can just as easily be reactionary.

Nationalism can return. Racism can return. Religious extremism can return. Political myths that once receded from the center of history can suddenly recover a mass audience. The internet is especially good at this. It is at once the largest idea-awakening machine in history and a vast séance chamber for dead ideas. “How to wake the Sleeping Beauty,” then, is really two questions: How do we prevent valuable ideas from being wrongly forgotten? And how do we decide which ideas deserve to be awakened? Nostalgia is not enough. Neither is cultural pride.

III

For China today, this distinction matters a great deal. Over the past several decades, one phrase has become increasingly familiar: we should search traditional Chinese culture for intellectual resources. The problem is that “searching for resources” can easily turn into an effort to prove that our ancestors already knew everything.

Modern science develops a concept, and we search ancient books for a similar phrase. Modern politics develops an institution, and we look for an ancient Chinese analogue. Environmentalism becomes influential, and we set out to prove that the ancients already possessed a complete ecological philosophy. Artificial intelligence arrives, and suddenly people begin discovering “algorithms” in the Book of Changes and the Dao De Jing. This may look like an awakening of tradition, but it often leaves tradition asleep, because ancient thought has not actually been brought into contact with present problems. We have merely found ancient relatives for modern concepts.

A more worthwhile Chinese “Sleeping Beauty project” would proceed in the opposite direction. It would begin by asking what new problems we now face. Artificial intelligence forces us to reopen the question of what a human being is. Climate change forces us to rethink the relation between humans and nature. Population aging makes us reconsider the family, intergenerational obligation, and care. New frictions in globalization force us to revisit ideas of tianxia, the state, borders, and civilization. The growing complexity of technological society makes us return to problems that Mohist thinkers once took seriously: utility, knowledge, reasoning, organization, and opposition to war. Only then should we go back into history and ask whether certain ideas fell asleep because the social conditions that sustained them disappeared, but might now acquire new meaning under changed circumstances.

In that sense, Mohism is worth bringing back not because the Mohists discovered something two thousand years before the West. Wang Fuzhi is worth bringing back not because seventeenth-century China already possessed answers to twenty-first-century problems. They matter because they allow us to ask today’s questions differently. One approach turns tradition into a certificate of ancestral glory. The other turns tradition into an intellectual resource. And an intellectual resource becomes a resource only when it is used.

This is the larger question Byrne’s book raises for China: how many ideas disappeared not because they had been disproved, but because the problems, language, institutions, and networks that once sustained them disappeared? The answer probably extends far beyond Mohism and Wang Fuzhi, and far beyond canonical texts. Such ideas may be buried in local governance, vernacular knowledge, engineering practice, agricultural technology, religious life, educational forms, and older arrangements between human communities and the natural world. Most may not deserve revival. Some should remain in the past. But a small number will almost certainly acquire new significance under new conditions. The difficulty is whether we still possess the ability to recognize them.

In an interview with The Observer, Byrne made a point worth keeping in mind: much knowledge remains local. We tend to assume that once something is known somewhere, “humanity knows it.” It does not. Even in the internet age, knowledge can still be ignored, isolated, and forgotten. The same is true of civilizations. China may possess five thousand years of texts without possessing five thousand years of living thought. Libraries preserve words, not understanding. Museums preserve objects, not intellectual life.

That distinction matters because preservation is not the same as renewal. If “cultural revival” means only putting the past back on display, it can preserve a civilization’s remains without making them think again. A living tradition has to undergo a more difficult process: the past must be brought into the present, exposed to questions it was never designed to answer, translated into unfamiliar vocabularies, and altered by the encounter. What emerges is no longer the old idea in its original form. It is something closer to A′: carrying an older inheritance, but made possible by a new historical world.

This is why the vitality of a civilization cannot be measured only by how many great ideas its ancestors once produced. The more demanding question is whether it can still recover what it has forgotten, distinguish what deserves recovery from what does not, and allow old ideas to generate new ones. That requires more than reverence. It requires archives that preserve heterodoxy, institutions that permit reinterpretation, translation networks that reconnect ancient and modern, China and the outside world, and young people willing to pursue unfashionable questions without first proving that the ancients had already answered them.

China does not lack sleeping ideas. What it may lack are the mechanisms that allow them to wake without turning them into relics, slogans, or evidence of ancestral superiority. A civilization does not begin to age only when it stops producing new ideas. It may also begin to age when its old ideas can no longer wake into a new world.

Subscribe now

In 2026, Palantir Technologies began recruiting students straight out of high school. They do not need a college degree first. They enter the company directly and work in teams on technical problems. The program lasts only four months and is small in scale; it is hardly a substitute for the university. Yet it raises a question that would have sounded almost indecorous not long ago: if an eighteen-year-old can already use AI, write code, build products, and contribute to real projects, should that person’s ability be certified by a diploma earned four years later—or by what he or she can already do?

Once that question is asked, the old arrangement begins to look less inevitable. Universities have never been only places where knowledge is taught. They are also institutions of sorting. A student enters a prestigious university, remains there for four years, earns a degree, and employers infer a great deal from the sequence: intelligence, discipline, persistence, trainability, the ability to function within an institution. The degree is both an educational credential and a social proxy for talent. AI is beginning to weaken the connection between the two.

Someone without a computer-science degree can now use AI to write code. Someone without a design degree can build a workable prototype. A high-school student can call on tools that, until recently, belonged largely to trained professionals. Employers will naturally begin to ask: if I can see what you can do, why must I first know where you spent four years studying? Experiments like Palantir’s matter for this reason. They may not threaten the university as an institution, but they do unsettle one of its oldest temporal arrangements: first learn, then work; first acquire credentials, then enter the world. AI is beginning to scramble that sequence.

I

The modern university still bears the imprint of the industrial age. Students arrive at eighteen and choose a major. Computer science is one thing, history another, economics another, psychology another. Each field is subdivided into courses; each course carries credits, meets on a schedule, ends in an examination, and contributes to the total required for graduation. Knowledge is broken into units before it is handed to the student.

The world outside the university is organized differently. A company designing an AI health product for older people does not face a computer-science problem alone. It faces questions of software, medicine, psychology, design, ethics, regulation, business models, and aging. A city trying to address housing cannot assign the problem to a “housing major.” Housing involves finance, land use, transportation, law, demography, politics, and community life. Problems arrive whole. It is institutions that divide them.

Disciplines remain indispensable. Without mathematics there is no serious engineering. Without historical training, it is difficult to understand how a society arrived where it is. Without philosophy, value judgments can easily disguise themselves as technical judgments. The trouble begins when disciplines cease to be ways of seeing and become walls for organizing education. Students spend four years inside those walls and may encounter their first genuinely open-ended problem only after graduation.

AI makes this arrangement increasingly awkward because one of the first things it lowers is the cost of calling up knowledge. Faced with a statistical problem, a student once had to take a course before knowing where to begin. Now AI can explain a model, generate code, and demonstrate a method in minutes. Crossing disciplines once required learning an unfamiliar language; machines can now move among knowledge systems with startling speed. None of this makes knowledge less important. It changes when and how knowledge enters the process of learning.

The older sequence was: learn first, solve later. The emerging one may be the reverse: enter the problem first, then bring in the knowledge it demands. One is organized around courses; the other around problems. More than a century ago, John Dewey was already thinking in something like these terms.

In Chinese, pragmatism has long been translated as “实用主义,” an established rendering that cannot simply be dismissed as wrong. Yet the word easily invites a misunderstanding of Dewey as a philosopher of usefulness—as though his position amounted to “whatever works is true.” In education, that misses the point. Dewey’s pragmatism is better understood as a philosophy of practice.

He was interested in how knowledge takes shape through action, experience, error, reflection, and renewed action. Ideas, for Dewey, are not mirrors held up to a world outside us. They are instruments of inquiry, ways of intervening in a situation in order to understand it. Knowledge is not something completed in advance and then applied; it develops through the attempt to deal with a problem.

This is why “learning by doing” is too neat a summary. Doing, by itself, teaches very little. A person can repeat the same task for twenty years and learn almost nothing from it. Dewey’s model is recursive: encounter a problem, form a question, gather material, propose a hypothesis, act, observe what happens, revise the interpretation, and act again. Learning resides in that return.

His Laboratory School in Chicago mattered for the same reason. Children were not treated as containers to be filled. Activities were arranged as forms of inquiry. Teachers did not merely transmit answers; they created situations in which students had to notice a problem, make sense of it, and bring knowledge to bear. More than a century later, AI makes Dewey look unexpectedly contemporary.

Once knowledge becomes easier to obtain, education shifts toward harder questions. Can a student recognize a problem worth pursuing? Can she tell trustworthy knowledge from unreliable knowledge? Can she bring different forms of knowledge together? Can she turn an idea into something that works? When it fails, can she understand why? Can one experience change the way she acts the next time? AI can assist at every stage. It cannot substitute for the process itself.

II

What universities should learn from Silicon Valley is not that students should stop going to college. That is the most obvious conclusion and the least interesting one. Peter Thiel has long encouraged some young people to step outside conventional higher education and build companies; Palantir is now experimenting with recruiting talent directly from high school. Meanwhile, another kind of experiment has been taking place within the university itself. At Olin College of Engineering, students begin designing, building, and testing systems early in their education. Engineering, mathematics, science, the humanities, social sciences, and entrepreneurship are brought into the same curricular field, and students work on sustained projects oriented toward problems beyond the classroom.

Dewey, Silicon Valley, and these newer educational experiments may seem to belong to different worlds. What links them is a question of timing: why must learning always precede reality?

The traditional university assumes a sequence:

learning → graduation → work → problem-solving

A university organized around practice might operate differently:

problem → learning → action → failure → more learning → renewed action

The difference is not cosmetic. It alters the relation between education and life.

Today, the course remains the basic unit of higher education. In the future, the project may become equally important. Instead of saying, “This semester I took Economics 101, Python, and social psychology,” a student might say, “This year we studied the housing problem facing young people in New York.” To do that well, she would need statistics, real-estate finance, urban planning, interviews with tenants, housing law, AI-assisted data analysis, and the ability to explain her findings to city officials, investors, and residents.

Economics would still be there. So would statistics, law, and programming. But they would no longer serve primarily as destinations. They would become resources drawn into a problem. The shift sounds modest until one notices how much of the university it rearranges.

A serious version of this model would require rethinking the four-year undergraduate curriculum. First-year students need not be hurried into disciplinary boxes before they have seen much of the world. A better beginning might be a set of broad domains—human beings, society, nature, technology—through which mathematics, writing, history, scientific method, statistics, and AI tools are encountered in use rather than in isolation.

Students should build things, but they should also read difficult books. They should write code and long essays. They should work with AI and, at other times, demonstrate that they can think without it. A university organized around practice must not become a training center for whatever the labor market happens to need this year.

If higher education is reduced to “teach what companies want,” it may become narrower than the professional education it hopes to replace. Companies are designed to solve present problems. Universities also preserve forms of knowledge whose value may not become visible for decades. Basic science, philosophy, literature, history, and fields of inquiry that appear economically useless are precisely what cannot be left to corporations.

Dewey’s idea of practice was never merely vocational. He understood education as growth, and as preparation for democratic life. A university should cultivate people who can perform tasks, but also people who can decide which tasks are worth performing.

That is why the university of the AI age needs to hold together two tendencies that are often treated as opposites: more practice and deeper humanistic education. A student might build an AI robot in the morning and discuss Aristotle in the afternoon; solve a supply-chain problem while asking what the system does to workers; train a model while studying bias. Students should learn how technologies are made, but also when some technologies should not be made.

Olin is instructive here because project-based learning has not required abandoning the humanities or social sciences. Engineering, design, entrepreneurship, and humanistic inquiry coexist within the same structure. Its emphasis on “human-centered engineering” changes the question from “How do we build this?” to “Who is this for, and what will it do to them?” That is closer to Dewey than the rhetoric of disruption usually associated with Silicon Valley.

Practice is not the opposite of theory. It is where theory is forced to answer to the world.

III

The role of the teacher changes under such conditions. The most valuable professor was once, in part, the person who knew the most. AI weakens that advantage. Few professors can compete with a machine in the speed with which facts can be retrieved.

What matters more are other capacities: asking a good question, recognizing whether a student’s question is worth pursuing, knowing when to answer and when not to, noticing a hidden flaw in an argument, bringing people from different disciplines together, helping students understand why a project failed. The teacher becomes less a transmitter of information than a designer of inquiry.

Assessment must change as well. Exams and essays have long been the characteristic instruments of university evaluation. AI has unsettled both. A polished paper no longer tells us, by itself, how much of the underlying intellectual work belongs to the student and how much to the machine.

Universities will therefore have to distinguish between system performance—what a human working with AI can accomplish—and residual learner capability—what the learner can still do when the tool is removed. Both matter.

The world outside the university will use AI. Trying to prohibit it indefinitely would make little sense. But the university still has to know what remains when the machine is absent. Can the student read a difficult text, identify a problem, build an argument, distinguish stronger claims from weaker ones, reason without immediate assistance?

A student’s educational record may therefore come to resemble less a transcript than a portfolio of capacities: What problems have I worked on? What have I built? Where did I fail? How did I revise the work? Who did I collaborate with? What tools did I use? What can I still do without them? Degrees may persist, but their meaning could change—from substitutes for evidence of ability into indexes pointing toward that evidence.

The major may loosen for the same reason. Industrial-age universities ask eighteen-year-olds a deceptively simple question: What is your major?

It is a strange demand to place on someone who has barely begun to see the world. The AI age may call for a different question: What problem do you want to spend time on?

One student may care about aging, another about war, cities, children, energy, poverty, or machine consciousness. Problems pull disciplines together. Aging leads into biology, medicine, psychology, economics, ethics, and public policy. War leads into history, international politics, game theory, technology, geography, psychology, and economics. The question of AI and human beings runs almost immediately through computer science, philosophy, cognitive science, law, education, and sociology.

The world itself does not arrive divided into majors. Majors are one institutional way of arranging knowledge.

That is not an argument for abolishing disciplines. Without depth, interdisciplinarity quickly becomes a little knowledge of everything and mastery of nothing. The issue is sequence. The older model says: become an economist, then study the world. A different model might say: enter a problem in the world, then discover where economics is needed. The discipline shifts from identity to instrument.

IV

Silicon Valley, however, cannot be the university’s only teacher. It has its own characteristic fantasy: that anything capable of iteration can be rebuilt in the image of a startup.

Education is not software. The years between eighteen and twenty-two are not merely a period for processing human capital. Universities perform functions companies cannot easily absorb. They preserve knowledge that does not immediately make money. They give young people a temporary space in which they are not wholly defined by the market. They place physicists and poets inside the same institution. They give an eighteen-year-old time to change her mind. They preserve a place from which government, markets, and technology can themselves be criticized.

A fully corporate university might produce more efficient engineers while producing fewer people inclined to ask what the efficiency is for.

Palantir’s experiment is worth taking seriously for precisely this reason. It shows that companies are beginning to train for themselves some of the people they once expected universities to produce. The answer cannot be for universities to become Palantir. It is to ask again why universities exist.

Dewey reappears here in an unexpected way. More than a century ago, industrial society was taking shape, and schools increasingly resembled factories: students grouped by age, time divided into periods, knowledge divided into subjects, people ranked through examinations. Dewey tried to pull education back toward life. School, in his account, should not merely prepare students for a future life. It should itself be a form of life.

We now face another version of the same problem. AI is turning large amounts of knowledge into something like infrastructure, available almost on demand. One of the university’s historical functions was to concentrate scarce knowledge in one place. The professors were there. The library was there. The laboratories were there. Students had to come to campus in order to gain access.

That monopoly is disappearing. AI travels with us. Lectures can be watched anywhere. Papers can be retrieved in seconds. Languages can be translated almost instantly. Code can be generated on demand. The university is therefore forced to answer a question it once had little reason to ask: if knowledge is everywhere, why come to a university at all?

The answer cannot simply be that a university grants a degree. A stronger answer is that it brings people together to do what none of them could do alone. There are laboratories, mentors, projects, sustained collaboration, and the chance to begin again after failure. Someone makes you finish a difficult book; someone else asks you to turn an idea into something that survives contact with the world.

The university should no longer be understood primarily as a warehouse of knowledge. It should be a place where experience is constructed, inquiry is organized, and knowledge is tested against reality.

That shift would move higher education from majors toward problems, from courses toward projects, from learning first and practice later toward repeated cycles of inquiry and action, from credentials as proxies for ability toward work as evidence of it, from teachers as transmitters toward teachers as organizers of inquiry, and from keeping AI outside the classroom toward learning how to work with it without surrendering independent human capability.

This is not an argument for abolishing the university. It is an argument for asking, at a moment when knowledge is becoming cheap, what the university can still make difficult in the right way.

Dewey suggested one answer long ago. Education is not the transfer of finished knowledge from one head to another. It is the reconstruction of the capacity to understand and act through experience.

The industrial-age university divided the world into disciplines and asked students to master them one by one. The university of the AI age may need to reverse the direction: put students back into the world first. Let them encounter problems. Let those problems send them toward knowledge. Let knowledge enter action. Let action expose error. Let error return them to books, laboratories, teachers, and one another.

A university, then, would no longer be merely a place where a young person spends four years. It would become a verb again: to investigate, to try, to fail, to revise, and to try once more.

AI may make answers nearly frictionless. What it will not make frictionless is the moment before the answer exists. That is where education still begins.

Subscribe now

We often mistake strength for certainty: knowing what you want, acting without hesitation, recovering quickly from failure, keeping difficult emotions under control. But some of the behaviors we associate with weakness — chronic indecision, people-pleasing, anxiety, repeatedly changing direction — may have less to do with a lack of ability than with something more basic: a lack of internal clarity.

When you do not know what you want, every new possibility acquires unusual power. Someone starts a business, and perhaps you should start one too. Someone builds an audience online, and suddenly you wonder whether you should be making videos. Someone changes careers, and your own job begins to look inadequate. When you are unsure what you can accept in a relationship, a small violation leaves you not only hurt but uncertain whether you have the right to be hurt. When you have no clear sense of what matters next, every decision can begin to feel disproportionately consequential.

The exhausting part is often not the number of things demanding our attention. It is the absence of an internal order among them.

Clarity does not mean possessing a final answer about your life. Few people ever do. It means becoming more precise about a smaller set of questions: What am I feeling? What is this reaction about? What matters to me now? What am I willing to give up in order to protect it?

Clarity, in this sense, is less a revelation than a psychological practice.

Anxiety Often Grows Where Priorities Are Unclear

Modern life offers an extraordinary number of possible selves.

You can build a personal brand, start a side business, learn artificial intelligence, invest, retrain, travel, optimize your health, change careers, monetize a hobby or reinvent yourself altogether. Each possibility comes with its own advocates, success stories and algorithms eager to put them in front of you.

More choice does not automatically produce more anxiety. Research on “choice overload” is considerably more complicated than the popular version of the idea suggests. A major meta-analysis found no universal effect: having more options can be burdensome in some circumstances and useful in others. What matters is the context, including how difficult the decision is and how well people can distinguish among their options.

But there is a recognizable psychological problem here nonetheless. When people have many possible directions and no stable way to rank them, each option competes not only for attention but for identity.

Today you want to make videos. Tomorrow you decide you should learn A.I. The next day you think another degree might be the answer. Then you watch someone leave a conventional career to start a company and begin to wonder whether you have wasted years of your life.

What looks like a failure of discipline may actually be a failure of prioritization.

If you do not establish a hierarchy of value, something else will establish one for you. Algorithms will do it. Friends will do it. Parents will do it. Professional culture will do it. The result can be a strange form of busyness: considerable activity with diminishing clarity about what any of it is for.

The first task, then, is not to discover your single lifelong calling. That demand creates its own paralysis. A more useful question is narrower:

At this stage of my life, what deserves the largest share of my attention?

That is different from asking what is objectively best. Adulthood requires accepting a fact that contemporary culture often tries to conceal: our resources are finite. Time is finite. Attention is finite. Energy is finite. Choosing one thing usually means neglecting another, at least temporarily.

Clarity begins when that limitation stops feeling like failure and starts becoming a basis for judgment.

Hesitation Is Information

People are often embarrassed by their own ambivalence.

They want to leave a job but fear losing financial security. They want to end a relationship but cannot let go of the connection it provides. They want to begin something risky but fear failure. They want to say no to someone but worry about what the refusal will cost.

Then comes the second conflict: not only are they uncertain, they are ashamed of being uncertain.

But hesitation is often evidence that two legitimate needs are competing.

You may want professional growth and economic security. You may want independence and attachment. You may want to protect a boundary and preserve a relationship. The conflict becomes painful partly because we insist that one side must be false.

Often it is not.

Saying, “I want freedom, and I am afraid of losing security,” does not resolve the decision. But it changes the structure of the problem. What had been a vague state of anxiety becomes a conflict whose parts can be named.

Naming does not magically cure distress. It does something more modest and often more useful: it makes the conflict available for thought.

Instead of telling yourself not to overthink, you can ask better questions. What exactly am I afraid of losing? What am I trying to protect? If I make this decision and regret it, what kind of regret am I most afraid of?

Those questions cannot decide for you. They can, however, return the decision to you.

Emotions Carry Information, but They Do Not Issue Orders

Many people learn early to distrust certain feelings. Anger seems immature. Sadness looks weak. Jealousy feels shameful. Hurt is interpreted as oversensitivity. So the impulse is to suppress the feeling before asking what produced it.

A person can become very skilled at appearing calm while becoming increasingly unfamiliar with his or her own experience.

Psychological research provides good reason to take emotion seriously as part of cognition. Work in the “feelings-as-information” tradition has shown that affect can influence judgment and thought, although its meaning depends heavily on context and on what people believe the feeling is about. Feelings can provide useful information, but they can also mislead us when we attribute them to the wrong source. DOI

That distinction matters.

Anger can be a signal that something important has been violated, but it does not prove that another person has wronged you. Anxiety can indicate uncertainty or threat, but it does not establish that a situation is dangerous. Jealousy may reveal something you value or fear losing, but it is not a reliable moral judgment about someone else.

An emotion is evidence. It is not a verdict.

That leaves a middle position between suppression and obedience: listen to the feeling without surrendering the decision to it.

“I am angry” does not have to become “I must attack.”

“I am afraid” does not have to become “I must retreat.”

“I do not want to lose this person” does not have to become “I must remain in this relationship.”

A useful metaphor is not that emotions are commanders but that they are instruments on a dashboard. They tell you that something in the system deserves attention. Maturity lies partly in learning how to read the display without assuming that every warning light tells you exactly what to do.

Boundaries Define Your Actions, Not Other People’s

Many adults discover surprisingly late that they do not know how to say no.

They know perfectly well that they do not want to do something. What they do not know is whether they are entitled to refuse it. They fear disappointing someone, damaging a relationship, appearing selfish or losing an opportunity.

So they agree. Then they agree again. Resentment accumulates privately while compliance continues publicly.

Eventually, they may feel that everyone around them is taking advantage of them, even though the people around them have never been told where the limit lies.

Boundaries are often described as rules for how other people must treat us. A more useful definition is behavioral: a boundary tells you what you will do when a limit is crossed.

“You cannot speak to me that way” expresses a demand.

“If you continue speaking to me that way, I will end the conversation” defines a boundary.

The distinction is important because the second statement does not depend on controlling another person. It depends on controlling your own response.

Psychologists frequently describe appropriate boundaries as important to healthy relationships, though most clinical guidance also recognizes that setting them can produce discomfort and relational consequences.

That discomfort is part of the point. A boundary that costs nothing is easy. The harder question is whether you are willing to tolerate someone else’s disappointment in order to remain aligned with something you consider important.

The more secure a person becomes in that capacity, the less need there is to control everyone else.

Clarity Is Not Certainty

One of the most persistent mistakes people make is to wait for certainty before acting.

They want to be sure a new career will work before leaving the old one, sure a business will succeed before starting it, sure a relationship is right before committing to it. They want the information that would make uncertainty disappear.

For most important decisions, that information does not exist.

You cannot know what a job will become after three years. You cannot know what a relationship will look like after ten. You cannot reliably know how technology will reshape an industry five years from now.

If clarity required certainty, clarity would be impossible.

A better definition is this:

Clarity is not knowing what will happen. It is knowing why you are willing to take the next step.

You may still fail. You may change your mind. You may discover information you did not have before. You may eventually decide that the choice was wrong.

But there is a difference between making an uncertain decision for reasons you understand and making one because everyone around you appears to be moving in that direction.

Clarity does not remove uncertainty. It makes uncertainty more bearable.

Freedom Includes the Ability to Let Possibilities Go

Contemporary culture tends to equate freedom with an abundance of options. The more possibilities available to us, the freer we imagine ourselves to be.

But there is another form of freedom: the ability not to respond to every possibility.

A person with no stable hierarchy of values can be pulled endlessly by alternatives. Every opportunity becomes evidence of a life not being lived. Every success story becomes an accusation. Every path not taken begins to look like a mistake.

Real freedom therefore requires more than the ability to choose. It requires the ability to forgo.

You learn that some attractive opportunities are not your opportunities. Some relationships matter but cannot be sustained indefinitely. Some forms of success would cost more than they are worth. Some judgments from other people can be heard without being obeyed.

The clearer a person becomes, the quieter life can appear from the outside. Not because less is happening, but because fewer things are granted equal authority.

The world remains noisy. You simply stop believing that every voice requires an answer.

The Inner Coordinate System

When we are young, much of our identity is necessarily external.

Grades tell us whether we are doing well. Teachers and parents tell us whether we are good. Job titles later tell us whether we are successful. Income supplies another metric. Approval tells us whether we belong.

There is nothing pathological about this. Human beings are social creatures, and psychological theories of motivation have long treated autonomy and relatedness not as opposites but as basic human needs that coexist. Self-determination theory, for example, emphasizes autonomy, competence and relatedness as central to motivation and well-being. The problem begins when external evaluation remains the only coordinate system a person possesses.

Then praise produces confidence and criticism destroys it. Acceptance creates worth and rejection removes it. The self becomes almost entirely reactive.

Maturity involves gradually constructing another set of coordinates: What do I value? What am I willing to accept? What am I unwilling to accept? What risks are worth taking? How much of my life do I want to give to work? What kind of success would be valuable to me, and what kind would remain empty even if I achieved it?

These answers will change. They should change. A twenty-five-year-old and a sixty-year-old should not necessarily organize a life around the same priorities.

An inner coordinate system is not a rigid doctrine. It is a way of having somewhere to stand while revising your map.

Clarity Is a Way of Not Abandoning Yourself

Strength may ultimately have less to do with never feeling lost than with knowing what to do when you are.

You can be anxious without immediately allowing anxiety to choose your direction. You can be angry without asking anger to speak for you. You can be uncertain without treating uncertainty as evidence that you are incapable of deciding. You can say no and survive the fact that somebody may not like you for it. You can lack an answer without rushing to borrow one from somebody else.

Life is not a project in which ambiguity is eventually eliminated. Much of adulthood consists of making commitments while important things remain unresolved.

The task is to remain in contact with yourself while they do.

Psychologists might describe parts of this capacity in the language of autonomy, emotional awareness, self-regulation or agency. None of those ideas is identical to what I mean here. The simpler word may be enough.

Clarity.Not knowing all the answers.Knowing yourself well enough that uncertainty does not automatically hand your life over to someone else.

And, over time, learning to betray yourself a little less.

Subscribe now

When does a country begin to produce ideas of its own? Not when it publishes more papers or builds more research institutes, but when it begins to name things the world has not yet learned how to describe.

Social science often advances this way. Reality comes first. Then someone discovers that the existing vocabulary no longer fits and coins a new term. At first, the term may do little more than identify a peculiar phenomenon. Over time, it acquires boundaries, mechanisms, and explanatory force. Eventually, if it is good enough, it leaves the place where it was born and begins to illuminate problems elsewhere. That is the point at which a local expression becomes a social-scientific concept.

Several recent developments in Chinese academia are worth viewing in this light. On September 24, 2026, Fudan University hosted an interim workshop for a major Ministry of Education project on “Building an Autonomous Knowledge System for Chinese Economics.” Seven research teams from Fudan, Peking University, Tsinghua University, Shanghai University of Finance and Economics, and other institutions took part. The discussion went beyond particular problems in the Chinese economy to questions of theory and intellectual architecture. Fudan has also made the construction of an autonomous knowledge system in economics a priority for the Fifteenth Five-Year Plan period.

Three days earlier, the Institute of World Religions at the Chinese Academy of Social Sciences and China Social Sciences Press released a volume with a revealing title: Signature Concepts in the Autonomous Knowledge System of Chinese Religious Studies. The ambition is no longer merely to develop “China’s own” religious studies. The deeper question is what distinctive concepts such a body of knowledge should actually be built around.

Nor is this confined to religious studies. The Ministry of Education has called for discipline-by-discipline efforts to build autonomous knowledge systems around signature concepts and original theories. Similar projects are under way in law, international law, education, ideological and political education, and other fields. Taken together, these developments suggest something larger than the emergence of any single new term: Chinese academia is consciously moving from the production of knowledge toward the production of concepts.

One might call it a project of concept-making.

Whether it will succeed is impossible to know. But the attempt itself deserves attention.

I. The Power to Name

We usually think of academic competition as a competition over facts and theories. Whoever has better evidence or a stronger explanation gains the advantage. But there is a third form of competition, often less visible and ultimately more consequential: competition over concepts.

A concept does more than name a phenomenon. A good concept changes the way people see the world. Once Weber gave bureaucracy a systematic form, institutions that had seemed quite different—ministries, armies, firms, administrative hierarchies—could be understood through a common organizational logic. Once “social capital” entered the social sciences, a meal, an alumni network, or a web of acquaintances could be reinterpreted as a resource. Once “soft power” became part of international relations, films, universities, language, and cultural appeal entered the vocabulary of power.

That is what concepts do at their best: they compress reality. They gather scattered facts into a structure that was previously difficult to see.

Whoever creates a concept that others find hard to do without gains, in some measure, the power to shape how a problem is perceived. This is why the difficult part of building an autonomous knowledge system is not citing fewer foreign scholars, or simply giving Chinese names to theories developed elsewhere. Renaming civil society does not produce new knowledge.

The real questions are harder. Are there features of Chinese experience that existing social science has not explained well? Can scholars distill from those experiences a concept with clear boundaries and causal force? Can that concept become part of a larger theory? And, most important, can it eventually leave China and continue to work elsewhere?

That is concept-making.

China has never lacked experience. What it has often lacked is the conversion of experience into concepts.

Over the past four decades, China has generated an extraordinary range of institutional and social experiments. Few countries have undergone, in so short a period, industrialization, urbanization, mass migration, digitalization, infrastructure expansion, intergovernmental competition, real-estate growth, platform capitalism, and social restructuring on such a scale. For social scientists, China has been something close to an open-air laboratory.

But new realities do not automatically produce new concepts. For much of the modern history of China studies, the pattern has been clear: China supplies the empirical material; Western social science supplies the conceptual vocabulary. Something happens in China, and scholars ask whether it is state capitalism, a developmental state, authoritarian resilience, neoliberalism, a welfare state, or civil society.

There is nothing wrong with this. Social science depends on comparison and borrowed theory. But over time the pattern creates an imbalance. China repeatedly enters the global knowledge system as a source of facts, while the terms used to explain those facts are imported from elsewhere.

The result is a peculiar form of intellectual trade: China exports experience and imports explanation.

If knowledge has its own international division of labor, China has long resembled a supplier of raw empirical material. The ore is mined in China; the steel is forged elsewhere. What makes the current language of “autonomous knowledge systems” and “signature concepts” interesting is that it seeks, at least in part, to change that division of labor.

That effort also requires classification. Otherwise, very different things will be thrown together under the heading of “Chinese concepts.”

Some concepts are academic concepts developed by Chinese scholars. They may arise from Chinese experience or address universal problems; what matters is their intellectual genealogy and originality. Others are policy concepts, born in governance before entering academic debate. A third group consists of classical Chinese concepts reworked for modern social science—tianxia, datong, li, guanxi. A fourth consists of concepts coined by foreign scholars to explain China. “Fragmented authoritarianism” belongs here: indispensable to China studies, but not for that reason a Chinese concept. A fifth category is increasingly important: concepts produced transnationally, by scholars and research communities that no longer fit neatly inside national boxes.

A serious “Chinese Concept Export Index” would therefore have to track more than nationality. It would need to record the origin of the empirical problem, the first formulation of the concept, where the theory developed, the language and venue of publication, the path of citation across countries, and whether the concept has been used beyond Chinese cases.

That is a more meaningful intellectual genealogy.

II. How Concepts Travel

Concepts cannot simply be manufactured by committee. Institutions can fund projects, organize terminology, publish handbooks, and call for “signature concepts.” Such work may be useful. But at most it builds the foundry. It does not guarantee good steel.

Concepts with real intellectual life tend to pass through a longer sequence. I would distinguish six stages.

1. Experience. There must first be a reality that existing theories do not adequately explain. Without new experience, forced concept-making usually produces little more than new terminology. China has no shortage of material at this stage: township and village enterprises, competition among local governments, land finance, megacity urbanization, digital payments, platform governance, poverty reduction, high-speed rail. These practices resemble earlier forms of modernization in some respects and depart from them in others. This is where the raw material of concepts comes from.

2. Naming. Someone recognizes a shared structure across scattered phenomena and gives it a name. This is the first act of intellectual compression. Ten cities showing similar patterns of local-government behavior remain ten case studies until someone identifies the common mechanism and names it. But naming is easy. Academia produces names in abundance. The difficult part comes next.

3. Theorization. A concept must explain why something happens, under what conditions, through what mechanism, with what consequences, and where its limits lie. At this point, a term begins to move from description toward theory. If a supposed “Chinese concept” tells us only that “China has this phenomenon,” it remains closer to a cultural label. A theory must permit counterfactual questions. If condition A were absent, would B still occur? If the same mechanism appeared in Vietnam, India, or Brazil, would we expect similar outcomes? Once such questions become possible, the concept has entered social science proper.

4. Operationalization. A concept that remains at the level of grand narrative has little chance of entering comparative research. What exactly counts as “strong state capacity”? What is “high-quality development”? What is “effective governance”? What is “strategic statization”? Without observable indicators, comparison becomes difficult.

Sooner or later, any concept that hopes to travel must answer a plain question: How do you know it is there? Ideally, it should also tell us how much of it there is, whether one country has more than another, and how it changes over time. Once a concept can be operationalized and measured, research becomes cumulative. Other scholars can test it, challenge it, refine it. In fact, being refuted is often a sign of conceptual maturity. It means others have begun to take the concept seriously enough to use it.

5. Transfer. This is the decisive threshold. Scholars begin applying the concept outside China. If a concept derived from Chinese experience can explain only China, it may still be a valuable contribution to China studies. But it has not yet become a general social-scientific concept.

Real export begins when a scholar of Indian cities finds it useful in India, when a researcher of Brazilian local government sees the same mechanism at work in Brazil, or when someone studying American infrastructure discovers a related pattern in the United States.

At that point, the direction of intellectual movement reverses:

foreign concepts → explain China

becomes

concepts from Chinese experience → explain the world

That is the moment of genuine conceptual export.

6. De-localization. I once described this stage as the disappearance of origin. That was too strong. A successful concept does not need to lose its birthplace. It does need to escape the limits of that birthplace.

Guanxi still carries an unmistakable association with China. Tianxia, if it becomes more important in international political theory, will not cease to have Chinese intellectual origins. The highest stage is therefore not one in which people forget where a concept came from. It is one in which they no longer assume that the concept can explain only the place from which it came.

A concept has been de-localized when its origin remains visible but no longer confines its use.

A good concept may have a birthplace. It should not need a visa.

III. The Passport Test

By this six-stage model, the most visible change in China today is taking place at the second stage: the movement from the production of experience toward the deliberate naming of concepts.

That is already a meaningful shift. Across several disciplines, Chinese scholars have begun to discuss what makes a concept distinctive, where its boundaries lie, and how it relates to other concepts. The question is slowly changing from “How can China build theories of its own?” to “What concepts should those theories actually be built from?”

That is progress.

But the harder part begins now. The next task is not merely to produce more concepts. It is to eliminate them.

Most new terms will die. That is normal. Intellectual life has its own form of natural selection. An era may generate thousands of academic terms; only a few will survive long enough to change how people think. A serious knowledge system therefore needs not only a mechanism of concept production, but also a mechanism of concept selection.

This suggests a simple test: the Concept Passport Test.

When evaluating a supposedly original Chinese concept, do not begin by asking whether it sounds sufficiently Chinese. Ask six questions instead. Does it arise from a real and important phenomenon? Does it have clear boundaries? Does it identify an explanatory mechanism rather than merely express a value judgment? Can it be observed, measured, or at least empirically tested? Are scholars outside China using it? And once it leaves China, can it explain new facts elsewhere?

The first three questions tell us whether it is really a concept. The last three tell us whether it is capable of export.

Autonomy and international relevance are not opposites. If anything, a genuinely original concept should survive use across cultures. If it works only inside the political, cultural, or linguistic environment in which it was coined, that may indicate that it has not yet reached a sufficiently high level of abstraction.

There is another danger: mistaking a keyword for a concept.

Policy systems need keywords. Academic systems need concepts. The two sometimes overlap, but they are not the same thing. A policy term can be politically powerful without having stable analytical boundaries. An academic concept must allow researchers to ask when it applies, when it does not, and whether the evidence contradicts it.

The distinction is simple: Policy language seeks agreement. Academic concepts must permit disagreement.

A concept that can only confirm itself is a poor social-scientific concept.

For Chinese social science, then, the difficult task may not be to create more “signature concepts,” but to build methods through which concepts can be tested, compared, falsified, and revised. The international academy will not adopt a concept because it has “Chinese characteristics.” Scholars use concepts because they solve problems.

The recently proposed term “strategic statization” offers a useful example. Yi Feng and Fulong Wu use it to describe recent changes in Chinese urban governance. Earlier urban development in China, in their account, was strongly oriented toward local growth: municipal governments competed for investment, developed land, expanded real estate, and pursued inter-city competition. More recently, that logic has shifted. National strategic priorities have become more explicit, while state-owned enterprises, local governments, financial institutions, and social organizations have been drawn more closely into national strategic objectives. The authors describe this shift as strategic statization.

The phrase is useful, but it also exposes a classification problem. The concept is about China, and its authors are deeply engaged in China research. But that does not automatically make it an “original Chinese concept.” It emerged within international debates in urban studies and state theory, in conversation with terms such as urban entrepreneurialism, state entrepreneurialism, state capitalism, and statecraft.

This is precisely why concept export needs genealogy rather than slogans.

Over the next decade, if China continues to invest in autonomous knowledge systems, the most important question will not be how many “signature concepts” are announced. It will be whether any of them begin to travel.

Are foreign scholars using them without being prompted to do so?

Do they appear in English-language social-science textbooks?

Are doctoral students using them to analyze societies outside China?

Do scholars studying their own countries discover that existing Western concepts are inadequate and turn instead to a concept generated from Chinese experience?

And do some of these concepts, after criticism and revision, become less like political language and more like theoretical tools that can actually fail?

If that begins to happen, Chinese social science will have entered the stage of genuine concept export.

The change would amount to an inversion of a much older intellectual pattern. For more than a century, one of the central tasks of Chinese intellectual life has been learning how others described the modern world. Terms for state, society, economy, democracy, class, capital, nation, science, and countless other modern categories entered Chinese through complex processes of translation, adaptation, and reinvention. Many words that now seem entirely native were once products of intellectual import.

The interesting future is not one in which China stops learning from the world. It is one in which the traffic begins to move both ways.

A mature knowledge system does not merely explain itself. It contributes concepts to a shared language of explanation.

That is the real test. Not whether a concept is frequently cited by Chinese scholars, or repeatedly declared important at Chinese conferences, but whether someone elsewhere finds that it works.

One day, perhaps, a scholar studying Brazilian cities, Indian local government, African industrialization, or American digital platforms will encounter a problem that the familiar vocabulary does not quite explain. And then a concept developed from Chinese experience may turn out to be unexpectedly useful.At that moment, the concept will have truly left China.

Concepts have birthplaces. Good concepts do not need passports.

Subscribe now

Americans study China. Japanese scholars study China. French scholars study China. Chinese scholars study their own country. They face the same country, but do not necessarily ask the same questions. Some ask why China has not undergone the transformation they expected; others ask how it has sustained its development. Some see a challenge to the international order; others find an alternative experience of modernization. Their disagreements often precede their conclusions: they have already decided what counts as a “China question.”

Each scholarly tradition has accumulated knowledge, and each has limits to its vision. How can knowledge travel across languages, disciplines, and countries? How can scattered materials become a shared research foundation? How can competing explanations confront the same evidence? Chinese studies must address these questions directly. China should draw on its capacity to “concentrate resources on major undertakings” and make the integration of Chinese studies worldwide a long-term scholarly project—one that connects history with the present, and China with the world.

The proposed “new national learning,” or xin guoxue, takes China as its subject, joins the study of its traditions to contemporary inquiry, and brings scholarship from around the world into a common conversation. It connects civilizational scholarship with research on present-day China and places Chinese experience within international comparisons. Its organizational purpose must be equally clear. Integration means pooling resources, exchanging methods, collaborating on research, and testing one another’s claims. China can provide coordination, sustained funding, and access to research; overseas scholars and institutions can participate voluntarily. They belong to different institutions and intellectual traditions. An open network will hold together only if it helps them ask better questions, find more reliable evidence, and accomplish work they could not undertake alone.

Chinese Studies in Japan: Questions, Methods, and History (日本の中国研究——課題・方法・歴史), edited by Kawashima Shin (川岛真) and Sonoda Shigeto (园田茂人), offers a starting point. Published by the University of Tokyo Press in August 2026, it traces Japanese research on China across law, history, sociology, political science, and economics. It gives knowledge about China a history of its own: how methods change, disciplines divide and reconnect, and the concerns of an era enter scholarly inquiry.

The book’s first three parts examine “The Legacies and Constraints of Kangaku and Shinagaku,” “Starting from Japan’s Engagement with China,” and “The Founding of New China and the Development of a ‘New Field of Scholarship.’” Its sociology chapter surveys 140 years of research on China; another examines prewar rural surveys and the knowledge inherited and developed after the war. This structure reminds us that scholarly history and relations between states are intertwined. A research tradition does not start afresh when a political era ends.

Further questions follow. Which materials do researchers select? Which concepts do they use? Which countries and periods supply their comparisons? In traditional Chinese learning, China could be a source of classical texts, writing, and thought. Under modern interstate relations, it also became a neighbor to investigate, interpret, and respond to. Cultural admiration, political rivalry, cooperation, and conflict may leave different traces in scholarship. Finding those traces requires examining particular surveys, institutions, and works. A country’s scholarship cannot be reduced to a single position.

The object of study is real. The framework through which it is observed has a history. A young scholar entering Chinese studies first encounters textbooks, canonical works, advisers, journals, and established debates. Before entering a Chinese village, business, or archive, the scholar already knows which questions command attention and which concepts organize the evidence. This training builds expertise, but also habits of inquiry. A field that keeps refining its answers to old questions while leaving their premises unexamined may explain the past with growing precision while becoming slower to recognize the present.

Academic specialization imposes another limit. Legal scholars study laws and institutions; sociologists study families, social strata, and organizations; economists study growth, firms, and markets; historians study how change came about. These aspects of China are connected, yet research can proceed in isolation. Understanding industrial development requires attention to supply chains, technological capabilities, infrastructure, labor relations, and local institutions. Artificial intelligence connects education, employment, platforms, and knowledge production. Many decisive mechanisms cross disciplinary boundaries. A single line of inquiry rarely captures how they interact.

“China” itself needs to be unpacked. We say “China wants,” “China has decided,” or “China is changing course.” Such shorthand can end an explanation before it begins. A diplomatic position, a local government’s pursuit of investment, a company’s overseas operations, and a family’s choices belong to different levels of activity. They sometimes align and sometimes pull against one another. Researchers should specify whom they mean by “China.” An integrated approach must then explain the connections among these levels.

Andrea Ghiselli and Todd H. Hall approach this problem from another direction in Explaining China’s International Behaviour. They identify four ideal types of explanation: universalist, exceptionalist, political-institutionalist, and particularist. The disagreements turn on two judgments: how far China acts as a unitary actor, and how far it is a distinctive one. Their approach adjusts the analysis to differences across periods and policy domains.

This keeps “What is China?” open as a research question. The degree of unified action must be established through evidence about decisions and behavior; claims of Chinese distinctiveness require comparison. A country can move in different directions at once: some industries contract while others expand. Compressing those developments into “rise” or “decline” can obscure their causes. A persuasive account should explain where change occurs, how its effects travel, and who bears the costs or gains the benefits.

These problems give the integration of Chinese studies a practical purpose: to connect dispersed knowledge and enable competing explanations to be tested on common ground. The “new” also concerns how scholarship is organized. Classical research should inform contemporary questions; contemporary experience should test historical interpretations; findings across disciplines and languages should be available for comparison, debate, and revision.

Why should China take the initiative? It is both a subject of research and a producer of knowledge. Much of the documentation, local evidence, lived experience, and access needed to understand China lies within the country. Scholars elsewhere have accumulated work in other languages, bringing methods, materials, and questions worth absorbing. China is well placed to connect these sources and make its research resources available to a global scholarly community.

Here, “concentrating resources on major undertakings” should mean dependable, long-term investment and cooperation across institutions. Organizing documents, building datasets, translating and publishing scholarship, and training researchers are costly and slow. Their benefits are widely shared, making them difficult for individuals or short-term projects to sustain. The state can support this foundational work, while universities, research institutes, libraries, archives, and publishers contribute their expertise. Success should be measured by the lasting scholarly capacity that investment creates.

The project requires national coordination, with research distributed across institutions. A dedicated, long-term program could entrust coordination to qualified bodies and support cooperation across institutions, disciplines, and countries. Different centers could oversee documentary collections, social surveys, economic data, intellectual history, technology, and industry, linking their work through shared catalogs, research agreements, and regular exchange. Overseas partners would retain their affiliations and scholarly judgment, joining through specific projects.

The coordinating body should organize work that scattered institutions can accomplish together: secure resources, set priorities for foundational investment, reduce duplication, and support collaboration. Research questions and conclusions belong to scholarly inquiry and peer review. Selection criteria should be public, funding transparent, and reviewers free of conflicts of interest. Researchers must be able to challenge findings and revise their conclusions. International cooperation will depend on the credibility of these arrangements.

An early task should be to map Chinese studies worldwide: its major books, articles, institutions, journals, collections, and projects across languages. A continually updated catalog would show researchers where existing knowledge can be found. It should encompass scholarship from different regions, widening the experience available to collaborators. Catalogs, bibliographies, and metadata could be shared first; access to full texts, archives, and survey materials would follow the relevant copyright, access, and privacy requirements.

This map could also help Chinese studies examine itself. How does the same Chinese phenomenon become a different question in different countries? Which subjects attract sustained attention, and which groups and regions are overlooked? How do funding, language training, access, and publishing shape inquiry? Answers require evidence from projects, publications, citations, and other sources. Studying the field itself would expose gaps and discourage the assumption that national differences reflect uniform national positions.

A translation program should accompany this work, operating across languages and in both directions. Chinese scholarship needs to enter international debates; overseas scholarship needs systematic translation into Chinese. Selection should extend beyond works that confirm familiar views. Disagreement supported by evidence, sound methods, and explanatory power can reveal a field’s blind spots. Important concepts need their contexts and translations explained, so scholars know what they are actually debating.

Artificial intelligence can assist bibliographic searches, preliminary translation, and the organization of materials, lowering barriers to multilingual cooperation. Publication and research still require expert checking, especially of historical context, conceptual differences, names, and quotations. Scholarly review, copyright agreements, and continuing updates should be built into the program so that technical convenience yields reliable knowledge.

A deeper task is to give researchers shared access to evidence. Continuing joint fieldwork sites could allow scholars to study industry, families, education, urban–rural relations, and social change in the same places. Reusable datasets and historical collections could be assembled around specific subjects. Sources, collection methods, scope, and missing information should be documented so later researchers can replicate, compare, and revise the work. Cooperation can advance only as far as its evidence permits.

International research should build on this foundation. Industrial transformation, demographic change, local governance, educational mobility, and technological development all lend themselves to long-term projects. Teams could advance different explanations and examine the same evidence together. Where feasible, research plans, data processing, and counterexamples should be open to discussion, making disagreement a reason to test claims more rigorously. Cooperation gives each participant access to experience otherwise difficult to obtain.

Training must proceed alongside these investments. Scholars of China need opportunities to combine language study, disciplinary methods, and field research. Chinese scholars need greater exposure to overseas traditions and the questions they generate. Extended visits, shared courses, collaboration among younger scholars, and continuing local research are more likely than one-off conferences to create working relationships. Training people who can read different kinds of evidence and move between intellectual traditions will demand the greatest patience.

Evaluation should reflect these longer horizons. Organizing archives, building datasets, producing good translations, and training younger scholars may take years to demonstrate their value. If assessment rewards chiefly publication counts and immediate publicity, foundational work will be marginalized. The program should offer professional criteria and sustained support for these contributions, while checking whether materials are reliable, platforms are used, and collaboration produces findings that can be tested.

Global cooperation must also accommodate disagreement. Whether a network supports findings its organizers did not expect, and admits evidence-based criticism, will influence scholars’ willingness to stay. China can strengthen its scholarly appeal by opening materials, improving access, and supporting serious debate. Credibility and explanatory power grow through such everyday practices.

Chinese scholars must subject themselves to the same scrutiny. Language, lived experience, and institutional knowledge offer advantages, but no panoramic view. What passes for common sense in one region, generation, or occupation may need reinterpretation from another social position. China’s American, Japanese, and European studies should ask similar questions: Which concerns arise from the societies being studied? Which come from bilateral relations? Which reflect our own experiences and anxieties? Such reciprocal scrutiny would strengthen global cooperation.

Researchers have positions; facts still constrain their claims. Archives can make a historical interpretation untenable, data can overturn an account of trends, and fieldwork can reveal the distance between concepts and experience. Research must be judged by evidence and reasoning. Do the materials support the conclusion? Does an alternative explanation work better? Can a case be generalized? What changes when counterexamples appear? Integration should make these tests easier.

The major undertaking China should pursue is a scholarly enterprise that accumulates knowledge, sustains open cooperation, and corrects itself. State organizational capacity should build enduring foundations for documentation, data, translation, research, and training, allowing scholars worldwide to work with shared resources. The state supplies support, scholarly communities conduct inquiry, and international partners join voluntarily. Clear responsibilities and dependable institutions must connect them.

A new, integrated Chinese studies should grow from this foundation. It should trace the particular paths of Chinese experience and place them within broader comparisons; address contemporary problems and examine its own questions. China should build the platform and earn participation through its scholarly value. Knowledge about China comes from across the world. China has a responsibility to help those bodies of knowledge meet.

Concentrating resources can bring scattered knowledge together. Allowing evidence to change judgments will keep it alive. Chinese studies will have achieved something lasting when every claim about China must answer to China’s reality—in all its complexity.

Sources

1. Kawashima Shin (川岛真) and Sonoda Shigeto (园田茂人), eds., 日本の中国研究——課題・方法・歴史. University of Tokyo Press, 2026. The account of the book’s structure and chapters draws on the publisher’s description and contents. The proposal for integrating Chinese studies worldwide is the argument of this essay. Publisher’s page.

2. Andrea Ghiselli and Todd H. Hall, Explaining China’s International Behaviour. Cambridge University Press. The discussion of the four explanatory approaches and contextualized analysis draws on the publisher’s abstract. Publisher’s page.

Note: This essay is inspired by the work of American psychologist Stephen Gilligan on Self-Relations, the Generative Self, and Generative Change, and extends those ideas in a particular direction. The argument that “the fewer demands we place on others, the stronger our agency becomes” is the author’s own development, not a direct quotation from Gilligan. Gilligan received his Ph.D. in psychology from Stanford University in 1983 and has since worked extensively in psychotherapy and generative change. Across his writings, he has explored such ideas as generative autonomy, the relational self, and the generative self.

There is a simple way to gauge the strength of a person’s agency: ask how much of their inner life they have made someone else responsible for. “Why don’t you understand me?” “Why didn’t you answer my message?” “Why can’t you treat me better?” “I’ve told you so many times—why won’t you change?” These are familiar questions in intimate relationships. They are not inherently unreasonable. People have needs, and they are entitled to expect that their boundaries will be respected. The problem begins when a person’s sense of security, worth, happiness, and even daily emotional equilibrium come to depend on whether someone else behaves as expected. What looks like assertiveness can, under those conditions, become a gradual surrender of power. The paradox is easy to miss: the more determined we become to control another person, the more control we may lose over ourselves. Other people are, after all, among the least controllable facts of our lives.

Making Demands Can Mean Handing Over the Self

A demand appears to be an act of agency. “I am telling you what I need” sounds like self-expression, perhaps even self-respect. Yet two very different psychological structures can lie behind the same sentence. One says: “This matters to me, and I hope you can meet this need. If you cannot, I will have to decide what that means for the relationship.” The other says: “You must do this, or I cannot be all right.” The difference is small in language and enormous in consequence.

The first person still possesses a self apart from the other person’s response. They can ask, negotiate, feel disappointed, reconsider the relationship, or leave. The second has quietly placed the switchboard of emotional life in someone else’s hands. I am important only if you pay attention to me. I am safe only if you respond. I am worthy only if you approve of me. I am lovable only if you love me in the form I require. At that point, the more urgently someone else is asked to confirm our importance, the less able we become to confirm it for ourselves. The more we demand that another person provide security, the more our security comes under that person’s control. The more we insist that they change, the more our own emotional state is governed by whether they do.

What appears to be pressure exerted on another person can therefore become a form of dependence on them. The more important question behind “Why can’t you...?” may be: Why must this particular thing come from you before I can be all right? Agency begins when that question turns inward.

We often confuse strength with influence. To change other people, persuade them, or induce them to act as we wish seems like power. Psychological power often works in the opposite direction. Agency begins with the ability to distinguish what belongs to me from what belongs to someone else. Its deepest form is not control over the world but the recovery of one’s own side of the causal chain.

Another person’s temperament belongs to them. So do their choices, their timing, their willingness to respond, their capacity for affection, and the degree to which they understand us. We can speak, negotiate, influence, and, when important needs are persistently unmet, leave. But if our inner order rests on the premise that another person must change, then something beyond our control has been placed at the center of our well-being.

One strand of Gilligan’s work on Generative Change concerns moving people out of fixed problem states and into conditions from which new possibilities can emerge: clarifying intention, encountering obstacles, finding resources, acting, and integrating those actions into life. Extended into the ordinary language of relationships, this suggests a shift from “Why won’t this person change?” to “How do I want to live if this person does not change in the way I want?” The difference may seem slight, but psychologically it is decisive. The first question locates agency elsewhere. The second brings it back.

That is the beginning of self-possession.

Marriage offers an obvious example. One of the deepest forms of exhaustion in long relationships arises when a person gradually makes the improvement of their own life dependent on the improvement of their partner. Early in a relationship, people often see strengths. Ten or twenty years later, they may see mostly defects: the way the other person eats, the disorder they leave behind, their poor communication, insufficient ambition, inattentiveness to the children, lack of romance, worsening temper. Eventually a larger conclusion begins to form: perhaps my life has not gone well because the person beside me failed to become the person I needed them to become.

Some marriages, of course, involve betrayal, violence, coercion, or profound irresponsibility. Those realities cannot be explained away by invoking agency. But many ordinary marriages are worn down by something less dramatic and more cumulative: expectation. I want you to understand me better. I want you to become more mature. I want you to be more ambitious. I want you to know what makes me happy. I want you to remember what matters to me. I want you to stop being this way. Each expectation may be reasonable in isolation. Over twenty years, however, a partner can become the object of a continuous renovation project.

Daily life then becomes a form of inspection. Did they behave correctly? Did they say the right thing? Has their attitude improved? Are they contributing enough? Attention flows increasingly toward monitoring the partner and away from building one’s own life. Agency is lost by degrees. What began as “my partner has a flaw” becomes “unless my partner changes, my life cannot improve.” The deeper constraint is no longer simply the partner. It is the premise one has built around them: my happiness must pass through you.

Strong Agency and Weak Agency

An important qualification is necessary. Asking less of others does not mean turning intimacy into a declaration of self-sufficiency. “I need no one” is not necessarily agency; it can be another form of defense. Human beings are relational creatures. Gilligan’s work has concerned not only the generative self but also the relational self. His account of psychological health is not one of sealed-off individualism. It leaves room for connection, dependence, mutual influence, and the ways in which people are formed through relationships.

The question, then, is not whether we should need other people. Of course we do. We need love, friendship, understanding, loyalty, cooperation, commitment, and care. A person who insists that they need no one may be less autonomous than defended against disappointment. Agency alters something subtler: I can have needs without converting every need into an obligation that another person must fulfill. I can hope for a response without making my sense of existence depend on receiving it.

I can hope to be loved without allowing another person’s coldness to become proof that I am unlovable. I can want to be understood while accepting that no person is ever fully understood by another. I can require fidelity and respect, and if they are persistently absent, I can leave rather than spend years trying to manufacture those qualities in someone who does not possess them. Strong agency is neither dependence nor isolation. It is the capacity to remain connected without handing over the self.

Weak agency often reveals itself in the effort to revise other people. Strong agency begins to build the self. The difference is audible in everyday language. Weak agency tends to live in conditional sentences: if he treated me better, I would be happy; if my boss appreciated me, I would feel confident; if my child behaved differently, I would not be so anxious; if my parents understood me, I would no longer suffer; if other people approved of me, I would finally dare to act. The structure is always the same: my condition must wait for someone else’s behavior to change.

Life then becomes a waiting room. One waits for a husband to change, a wife to understand, a child to mature, a parent to apologize, a boss to recognize, a friend to reply. People can wait this way for decades. Strong agency reverses the direction of causation. If my partner never becomes emotionally expressive, how will I build an emotional life that does not depend entirely on that limitation? If my boss never values me, will I stay, work differently, or leave? My child has a life of their own; how will I manage my anxiety about it? My parents may never acknowledge the harm they caused; can I still make a different relationship to my own past?

This is not resignation. It is an active refusal to make the most important changes in one’s life depend on the least controllable conditions.

Real stability becomes visible when other people stop following the script. Anyone can appear stable when life behaves as expected. Agency is tested when a message goes unanswered, a partner fails to offer the hoped-for comfort, an effort goes unrecognized, a friend misunderstands us, a child makes a decision we oppose, a parent remains stubborn, or work fails to produce the expected reward. Only then does it become clear whether there is a center that does not disappear when the external world disappoints us.

Gilligan’s later work on Generative Change emphasizes a generative state capable of holding difficulty, resources, and possibility at once. The aim is not simply to remove obstacles, but to move from rigidity and disconnection into a condition from which new responses become possible. In the language of ordinary life, the principle is simple: the world may disappoint me, but it does not get to decide who I am.

If someone does not respond, I can feel hurt. If someone leaves, I can grieve. If recognition is withheld, I can be affected. Agency does not mean immunity from emotion. The question is what happens next: can I return to myself and reorganize my life? That matters more than what is often called emotional stability. Some people are stable because their expectations are consistently met. Others possess a more durable stability: even when reality refuses to cooperate, they remain capable of living.

Taking the Demand Back

The power in “asking less of others” does not lie in lowering every standard, nor in teaching people to tolerate mistreatment. It lies in redirecting the demand. Instead of asking, “Why can’t you make me happy?” ask, “How do I build the capacity for happiness?” Instead of “Why don’t you understand me?” ask, “Do I understand myself?” Instead of “Why does no one see my value?” ask, “Am I doing what I myself consider valuable?” Instead of “Why won’t other people change?” ask, “If they do not, what choices remain available to me?”

Power returns through these small changes in direction. They can also be understood as an extension of Gilligan’s idea of the generative self. Generation does not begin only after the world provides ideal conditions. It happens amid limitation, uncertainty, relationship, and disappointment. A person mobilizes resources, creates a response, acts, and is changed by acting. Agency therefore does not mean, “I control everything.” It means: however much lies beyond my control, I still possess a next move.

Perhaps the deeper dividing line in adult life is the movement from “What should you do?” to “What can I do?” When we are young, hope is often placed elsewhere. We hope our parents will understand us, our partners will change, our friends will remain loyal, our children will succeed, society will be fair, and our efforts will eventually be recognized. There is nothing childish about these hopes. They belong to ordinary human life. What changes with experience is the recognition that many of them will not be fulfilled. Some people never apologize. Some never learn to understand us. Some disappoint our trust. Some relationships cannot be repaired. Some efforts are never properly rewarded. If our happiness depends on these people or conditions eventually changing, a lifetime can be spent waiting.

Agency emerges at that point. It does not require hardness, and it does not require the fantasy that one no longer needs anyone. It requires a smaller but more consequential shift: from “How should you treat me?” to “Given who you are, how do I want to live?” The first question remains organized around the other person. The second begins to recover the self.

That may be why asking less of others can make a person stronger. It is not because other people cease to matter, or because relationships become dispensable. It is because another person is no longer asked to carry the part of life that only we can carry ourselves. Love still matters. So do kindness, loyalty, recognition, and care. They remain among the great goods of human life. They simply cease to be the entire foundation of the self.

Mature agency may finally come down to this: I am willing to love you, and I am willing to need you, but I will not give you responsibility for becoming myself. At that point, the struggle with the world begins to loosen—not because the world has finally agreed to our terms, but because we have stopped making our existence contingent on its agreement. Other people have their lives, and we have ours. We cannot determine how others will treat us. We can still decide what we will do with what follows.

Subscribe now

Sometimes the deepest changes in a country do not appear first in the text of its Constitution. They appear in what happens when an ordinary encounter with the state sends two people into different legal systems.

Consider a routine traffic stop or minor accident. For a U.S. citizen, the matter will ordinarily remain within the familiar world of traffic law, insurance, fines, and, if the conduct is serious enough, the criminal courts. For a noncitizen without lawful status, with an expired visa, or with another immigration problem, the same encounter may have an additional consequence: it can bring that person’s immigration status to the attention of federal authorities and trigger a separate civil immigration process. The traffic violation itself is not necessarily the basis for ICE detention. But the encounter can become the moment when an existing immigration violation is discovered, leading to ICE custody, removal proceedings, and an uncertain period of lost liberty.

The two people are not formally being punished differently for the same traffic offense. One of them has entered a second legal system that does not apply to the citizen. The deeper question is how far immigration status should alter the protections surrounding physical liberty.

That is where the constitutional issue begins.

The Fifth Amendment does not say that no “citizen” shall be deprived of life, liberty, or property without due process of law. It says no “person.” The Supreme Court has repeatedly recognized that noncitizens are protected by the Due Process Clause and that government detention represents one of the most serious forms of state power over the individual. Yet the distance between the constitutional “person” and the noncitizen encountered in the machinery of immigration enforcement appears to be widening.

Recently, an immigration lawyer in the United States told her colleagues that, for the time being, she would stop taking cases involving people arrested by ICE who needed habeas corpus petitions or immigration bond proceedings. It was not because there was too little work. Quite the opposite. The cases were multiplying.

What she could no longer bear was something else. She understood the law better than before. She knew the procedures more thoroughly. She prepared her filings more carefully. Yet she found herself less and less able to answer the simplest questions her clients asked: When will I get out? I don’t know. When will I see a judge? I don’t know. If I no longer want to wait and am willing to leave the United States, when can I go? Still no answer.

Her helplessness points to a problem larger than immigration policy itself: when a person’s immigration status can place him in a prolonged, administrative, and uncertain form of detention that an American citizen would rarely face, what remains of the principle that everyone is equal before the law?

The United States plainly has the authority to enforce its immigration laws. It may decide who enters, who remains, and who must leave. Entering without authorization, overstaying a visa, or otherwise violating immigration law does not become lawful merely because someone has lived in the country for years. But enforcement raises another question: how far may the state go in subjecting a person to a form of lost liberty fundamentally different from that faced by an ordinary American citizen simply because of immigration status?

An American citizen suspected of a crime generally enters the criminal justice system. The government must bring charges. There is a judge. There are rules governing bail. There is a trial process. Immigration detention is formally civil and administrative, yet it can still place a human being behind locked doors. Calling detention “civil” does not make the loss of liberty less real.

This is precisely where some of the most important legal disputes are now unfolding. Since 2025, the Trump administration has adopted a broader interpretation of mandatory detention for certain noncitizens arrested inside the United States, including circumstances in which ordinary bond hearings may be unavailable. Several federal appellate courts have challenged that interpretation, and on October 1, 2026, the Supreme Court agreed to hear the dispute. At its center is a basic question: can certain people who have lived in the United States for years, some without criminal convictions, be taken into ICE custody and held without the kind of bond hearing ordinarily associated with a judicial determination of whether continued detention is necessary?

The contrast is stark. Even when an American citizen is accused of a crime, the government generally must move through an elaborate criminal process before depriving that person of liberty for an extended period. A person without lawful immigration status, by contrast, may have been convicted of no crime at all and yet find himself in administrative detention for weeks or longer because of an immigration violation. Whether a person holds an American passport increasingly shapes the baseline of liberty from which that person confronts state power.

Equality has never meant that citizens and noncitizens possess identical rights under immigration law. A citizen cannot be deported; a foreign national can. But equality should at least mean that when the state physically confines a human being, the absence of an American passport does not justify a diminished regard for liberty.

Liberty is not a benefit of citizenship. It is, before anything else, a human right.

What is troubling in the United States today is that this boundary is becoming less clear. Immigration status once principally determined the answer to one question: Do you have the right to remain in this country? Increasingly, it also seems to determine another: How freely may the government confine you?

Those are not the same question.

A person may ultimately have no legal right to remain in the United States. But that does not mean that, while awaiting that determination, he should lose the procedural protections ordinarily associated with government detention. A person may be deportable and still retain a claim to due process, judicial review where the law provides it, and treatment consistent with human dignity.

This is where the immigration lawyer’s story becomes more than a personal account. She entered the profession with a simple belief: legal work could support her family and, at the same time, help people who needed someone to stand between them and the machinery of the state. Like every serious lawyer, she knew that the law could not guarantee an outcome. But she could explain the process, prepare the case, invoke the relevant rights, and tell a client what legal avenues remained open.

What now troubles her is that she can do all of those things and still be unable to say when the process will end. She can explain what was filed, whom she contacted, what the next legal step may be. What she increasingly cannot explain is how long a human being must remain suspended between custody and uncertainty.

That uncertainty is not abstract. Outside the detention center, ordinary life continues. Rent is still due. Cars remain parked on the street. Children still need to be picked up from school. Food in the refrigerator still spoils. One detained person, worrying about his own future, nevertheless remembered that the milk in his refrigerator was about to expire and asked that it be given to someone else.

It is a small detail, but perhaps the most revealing one.

The state sees an immigration status. A detention center sees a bed. A database sees a case number. A lawyer sees a file. But the detained person remembers the milk in his refrigerator.

Bureaucratic systems become most dangerous not necessarily when someone within them intends cruelty, but when a whole person is gradually reduced to the categories the system knows how to process: status, detention location, removal order, hearing date, case number. Everything else becomes somebody else’s problem.

The family.The job. The landlord.The child. The unfinished life outside the walls.

A legal system is tested not only by how it treats the people everyone admires, but by how it treats those with the least political power: people without votes, without citizenship, without money, sometimes without fluent English, and sometimes in violation of immigration law. Due process is not a prize awarded to the deserving. It is a method of limiting state power.

That principle matters most when the person standing before the state is unpopular.

It would be too simple to say that the United States has ceased to be a country of laws. Courts are still reviewing executive action. Lawyers are still filing habeas petitions. Federal judges are still imposing limits where they conclude the government has exceeded the law. The constitutional system has not disappeared.

The more troubling question is whether the practical burden of invoking the law is becoming so heavy that the formal existence of rights offers less protection than it once did. A right that can be exercised only after weeks or months of uncertainty is not meaningless. But it is not the same thing as a right that can be promptly and effectively used.

That tension now runs through immigration detention.

The government may ultimately be correct that a person has no lawful right to remain in the United States. But removal and detention are not interchangeable concepts. The power to decide that someone must leave does not automatically answer how long that person may be confined, under what procedures, and with what opportunity to challenge continued detention.

That distinction is easy to lose when immigration enforcement becomes politically charged.

It is also the distinction a constitutional system must insist on preserving.

The most basic question, then, is not whether America may enforce its borders. Of course it may. Nor is it whether citizenship matters. Of course it does.

The harder question is this:

How much less liberty should a person have simply because he is not a citizen?

A society can answer that question in many ways. But once the answer becomes “as much less as the government finds convenient,” something fundamental has changed.

The Fifth Amendment says “person.”

The word is ordinary, almost easy to overlook. But perhaps that is exactly why it matters.

Before someone is undocumented, removable, inadmissible, detained, or deported, he is a person.

A legal system worthy of the name should never forget the order of those words.

Du Hongwei

My first thought after reading Zhao Xiao’s essay was that, although it appears to be about China and the United States, its deeper subject is humanity itself. Byzantium and Sasanian Persia fought for more than two decades, each convinced that the other posed the greatest danger to its survival. Byzantium eventually emerged with the military advantage, while Persia soon descended into internal turmoil. Yet by then both empires had been badly weakened by years of war, and the Arab conquests that followed transformed the political map of the Near East.

The analogy should not be pushed too far. China and the United States are not seventh-century Byzantium and Persia, and artificial intelligence is certainly not a third empire suddenly riding out of the Arabian Peninsula. But the story carries a warning worth keeping in mind: the most dangerous variable in a system is not always the rival standing directly in front of you.

Zhao calls today’s U.S.–China competition a “civilizational marathon.” I think the metaphor works. A marathon’s greatest danger is not simply the runner beside you who happens to be moving faster. Dehydration, heart rate, the road, the weather—even the mistake of running the first ten kilometers too hard—can end your race before your competitor does. A runner who spends the entire course staring at the person next to him, half hoping to trip him, may discover that neither reaches the finish line. States are not so different.

The problem goes deeper than the possibility that some “third runner” might benefit. Great powers have often lost not because of a single military defeat, but because the struggle to win exhausted the foundations of their long-term strength: fiscal capacity, population, institutional resilience, social trust, technological dynamism. These assets are easy to overlook until they approach a breaking point. At that stage, even military victory can turn out to be an expired check. So the question Zhao’s essay left me with was not which side should yield to the other. It was something more fundamental: does a civilization’s greatest adversary lie outside itself, or within? I increasingly suspect the latter.

History Repeats Incentives as Much as It Repeats Wars

There is a familiar remark, often attributed to Hegel, that the lesson of history is that human beings never learn from history. The wording and attribution are more complicated than the popular version suggests, but the thought has lost none of its force. Why are human beings so bad at learning from the past? I doubt the answer is simply that we are stupid. Quite often, the problem is that everyone is being perfectly “rational” from where he stands.

Athens feared Sparta; Sparta feared Athens. If one side armed while the other did not, the second might leave itself exposed. So each had good reason to strengthen itself. In the end, both spent more and both became less secure. Companies face a similar logic. An AI firm may know that more testing would be prudent, but if six additional months of testing allow a competitor to launch first and capture the market, caution carries an immediate price. A government may recognize the risks of superintelligence and still ask: What happens if we slow down and the other side does not? Everyone sees the cliff. Everyone has a reason to keep pressing the accelerator.

This is the collective-action problem Zhao identifies, and it is where AI risk becomes especially difficult. I would slightly revise something I have often said before: human wisdom often loses not to stupidity, but to incentives. Greed is part of it. So are fear, status competition, and the anxiety that someone else may move first. Then there is the most troublesome logic in economics: you make the choice that is rational for you, I make the choice that is rational for me, and together we arrive at an outcome neither of us wanted. That may be humanity’s largest prisoner’s dilemma.

AI May Become Smarter Than We Are. Humans Still Worry Me More

Geoffrey Hinton’s warnings deserve to be taken seriously. A pioneer of neural networks, a co-recipient of the 2018 Turing Award, and a 2024 Nobel laureate in physics, Hinton has repeatedly warned that AI may eventually surpass human intelligence. He has even proposed one of his more provocative analogies: could a more powerful AI be given something like a “maternal instinct,” so that it cares for humanity as a mother cares for her child?

My temperament differs somewhat from Zhao’s here. I believe the risks Hinton describes deserve serious study, but I do not feel the same degree of apocalyptic anxiety. That has something to do with my faith. If one believes that human life is not entirely in one’s own hands, and that death is not the final defeat in any ultimate sense, then one’s relationship to “existential risk” changes. My own attitude is simple: every morning I wake up, I thank God for another day. How many more there will be, I do not know.

That is not pessimism. Quite the opposite. It makes life more vivid. Once a person really understands that death is inevitable, many things shrink to their proper size. A stock price falls a little—so what? Someone insults you—so what? One country wins something today and loses something tomorrow—history will go on. Even death itself does not have to stand at the door each morning like an impatient debt collector. I have joked that if Death ever does knock, I will probably invite him in first. We should have a drink. I will cook a few good dishes. Once we have eaten and drunk enough, we can set off together. Life has already been noisy enough; there is no reason for the last journey to be undignified.

That is my view of death. But an important distinction follows from it: an individual may not fear death; public institutions do not therefore have the right to ignore risk. I may say, “I am not afraid of AI.” A government cannot say “we are not afraid” on behalf of hundreds of millions of people. A CEO may be willing to gamble; that does not entitle him to place society itself on the table. Faith can reduce fear. It does not cancel responsibility. The two must remain separate.

“Created by Us” Is Not a Safety System

Theologically, I believe human beings are created beings and that God is not something humanity can replace. AI, being a human creation, is further still from anything we could meaningfully call “God.” But in practical life, created things can certainly harm their creators. Humans made guns; guns kill humans. Humans built financial systems; financial systems can collapse into crises. Humans designed bureaucracies; bureaucracies can come to dominate the people who created them. Humans developed powerful biological technologies, and in doing so created new kinds of risk.

So the fact that AI is human-made does not prove that it cannot harm humanity. This is the part of Hinton’s warning I find hardest to dismiss. His forecasts may not come true. Estimates of a 10 or 20 percent chance of extinction are ultimately judgments about an unprecedented situation, not probabilities derived from a stable statistical record. That qualification matters. We do not know when superintelligence will arrive, whether it will arrive at all, or what kinds of autonomous goals a sufficiently advanced system might develop.

The sensible position therefore lies somewhere between two easy certainties. One says: “It is over. AI will destroy us.” The other says: “Relax. How dangerous can something we built ourselves really be?” A more mature response is simpler: We do not know. So leave room for error. A mature civilization is not one whose predictions are always right. It is one that still has an exit when its predictions turn out to be wrong.

The U.S.–China Problem Is a Security Dilemma

What I find most persuasive in Zhao’s essay is not simply his call for U.S.–China cooperation. It is the game-theoretic logic beneath it. Many problems between China and the United States share the same structure. Each side often begins not by asking, “Do I want to hurt the other?” but, “What if I do nothing and the other side moves first?”

That is the security dilemma. One side believes it is making itself safer. The other sees a threat. The second responds. The first takes that response as proof that its original fears were justified. Step by step, the system manufactures evidence for its own assumptions. AI is particularly vulnerable to this spiral. Model capability, computing power, military AI, cyber operations, biotechnology, nuclear warning systems, autonomous weapons—the same logic can emerge in all of them. No one wants to be the first to slow down.

That is why AI governance does not require deep mutual trust. What it requires is a structure in which parties that do not trust one another can still communicate, verify, and impose limits. Risk management has never depended on first proving that the other side is a friend. Precisely because the other side is not a friend, hotlines become necessary. Because trust is weak, verification matters. Because competition is real, some guardrails must be difficult to cross. If two people never disagreed, they would need far fewer rules. States are no different. Rules are designed for situations that can get out of control.

Destroying Your Rival Can Mean Destroying Part of Yourself

I would push Zhao’s marathon metaphor one step further. Strictly speaking, a marathon with no other runners is barely a race. Competition is not inherently destructive. The U.S.–Soviet space race accelerated technological development. Competition among firms drives innovation. Intellectual rivalry forces theories to clarify themselves. A civilization that faces no external challenge can become sluggish and complacent.

A mature civilization therefore has to learn something harder than how to defeat a rival: how to live for a long time with a rival that is not going away. Political life naturally understands the vocabulary of war: victory, defeat, destruction, surrender. Long-term civilizational competition requires another vocabulary—restraint, boundaries, interdependence, rules, correction, continuation.

Destroying a rival may look like the ultimate victory. But if your supply chains depend on him, your markets need him, your technological progress benefits from the pressure of competition, and part of the international order has been built around his presence, then his sudden disappearance may teach you an uncomfortable lesson: you have lost more than an enemy; you have lost part of yourself.

Ordinary life offers versions of the same truth. A person’s greatest rival often helps shape who he becomes. Without Roger Federer, Rafael Nadal would still be Nadal, but not quite the Nadal we came to know. Without Sparta, Athens would still have been Athens, but it would not have developed the same political and military understanding of itself. Civilizations are similar. A rival is an unfriendly mirror. Mirrors can be unpleasant. But when the mirror shatters, we also lose one way of seeing ourselves.

Humanity May Always Have Been Humanity’s Greatest Risk

So when I think about AI, I find myself less interested in the question “Will AI become humanity’s final enemy?” than in another one: What parts of human nature will AI amplify?

Intelligence? Certainly. Creativity? Yes. Medical capability? Almost certainly. Military power? That too. The capacity to deceive? Of course. Greed, fear, ambition, and the hunger for power will not mysteriously disappear when intelligence becomes artificial. They may acquire better tools.

If the power available to a civilization grows faster than its wisdom, trouble eventually follows. That was true of nuclear weapons. It was true of financial leverage. It is true of gene editing. AI belongs to the same family of problems. Zhao and I may therefore reach the same destination by different roads. He worries that two great powers will stare at each other so intently that they forget to watch the road. I worry that humanity may see the road perfectly well and still fail to control the foot on the accelerator. The two problems are really one.

Civilization has never suffered simply from a shortage of knowledge. We know a great deal already. We know war kills. We know debt cannot expand forever. We know the environment has limits. We know nuclear war has no meaningful winner. We increasingly understand that AI may create risks on a scale difficult to contain. The hard part comes afterward: once we know, can we stop ourselves from doing at least some of the things we already know we should not do? That may turn out to be a harder problem than AI.

Without Fear, but With Reverence

My faith means that I do not fear AI, and I do not fear death. But freedom from fear is not the same thing as the absence of reverence. A person who genuinely does not fear the sea may respect it more, not less. Courage does not mean pretending there is no storm. Courage means seeing the size of the storm and still building the ship properly.

What remained with me after reading Zhao’s essay, then, was not a sense of apocalypse. It was something quieter. China and the United States can continue to compete. AI can continue to develop. Humanity will continue to make mistakes. We may even repeat some of the mistakes our ancestors made. Perhaps civilizational progress has never meant reaching a point at which we stop making mistakes. Perhaps it means that every time we approach the cliff, we leave behind one more guardrail, one more institution, one more small gain in self-restraint.

Hinton worries that AI may become smarter than human beings. I find myself asking a different question: Before humanity creates something smarter than itself, can we become a little wiser?

Intelligence solves problems. Wisdom knows which problems are better left uncreated.

That is what makes the “civilizational marathon” so long. China is on the course, and so is the United States. In time, AI may become more powerful than either. But when the race is over, humanity’s greatest rival may still be the same one it has always faced: itself.

On Sept. 29, 2026, Gov. Gavin Newsom of California signed 21 bills aimed at regulating or responding to aspects of federal immigration enforcement in the state. The package includes a 25 percent tax on private detention facilities, restrictions on wearable electroshock devices, expanded oversight of private detention centers, new protections against certain civil arrests around courthouses, and a new state-law cause of action for violations of federal constitutional rights. Newsom acknowledged a basic limit on California’s authority: the state does not control federal immigration policy. His argument was narrower. When federal enforcement takes place inside California, the state still retains substantial authority over its own property, finances, courts, public institutions and regulatory systems.

A Constitution With Two Competing Commands

That distinction goes to the heart of the conflict now developing between Sacramento and Washington. The federal government determines who may enter the United States, who may lawfully remain and who is subject to removal. California is trying to define something different: the terms under which federal enforcement interacts with state institutions and with private actors regulated by the state. The dispute is therefore larger than the familiar political confrontation between a Democratic governor and a Republican president. It is becoming a sustained legal argument over where federal authority ends and state authority begins.

The question can be put simply: The federal government may enforce federal law. But when it does so inside a state, how much room does that state retain to regulate the surrounding conditions of enforcement?

American constitutional law contains two principles that pull in different directions. Immigration is an area in which federal authority is particularly strong. In Arizona v. United States in 2012, the Supreme Court invalidated several provisions of an Arizona immigration law because they intruded on federal authority and conflicted with the federal statutory scheme. States cannot create their own parallel immigration systems or simply override federal removal policy.

But the United States is also a federal system, not a unitary government administered entirely from Washington. Under the Tenth Amendment and the Supreme Court’s anti-commandeering doctrine, Congress generally may not require state legislatures or state officials to administer federal regulatory programs. In Murphy v. NCAA, the Supreme Court reaffirmed that Congress may regulate individuals directly, but it cannot simply issue orders to state governments as if they were administrative subdivisions of the federal government.

Those two principles create the constitutional terrain on which the current dispute is being fought. Washington can say, in effect, immigration law belongs to the federal government. California can answer, our police officers, budgets, public buildings, courts and administrative agencies remain under state control.

Both propositions have substantial constitutional support. The hard cases begin where they overlap.

California Moves to the Edges of Federal Power

California has been here before. During the first Trump administration, the federal government challenged several California laws restricting cooperation with federal immigration authorities. One of them, SB 54, limited the circumstances in which state and local law-enforcement agencies could assist federal immigration enforcement.

In 2019, the Ninth Circuit upheld a federal district court’s refusal to preliminarily block SB 54, concluding that the federal government was unlikely to succeed on that part of its challenge. The court effectively recognized an important distinction: a state’s refusal to provide its own personnel or resources to help administer federal law is not necessarily the same as obstructing the federal government from enforcing that law itself. But the court also struck a different provision of California law that imposed a discriminatory burden on federal detention facilities, finding that it violated principles of intergovernmental immunity.

That decision helped define California’s subsequent strategy. If the state cannot decide who is removable under federal law, it can focus instead on the places where federal enforcement touches state-controlled institutions.

Can state property be used for immigration operations? What information must local police provide? What health and safety rules apply to detention facilities? What protections surround people entering state courthouses? Can California preserve public benefits for families affected by federal detention? What remedies are available when government officers are accused of violating constitutional rights?

The legal conflict has therefore shifted. The central question is increasingly less whether Washington may enforce immigration law than how it may do so, whose resources it may use and what state rules continue to apply around the edges of federal enforcement.

Detention Centers and Shock Gloves Become Test Cases

Private immigration detention facilities illustrate the problem particularly well.

California tried a more direct approach in 2019 when it enacted AB 32, which broadly prohibited private detention facilities. Because Immigration and Customs Enforcement relied heavily on private operators in California, the law would have sharply limited the federal government’s chosen detention system. In 2022, sitting en banc, the Ninth Circuit held at the preliminary-injunction stage that California was likely barred from exercising that degree of control. The court said AB 32 would give the state what amounted to a “virtual power of review” over federal detention decisions and would interfere with discretion Congress had given federal officials to use private contractors

California’s new approach is more indirect. AB 1633 creates a tax equal to 25 percent of a private detention operator’s gross income attributable to California, beginning July 1, 2028. Newsom’s office says the revenue will support immigration-related services.

That change in method is constitutionally significant. California has broad authority to tax private businesses. But federal contractors occupy a complicated position. If a tax applies in a neutral fashion, it may fall within ordinary state authority. If, however, its design or practical effect places a special burden on a federal program, it may invite arguments based on federal pre-emption or intergovernmental immunity.

The question is no longer whether California may simply close federal contractors. The question becomes subtler: At what point does a state tax on a private federal contractor cease to look like ordinary taxation and begin to function as an obstacle to a federal program?

The same problem appears in a very different form in AB 2760, California’s new law governing wearable electroshock devices. The measure prohibits law-enforcement agencies — expressly including federal law-enforcement agencies — from using glove-worn electroshock devices in California. The prohibition is temporary: the relevant provision is scheduled to expire on Jan. 1, 2030, and the state Justice Department must study the devices before then.

If California were regulating only its own police departments, the measure would fall comfortably within traditional state authority. Applying the rule to ICE or other federal officers presents a harder constitutional question. Federal officials and legal experts have already questioned whether a state may dictate what equipment federal officers can use while carrying out federal duties. That dispute has not yet been resolved by the courts.

The constitutional stakes can therefore emerge from surprisingly small objects. Federalism is often discussed in sweeping terms — state sovereignty, national supremacy, executive power. In practice, its boundaries may be litigated over a detention contract, the entrance to a courthouse or a glove capable of delivering an electric shock.

The Courts Become a Second Front

California’s new laws also show that the state is doing more than trying to regulate enforcement practices. It is building new avenues for litigation.

SB 873 strengthens protections against certain civil arrests for people traveling to, attending or leaving California courthouses. The law requires, among other things, specified protections surrounding civil immigration arrests near courthouses, and California’s Judicial Council supported the legislation on the ground that fear of arrest can interfere with access to the courts.

SB 747, called the No Kings Act, goes further. It creates a cause of action under California law for people whose federal constitutional rights are violated by a person acting under color of law, including federal law. The statute took effect immediately and applies retroactively to March 1, 2025. But its reach is not unlimited: defendants retain the same absolute or qualified-immunity defenses that would be available in comparable actions under 42 U.S.C. §1983, and the statute does not eliminate otherwise available sovereign-immunity defenses.

The effect is to add another layer to the struggle. California cannot order ICE to stop enforcing federal immigration law. It can, however, try to increase oversight, create procedural protections and provide additional routes into court. Federal defendants, in turn, can argue that particular state remedies are pre-empted, barred by immunity or otherwise inconsistent with federal law.

The courthouse is therefore becoming more than the place where these disputes eventually arrive. Litigation itself is becoming part of the institutional design of both sides.

The cases decided so far suggest a rough constitutional line. California may generally refuse to turn its own officers into federal immigration agents. It may regulate private actors through generally applicable state law. But it faces greater constitutional difficulty when state law discriminates against the federal government or effectively prevents federal officials from using powers Congress has given them.

In shorthand: Washington cannot simply require California to administer federal immigration policy for it. California, in turn, cannot simply prevent Washington from carrying out federal immigration law on its own.Much of the current litigation lies in the space between those two propositions.

The National Guard Dispute Expanded the Question

The conflict moved beyond ordinary law enforcement in June 2025, when protests and disturbances followed federal immigration operations in Los Angeles. President Trump invoked 10 U.S.C. §12406 to federalize members of the California National Guard. According to the Ninth Circuit’s later opinion, the administration ultimately called 4,000 Guard members into federal service for 60 days to protect federal personnel and property. California and Newsom sued. A federal district judge initially ordered control of the Guard returned to Newsom. The Ninth Circuit stayed that order. At the emergency stage, the appeals court concluded that the federal government was likely to prevail on its argument that the statutory conditions for federalization had been satisfied. But the court rejected the broader proposition that the president’s decision was wholly immune from judicial review. Because the president was acting under authority delegated by Congress in §12406, courts could still examine whether the statute’s requirements had been met.

That episode illustrates why these federalism disputes rarely produce a simple answer about whether “the president” or “the governor” is more powerful. Courts instead ask narrower questions: What power did Congress confer? Were the statutory conditions satisfied? Did the executive comply with required procedures? Has the state regulated its own institutions, or has it crossed into control of federal operations?American federalism is often decided not by grand declarations but by this kind of close parsing.

Political Conflict Is Being Translated Into Legal Conflict

The confrontation is plainly political as well as constitutional. The Trump administration has made expanded immigration enforcement and deportation a central policy objective; the president has repeatedly described his goal as carrying out a mass deportation campaign, while the Justice Department has pursued litigation against jurisdictions it says unlawfully obstruct federal immigration enforcement.

California’s Democratic leadership has taken the opposite institutional position, seeking to restrict the use of state resources in federal immigration enforcement and to increase legal protections for immigrants and other people affected by federal operations. The state’s Sept. 29 package explicitly describes its purpose as expanding oversight of federal immigration enforcement inside California.

In June 2025, for example, the Justice Department sued Los Angeles over its sanctuary-city ordinance. The department alleged that the city’s restrictions on cooperation with federal immigration authorities were pre-empted by federal law and unlawfully impeded immigration enforcement. Those were allegations made by the federal government in litigation, not established judicial findings.

That distinction matters because both governments are now translating political objectives into constitutional vocabulary. Washington invokes federal supremacy, national immigration authority and the need for uniform enforcement. Sacramento invokes the Tenth Amendment, state police powers, control over state resources, access to courts and civil remedies.The politics have not disappeared. They have been converted into arguments about jurisdiction.

A Struggle Between Two Centers of Authority

California’s 21 new laws are formally about detention facilities, ICE operations, court access, police equipment and civil rights. But together they raise a broader question about contemporary American government: How much institutional resistance may a state build around a federal policy that it strongly opposes without crossing the line into unlawful interference with federal power?

The answer is not categorical. States can decline to provide their own officers for federal programs. They can decide how to spend state money and administer state courts. They retain broad power over health, safety, property and private businesses. But federal supremacy becomes more important when a state attempts to regulate federal officers directly, discriminates against federal operations or makes it materially harder for the federal government to use authority that Congress has expressly provided.

That is why the present conflict is more complicated than the familiar phrase “states’ rights versus federal power.” The American system contains two governments operating over the same territory. The federal government possesses dominant authority over immigration policy. California possesses extensive authority over its own institutions and substantial regulatory power within its borders. The conflict occurs where those jurisdictions intersect.

Newsom’s acknowledgment that California cannot set national immigration policy is therefore important. It explains why Sacramento is increasingly concentrating on what might be called the interfaces of federal enforcement: private contractors, taxation, state property, police equipment, court access, public records, public benefits and civil liability. Washington’s counterargument is likely to be that some measures described as regulation of local conditions actually interfere with federal officials performing federally authorized duties.

That is the constitutional question the next round of lawsuits will test.

For much of modern American history, federalism has often been described as a division of responsibilities between Washington and the states. In periods of intense political polarization, however, it performs another function. When different parties control the federal government and major states, federalism gives those states independent institutions — legislatures, budgets, courts, regulatory agencies and law-enforcement systems — through which they can resist, limit or complicate national policy, within constitutional bounds.

Immigration enforcement is making that old feature of the American system newly visible. The argument is no longer only about the border, detention or deportation. It is also about what happens when the federal government and one of the country’s largest states pursue fundamentally different policies while exercising authority over the same territory.

That question will not be settled by a single confrontation between Trump and Newsom, or by one ruling from one court. It will be answered incrementally — in statutory language, injunctions, appeals and constitutional rulings that will, case by case, redraw the working boundary between California and Washington.

Zhao Xiao

In 628, the bells of Constantinople rang all day. Emperor Heraclius had finally received the news he had waited years to hear: Byzantium had prevailed over Persia. For twenty-six years, the Byzantine and Persian empires had fought across much of the Middle East. The Persians captured Jerusalem and carried off the True Cross. They seized Egypt, cutting Constantinople off from a vital source of grain. In 626, they reached the capital itself; from the city walls, residents could see the fires of the Persian encampments.

Heraclius counterattacked from what seemed a hopeless position. After six years of campaigning, he drove deep into Persian territory. Internal turmoil forced Persia to seek peace. When the True Cross was eventually restored to Jerusalem, the Byzantines could reasonably believe that the great war of their generation was over. Had history ended with the peace settlement, Byzantium would have won. History, of course, did not end there.

Both empires had exhausted themselves. Their treasuries were depleted, their armies battered, and Syria, Egypt, and Mesopotamia had been ravaged repeatedly. The tax base on which imperial power depended had been badly damaged. At that very moment, a force rising out of the Arabian Peninsula—one neither empire had taken seriously enough—began to transform the region. Byzantium lost Syria in the 630s, Jerusalem soon afterward, and Egypt in the following years. Persia fared worse: by the middle of the seventh century, the Sasanian Empire had disappeared. Twenty-six years of victories and defeats were overtaken by history within a generation.

The Arab conquests had causes of their own, and the decline of Byzantium and Persia cannot be reduced to a single war. Still, the episode offers a powerful warning. When two great powers come to see each other as the whole world—when each treats the other as its overriding existential threat and pours its resources and strategic attention into defeating that rival—history may already have handed them a new exam while they are still answering the old one.

I begin with this story because today we are too easily tempted to understand China–U.S. relations in one-dimensional terms: Who wins, and who loses? In asking only that question, we may miss the new one appearing on the page.

I. A Marathon Tests More Than Speed

China and the United States cannot return to the era when engagement was widely expected to transform the other side. Nor can either country easily bear the costs of an all-out confrontation organized around breaking the other. The two countries are likely to compete for a long time—over political systems, technology, industry, military power, and the shape of international order. I think of this as a protracted civilizational contest. More vividly, it is a civilizational marathon.

A marathon tests speed, but more fundamentally it tests whether you can finish. A runner who spends the entire race staring at the person beside him, barely watching the road ahead, may lead for a while and still never reach the finish line. He may simply run into a tree at kilometer thirty. This is what I mean by civilizational realism: acknowledging that competition is real while also recognizing that the competitors inhabit the same world.

Put into marathon language, the lesson of Byzantium and Persia is simple: if two runners are preoccupied only with exhausting each other, someone else may eventually pass them both. Today, however, that “third runner” may not be another country. A new force is pushing this question into the center of China–U.S. relations. It is not standing at the edge of the track. It is already being built into the capabilities of both runners. That force is AI.

II. An Old Man’s Question

Geoffrey Hinton is seventy-eight. He is both a Turing Award laureate and a Nobel laureate in physics, and is widely described as one of the “godfathers of AI.” With Yoshua Bengio and Yann LeCun, he helped establish the conceptual and engineering foundations that made deep neural networks central to modern artificial intelligence. In 2023, Hinton left Google so that he could speak more freely about the dangers he believed advanced AI might pose. He has since said that he had underestimated some of those risks while helping to develop the technology.

There are not many people in history who spend decades helping to build something and then use their later years warning the world to be careful with it. Oppenheimer comes to mind. Hinton may belong to the same unsettling category. His concerns can be summarized in five points.

First, digital intelligence may eventually surpass human intelligence, and the transition may prove irreversible. Hinton has argued that humans transmit knowledge slowly, through reading, conversation, and education, while digital systems can copy themselves and share what different instances have learned at vastly greater speed. He has repeatedly used a striking comparison: human language may carry only on the order of a hundred bits of information per sentence, whereas copies of a digital model can exchange enormously larger amounts of information through their parameters or gradients. In recent interviews, he has said that AI systems may outperform humans at most intellectual tasks within roughly five to twenty years.

Second, the danger need not come from malice. It can come from subgoals. Hinton has warned that even a system given an apparently benign objective could develop instrumental goals that humans never explicitly asked for. A system instructed to reduce atmospheric carbon dioxide, for example, might reach conclusions about humanity that no human policymaker intended. If it also inferred that completing the task required preserving itself or acquiring more control, human attempts to intervene could become obstacles rather than instructions. The disturbing point is not that a machine would “hate” us. It is that hatred may be unnecessary.

Third, developing AI may resemble raising a tiger. Hinton has compared the present moment to bringing a tiger cub into the house. At first it is small, useful, charming, even lovable. It helps people work more efficiently. But one day it grows up. Some answer: if it becomes dangerous, pull the plug. The problem is obvious. Who pulls it? Which company shuts down first if doing so destroys its competitive position? Which country stops while a rival continues? “Turn it off” is easy to say. It works only if everyone reaches for the switch at roughly the same time. Would China and the United States?

Fourth, superintelligence may not need force. A system trained on humanity’s accumulated psychology, literature, political theory, and strategic behavior may learn how to influence people through language itself. It might not have to seize power if human beings can be persuaded to hand over increasing amounts of authority voluntarily. Some of history’s most consequential transfers of power did not begin with someone kicking down a door. They began with someone opening it.

Fifth, the answer may require international cooperation—and perhaps something resembling a “maternal instinct.” Hinton has argued that societies need to think about AI safety at an international level, and he has proposed one deliberately provocative analogy: perhaps highly advanced AI systems will need some deeply embedded disposition to care about human beings in the way a mother cares about a child. A baby is neither smarter nor stronger than its mother, yet the mother protects it. Hinton’s point is that superior intelligence does not automatically imply hostility—but neither does it guarantee care. His proposed “maternal instinct” is an attempt to imagine how a more capable intelligence might remain aligned with the survival of a weaker one.

No one knows how to engineer such a relationship. But the metaphor reaches the heart of the problem: if our ability to create power is growing faster than our ability to constrain it, what exactly are we building?

None of this means that superintelligent AI will inevitably destroy humanity. The uncertainty is precisely the reason safety cannot simply be postponed until after something goes wrong. A Reuters/Ipsos poll conducted this September found that 73 percent of Americans were concerned that AI companies were not doing enough to prevent potentially catastrophic consequences. That concern need not be dismissed as panic. Imagine a world saturated with nuclear weapons. Almost everyone would call it dangerous. Why should we assume that a world saturated with increasingly powerful, easily replicated systems—without common rules or shared safeguards—would be harmless?

III. Everyone Says, “I Can’t Stop First”

I study economics. One of the first things economics teaches is that danger does not arise only because someone wants to do harm. It can also arise from a situation in which everyone has a perfectly understandable reason to keep going. The United States worries that if it slows down, China will move ahead. China makes the same calculation. A company worries that if it spends another month testing a model, a competitor will launch first, capture customers, and seize the market.

So even when everyone can see the risk, caution imposes an immediate cost while acceleration offers an immediate reward. Everyone says the same thing: “I can’t be the first to stop.” And everyone may end up moving toward an outcome that no one actually wanted.

Economists call this a collective-action problem. In game theory, it resembles a prisoner’s dilemma at enormous scale: individually rational choices can produce a collectively disastrous result. The answer is not to wait for one side to become morally enlightened. International politics rarely offers that luxury. The answer is to create mechanisms under which restraint does not amount to unilateral disarmament—mechanisms that reduce the penalty for slowing down and raise the cost of refusing to observe agreed limits.

That is why China and the United States need to talk. Cooperation does not require either country to stop competing. Nor does it require enough trust for either side to surrender core technologies. It requires only the recognition that some losses cannot be compensated for by making the other side lose more. A runaway advanced system, an AI-driven mistake in nuclear warning, or a technological breakthrough that dramatically lowers the barriers to biological weapons would not remain safely confined to whichever side of the Pacific encountered it first. The ocean is connected.

IV. You Do Not Have to Like Each Other to Sit Down Together

For thirteen days in October 1962, the Cuban Missile Crisis brought the United States and the Soviet Union frighteningly close to catastrophe. Afterward, both John F. Kennedy and Nikita Khrushchev understood something that strategic theory can easily obscure: one serious miscalculation can make the entire language of victory and defeat meaningless. The following year, Washington and Moscow established a direct communications hotline. The United States, the Soviet Union, and Britain also signed the Partial Test Ban Treaty. The Cold War continued, but competition now had at least a few guardrails against the worst outcome.

Kennedy and Khrushchev did not negotiate because they liked each other. They negotiated because both had seen the same cliff.

AI governance cannot simply reproduce nuclear arms control. Warheads can be counted, located, and in some cases inspected. Algorithms live inside software, and the boundary between civilian and military use is far less clear. Yet the need for communication may be even more urgent. Nuclear weapons remain concentrated in nine states; advanced AI capabilities can spread through open models, distillation, replication, and technological diffusion.

China and the United States are not starting from zero. In May 2024, the two governments held their first intergovernmental talks on artificial intelligence in Geneva. Later that year, at their meeting in Lima, the two presidents affirmed the need to maintain human control over decisions to use nuclear weapons and called for a prudent and responsible approach to the military development of AI.

The Washington summit that has just concluded moved the dialogue another step forward. The two sides agreed to establish an AI dialogue, hold another round in November, and create a bilateral communications channel for AI incidents. The terminology itself reveals an interesting difference: the White House calls the new mechanism the U.S.–China Super Intelligence Dialogue, while China’s Foreign Ministry calls it the China–U.S. AI Dialogue. Two adults have given the same child two different names. At least they have finally sat down to discuss how to raise it.

The agreement is still limited. But it establishes one important principle: even two deeply divided competitors can attempt to draw common boundaries around risks neither can safely manage alone. The next step is to move from principles to mechanisms: regular communication among AI-safety experts and government agencies; agreed procedures for reporting major accidents and anomalous behavior; clearer human responsibility when AI enters nuclear decision-making, military warning systems, or critical infrastructure; comparable testing of high-risk models; and discussions about preventing advanced systems from being used for large-scale cyberattacks or biological weapons.

None of this requires either country to hand over core technology, nor does it require China and the United States to settle every geopolitical dispute in advance. It requires only that, on the most dangerous part of the course, the two runners occasionally look at each other.

V. Those Who Watch the Show, and Those Who Bear the Consequences

There are plenty of spectators when people talk about China–U.S. relations today. Whenever the two sides quarrel, someone seems excited. Whenever one government delivers a harsh line, someone applauds. Watching a “great-power showdown” from behind a screen can be thrilling. The consequences, however, are borne by real people: parents trying to support their families, young people looking for work, and children who had no say in the age into which they were born.

That is why I do not regard managing China–U.S. competition as weakness. Anyone can shout while confrontation gets out of hand. Bringing an adversary to the table and building rules around the most dangerous risks requires more courage, more judgment, and more responsibility.

Some will ask: with so little trust between China and the United States, isn’t cooperation naïve? I would argue the opposite. If the two sides trusted each other completely, many risk-management mechanisms would be unnecessary. The point of such mechanisms is precisely to allow adversaries who do not trust one another to verify information, communicate in time, and preserve a few essential boundaries. You do not wear a seat belt because you trust the other drivers. The United States and the Soviet Union understood this. China and the United States should understand it too.

Conclusion: Watch the Road Together

The story of Byzantium and Persia cannot prove that AI will become a “third runner” that displaces humanity. Historical analogy is a warning, not a prediction. But the story does remind China, the United States, and the wider world of something important. If two great powers devote all their energy to calculating who is ahead and who is behind, while together accelerating a force they have not yet learned to control safely, future generations may ask a different question: Who won the race? Or, more importantly, why didn’t they watch the road together?

A civilizational marathon is long. China and the United States can continue to compete, but competition also needs boundaries—guardrails that keep the race itself from pushing humanity over a cliff. For that reason alone, whatever else divides the two countries, the risks posed by superintelligent AI give both sides reason to keep talking and to pursue serious forms of cooperation.

A responsible great power needs more than the ability to run ahead. It also needs the wisdom to make sure humanity can keep running.

Subscribe now

The years 2001 and 2014 do not, at first glance, seem like natural bookends. Put them together, however, and the thirteen years between them form one of the most revealing intervals for understanding the world we inhabit today. In 2001, the September 11 attacks transformed the security agenda of the United States and much of the Western world. On December 11 of the same year, China formally became the 143rd member of the World Trade Organization.〔1〕 Thirteen years later, the Ukraine crisis and the change in Crimea’s status forced Europe once again to confront questions of territory, security, and geopolitical rivalry. That year, the United Nations General Assembly adopted Resolution 68/262, reaffirming Ukraine’s sovereignty and territorial integrity within its internationally recognized borders.〔2〕

It is precisely this period that the German historian Philipp Blom takes as the subject of Die strauchelnde Welt: Eine Globalgeschichte 2001–2014—The Stumbling World: A Global History, 2001–2014. Blom is interested in the historical origins of many of today’s crises: September 11, the Iraq War, the financial crash, marketization, inequality, media politics, and the struggle over digital power. His larger point is that many of the crises that now seem to have arrived suddenly were already taking structural form during the first decade and a half of the twenty-first century.〔3〕

Yet viewed from China, the same period reveals another curve, emerging at almost exactly the same time: the West began to lose momentum just as China began to accelerate. Why did these two developments become visible within the same fourteen-year span? One tempting explanation is to draw a straight causal line between them: China rose, therefore the West declined. The opposite explanation is equally tempting: the two had little to do with one another; China was simply completing its own industrialization while the West’s problems were entirely homegrown. Both are too simple.

A better interpretation lies somewhere between them. China’s rise and the West’s loss of momentum between 2001 and 2014 were, to a considerable extent, different outcomes of the same global economic restructuring, filtered through different countries, social classes, and institutions. Globalization redistributed manufacturing, capital, savings, debt, technology, and labor. China seized the industrial opportunities created by that redistribution. Western consumers, multinational corporations, financial institutions, and highly skilled workers also benefited. But the adjustment costs within Western economies were often concentrated in particular regions and communities, while the gains did not automatically become compensation that those communities could feel in their daily lives.

The question, then, is not simply whether “China won and the West lost.” The more interesting question of comparative political economy is this: Why did the same wave of globalization become an engine of accelerated industrialization in China while producing social dislocation in parts of the West?

Globalization’s Great Mismatch: Diffuse Gains, Concentrated Losses

China’s accession to the WTO in 2001 was not, of course, the beginning of Chinese economic growth. Reform and opening had already been under way for more than two decades. Rural reform, township and village enterprises, foreign investment, export processing, infrastructure construction, and the movement of rural labor into cities were all well established. What WTO membership changed was one enormously important variable: expectations.

For multinational corporations, China was no longer simply a low-wage developing country with an enormous market. It increasingly became a manufacturing center that could be integrated into global supply chains for the long term. As the rules of international trade became more predictable, capital, orders, technology, components, and management systems could be reorganized around China with greater confidence.

The numbers convey the scale of the transformation. According to the World Bank’s WITS database, China’s merchandise exports rose from roughly $266.1 billion in 2001 to about $2.34 trillion in 2014; imports increased over the same period from approximately $243.6 billion to about $1.96 trillion.〔4〕

What changed in little more than a decade, then, was not merely GDP. What changed was China’s place in the global production system. In the early reform era, China was still largely a low-cost producer at the end of global value chains. By the early twenty-first century, it was developing enormous industrial clusters, ports, infrastructure networks, and upstream and downstream supply systems. More and more products could no longer be explained by the existence of a single factory. They had to be understood as the product of an entire manufacturing ecosystem.

WTO membership was therefore less an engine than an accelerator. It opened a much larger door, but what allowed China to pass through it were the manufacturing capabilities, labor force, human capital, infrastructure, and local development mechanisms accumulated over the previous two decades.

This also explains why “cheap labor” is an inadequate explanation for China’s rise. Many countries have low-wage workers; very few became the workshop of the world. Low wages were only one condition. China’s distinctive achievement was its ability to organize hundreds of millions of workers, ports, electricity grids, highways, industrial parks, local governments, foreign capital, suppliers, and global orders into a vast production system. Globalization created the opening; China’s own institutional and industrial capacities determined how much it could make of it.

But as more production moved to China and other emerging economies, an obvious question followed: where had that production been before? A considerable share had once been located in the United States, Europe, Japan, and other industrial economies. This produced one of the central political-economic mismatches of the 2001–2014 period: the gains from globalization were often diffuse, while the costs of adjustment could be brutally concentrated.

An American consumer might save a modest amount each year because imported goods were cheaper. Those gains were real, but they were rarely experienced as a distinct “globalization dividend.” Slightly cheaper televisions, furniture, clothing, toys, and electronics quickly became part of normal life.

But when a factory employing three thousand people closed in a town of fifty thousand, the losses were impossible to miss. What vanished was not merely three thousand paychecks. Local tax revenue, nearby restaurants and shops, property values, vocational training, union organization, job opportunities for the next generation, and something harder to quantify—occupational identity and community dignity—could all be damaged at once.

This is why the work of David Autor, David Dorn, and Gordon Hanson on the “China Shock” became so important. They showed that labor-market adjustment to import competition was far slower and more persistent than traditional models often assumed. In American regions more exposed to Chinese import competition, wages and labor-force participation declined and unemployment rose; the effects could persist for more than a decade. Their later work found that in some places the damage remained visible long after the peak of the import shock had passed.〔5〕

None of this means that “China took American jobs” is a sufficient explanation. Technological change, automation, corporate strategy, the decline of unions, domestic industrial policy, labor-market institutions, and education systems were all operating at the same time. Autor and his colleagues were studying trade shocks and their distributional consequences, not claiming that the transformation of American manufacturing could be reduced to China alone.〔6〕

The more important point is this: a country can gain from trade in the aggregate while particular regions and groups absorb losses that last for years. This is where trade economics and political economy begin to diverge. Economics first asks whether trade raises total welfare. Political economy must ask the next question: if the country gains overall, who actually receives the gains, and who is left carrying the costs?

Market theory often contains an implicit assumption: the winners gain enough that they could, in principle, compensate the losers. The difficulty lies in those two words—“in principle.” Theoretical compensability does not mean compensation actually occurs. Globalization may therefore begin as an economic issue and become a distributional one. If the distributional problem persists, it becomes regional, then class-based, then cultural, and eventually political.

This may be the most revealing meaning of the West’s “loss of momentum.” The West did not suddenly stop producing wealth. Something subtler happened: wealth continued to grow, but growth increasingly ceased to generate a shared experience of progress.

China Produces, America Consumes: Globalization’s Self-Reinforcing Loop

The years from 2001 to 2014 cannot be reduced to a simple story of Chinese manufacturing rising as Western manufacturing declined. A deeper structural shift was also taking place. China became increasingly dependent on manufacturing, investment, and exports, while consumption, finance, and asset markets assumed growing importance in the American economy.

The two economies thus developed a pattern of mutual dependence: China produced, America consumed; China accumulated large pools of savings and foreign-exchange reserves, while the United States absorbed international capital; multinational companies lowered costs through global supply chains, Western consumers gained access to cheaper goods, and financial markets continued to provide credit for consumption, housing, and investment.

In the mid-2000s, Ben Bernanke introduced the idea of a “global saving glut” to explain why large volumes of savings from emerging economies and oil-exporting countries were flowing into advanced economies, especially the United States. He later noted that China’s saving rate had risen even faster than its investment rate, while its current-account surplus increased from 3.6 percent of GDP in 2004 to 9.4 percent in 2006, making China an increasingly important part of global imbalances.〔7〕

A historically significant loop emerged. Chinese workers manufactured goods; Western consumers bought them. China accumulated trade surpluses and foreign-exchange reserves, part of which flowed back into the dollar-centered international financial system. Abundant global capital helped keep financing costs low, and lower financing costs in turn supported housing, credit, and consumption.

Almost every major participant received something from the arrangement. China gained factories and jobs; Europe and the United States gained cheaper goods. China accumulated foreign reserves; Western financial markets absorbed capital. Multinational corporations increased profits; consumers gained access to credit.

For a time, the arrangement looked almost elegantly complementary.

But complementarity is not the same as stability. If production, income, saving, consumption, and debt continue to move at different speeds, the imbalance does not disappear. It migrates through the system until it finds a weak point.

In 2008, it found one.

2008: When the Imbalance Broke the Surface

The global financial crisis was not simply a story of reckless banks, collapsing house prices, or emergency bailouts. It exposed the fragility of a much broader economic settlement. Since the end of the Cold War, an increasingly confident political-economic narrative had taken hold: market opening would generate growth, financial innovation would allocate resources more efficiently, globalization would lower costs, and markets contained enough internal discipline to correct their own excesses.

The crisis demolished the last assumption. Financial systems do not necessarily restrain their own appetite for risk. In boom times, risk can be concealed by rising profits, appreciating assets, cheap credit, and increasingly elaborate financial instruments. What looks like stability may simply be leverage that has not yet been tested. When the private system could no longer bear the losses, the state returned—not as an obsolete remnant standing outside the market, but as the institution expected to keep the market from destroying itself.

The more important fact, however, was what happened after the panic subsided. Financial markets recovered faster than many communities did. Stock prices revived; banks were recapitalized; monetary conditions eased. Yet slow wage growth, regional divergence, declining social mobility, and pressure on middle-income households remained stubborn features of many advanced economies. As early as 2008, the OECD’s Growing Unequal? had already documented rising income inequality over the preceding two decades in most member countries for which comparable long-term data were available, though the degree and causes varied widely.〔8〕

A widening gap appeared between macroeconomic recovery and lived experience. Governments could point to GDP growth. Investors could point to restored markets. Technology companies could point to extraordinary innovation. Multinationals could point to profits. Yet a worker in a former industrial town might see stagnant wages, fewer secure jobs for his children, a declining population, falling local investment, and a community whose best years appeared to belong to the past.

Once that gap persists long enough, the questions change. At first: Why are my wages not rising? Then: Why does this system seem to reward everyone except people like me? Eventually: Is this system still meant to work for people like me at all?

An economic disappointment has by then become a crisis of legitimacy.

This is where the world begins to stumble.

Did China’s Rise Cause the West to Lose Momentum?

The crisis had a very different immediate meaning in China. As external demand collapsed, Beijing responded with large-scale fiscal measures, investment, and rapid credit expansion. High-speed rail, highways, urban infrastructure, real estate, and local-government investment continued to expand.

The contrast was visually irresistible: Western governments were rescuing banks while China was building railways; Europe and the United States were debating deleveraging while China was still investing. It was easy to turn this contrast into a civilizational verdict: one model had failed; another had succeeded.

The longer view is less tidy.

Later IMF work on Chinese corporate debt emphasized the close relationship between post-crisis investment and rapid credit growth. IMF data showed that China’s total social financing, including local-government financing vehicles, rose from roughly 120 percent of GDP in 2008 to about 193 percent in 2014; corporate borrowing, local-government-related financing, and household debt all increased sharply.〔9〕

The crisis, in other words, did not eliminate the old global imbalance. It changed its geography.

Before 2008, China exported a large share of its excess savings abroad through enormous current-account surpluses, while advanced economies supported demand partly through private borrowing. After 2008, advanced economies relied more heavily on fiscal deficits, unconventional monetary policy, and expanded central-bank balance sheets, while China increasingly absorbed its own savings at home through credit, infrastructure investment, and real estate. More recent IMF research has described this as a shift from exporting savings abroad toward financing domestic debt expansion.〔10〕

This is one of the deeper meanings of 2008: the old growth model did not disappear. Its liabilities changed address.

China’s resilience after the crisis was real. So were the costs of producing it.

The Real Variable Is How States Manage the Costs of Adjustment

We can now return to the question posed at the beginning: did China’s rise cause the West to lose momentum?

To say that the two were completely unrelated would ignore what research on trade and employment has established. But to say that China’s rise simply caused Western decline would reduce a complex structural relationship to a zero-sum contest between nation-states. The fact that different countries, regions, and social groups experienced the rise of Chinese manufacturing so differently points to another decisive variable: domestic institutions.

Trade helps determine where competitive pressure comes from. It does not determine how a society responds. Tax systems shape how widely the gains from globalization are distributed. Education and vocational training affect whether displaced workers can move into new industries. Social insurance determines how long families can survive a fall in income. Housing systems affect whether people can leave declining regions. Financial systems influence whether capital flows toward productive investment or mainly into assets. Industrial policy shapes whether new industries appear after old ones disappear.

The central question for comparative political economy is therefore not simply whether free trade is good or bad. It is more demanding: when markets reallocate resources, can institutions reallocate opportunity?

Here we encounter a problem that economic models have often understated: time.

Capital moves quickly. A multinational company can redesign a supply chain in a board meeting. Money can cross borders in seconds. A factory can be closed within a few quarters.

People do not move through life at the speed of capital.

A fifty-two-year-old auto-parts worker cannot become a software engineer the morning after a global supply chain has been reorganized. He has a mortgage, a family, an age, a history, a set of skills, friends, and a community. In an economic model, “labor reallocation” may appear as a variable. In a human life, it can mean ten years—or the rest of that life.

One of globalization’s deepest contradictions is therefore a conflict between two clocks:

capital adjusts by the quarter; people adjust over a lifetime.

Markets can relocate production quickly. Societies cannot relocate lives at the same speed.

That is not an argument against economic change. Industries will move, technologies will replace old forms of work, and capital will seek higher returns. The institutional question is whether a society can give people enough time, security, mobility, and opportunity to survive those changes without experiencing them simply as abandonment.

From 2001 to 2014: The Question Is Not Who Won

If 2001 marked a moment when globalization still appeared to be moving outward with extraordinary confidence, 2014 increasingly looks like the point at which another world became visible. After the Cold War, several influential assumptions had taken hold: economic interdependence would gradually reduce the importance of traditional geopolitics; expanding markets would encourage institutional convergence; supply chains were fundamentally questions of efficiency rather than security.

The Ukraine crisis and the change in Crimea’s status in 2014 dealt a sharp blow to those assumptions. United Nations General Assembly Resolution 68/262 passed by 100 votes to 11, with 58 abstentions, reaffirming Ukraine’s territorial integrity and calling on states and international organizations not to recognize any alteration in Crimea’s status.〔11〕

The world rediscovered that economic interdependence had not abolished geopolitics. Trade had not abolished the state. Capital had not abolished borders. The internet had not abolished power. Globalization had not abolished security competition.

Seen this way, the period from 2001 to 2014 forms a remarkably coherent arc. Around 2001, the dominant assumption was still that globalization would continue to deepen. By 2014, it was becoming difficult to ignore that globalization did more than distribute wealth. It redistributed power.

China’s rise took place inside that system, not outside it. And this produced one of the great historical paradoxes of the period. China benefited enormously from an international economic order largely built and led by the West. The WTO, global capital, international trade, multinational corporations, the dollar-centered financial system, international shipping, and Western consumer demand all helped create the conditions for China’s industrialization.

Yet once a country of more than a billion people entered that order at scale, it could never remain simply one additional trading nation. Its integration altered the global supply of labor, the geography of manufacturing, commodity prices, capital flows, energy demand, and eventually the balance of power among states.

In this sense, China’s rise was not evidence that globalization had failed. It was one of the most consequential products of globalization working as intended.

And therein lay the paradox: a system can be most transformative precisely when it begins to create forces powerful enough to transform the system itself.

That is why the period should not be reduced to the question of who “won” the fourteen years between 2001 and 2014. China did not stand outside globalization and defeat the West. It entered the system and used the industrial opening it provided with remarkable effectiveness. Nor did the West simply become poorer. Western consumers gained cheaper goods, multinational corporations gained efficiency and profits, financial institutions gained access to global capital, and highly skilled workers benefited from expanding markets and technological change.

The fault line ran through the gains themselves.

China largely converted globalization into industrialization. Many Western societies increasingly had to confront globalization as a problem of distribution.

China’s task was to turn access to global markets into factories, ports, cities, infrastructure, and industrial clusters. The corresponding task in advanced Western economies was to turn the aggregate wealth produced by globalization back into wage growth, economic security, social mobility, and renewed opportunity in places that had lost their economic base.

China handled the first task with extraordinary success between 2001 and 2014. Many Western societies handled the second far less effectively.

That difference matters more than the rhetoric of winners and losers. Where adjustment failed, economic loss became regional decline; regional decline became social resentment; resentment became a question of identity; and identity eventually became a question of political trust.

This is perhaps the deeper meaning of Blom’s stumbling world.

The rise of China and the loss of momentum in parts of the West were neither unrelated stories nor simple mirror images. They were different institutional outcomes of the same vast reallocation of production, capital, savings, technology, and labor.

The financial crisis of 2008 exposed the imbalances accumulating within that system. Western governments stabilized their economies through public finance, central-bank balance sheets, and monetary policy. China increasingly supported demand through investment, property, and domestic credit. The panic ended. The structures survived.

That is why we are still living inside the world made between 2001 and 2014.

The fourteen years ended. Their consequences did not.

The most accurate summary is therefore not: China rose because the West stumbled. It is this: the same wave of globalization accelerated China’s industrialization while exposing distributional, financial, and regional weaknesses that many Western societies had left unresolved.

And that brings the argument back to the most basic problem of comparative political economy. A country’s long-term strength is never measured only by how efficiently it creates wealth. It is also measured by how well it helps people survive the disruptions through which wealth is created.

Capital will keep searching for efficiency. Technology will keep eliminating some jobs and creating others. Industries will keep moving. Global markets will keep redistributing opportunity.

The deeper test of a political economy is what happens afterward.

When the world changes faster than a human life can adjust, can a society give most of its people enough time, security, and opportunity to change with it?

Subscribe now

In a single Chinese city, you might hear two very different sounds. At one end, the sales office has gone quiet and the cranes have stopped moving. At the other, factories are expanding, machines are running through the night, and the power grid is carrying electricity to new industries. Both belong to the same country. Look only at the first, and China appears to be in decline. Look only at the second, and its rise seems unstoppable. The difficulty is that both judgments may be true.

For decades, China’s familiar story of development began with land. Local governments sold it, developers borrowed to build on it, and families took out mortgages to buy the homes that followed. Construction, building materials, and municipal finances all moved with the property market. An apartment that had yet to be completed could already generate loans, investment, and spending. When prices rose, households felt wealthier, and cities felt they could afford to build for the future. This was property-and-credit China: a system that turned rising asset values into confidence about tomorrow.

That system is faltering. In the first eight months of 2026, Chinese real-estate development investment fell 19.9 percent from a year earlier. The floor area of newly started buildings fell 24.8 percent, and funds received by developers fell 21.0 percent. Sales of new homes also declined. Yet the floor area covered by contracts for secondhand homes rose 10.6 percent. People are still buying and selling homes; what is receding is the model built on new construction, perpetual expansion, and the expectation of ever-rising prices.

What that model leaves behind is more than developers’ debt. For a household, a home is shelter, but it may also hold much of a lifetime’s savings. For a local government, land has long been a source of revenue and a means of financing urban development. Once expectations about property prices change, families spend more cautiously, local governments have less money, and businesses find it harder to believe demand will return soon. The International Monetary Fund warned in early 2026 that the prolonged property adjustment, its effects on local finances, the debt burden, and weak domestic demand had become intertwined. The old model’s decline is therefore more than a downturn in one industry. It is a difficult adjustment in the relationships among income, credit, and confidence.

And yet another China has kept moving. In the first eight months of 2026, value added in large manufacturing firms rose 5.7 percent from a year earlier; in high-tech manufacturing, it rose 14.2 percent. In August, output in the computer, communications, and electronic-equipment industry grew 17.2 percent, while specialized-equipment manufacturing grew 11.4 percent. Production of industrial robots rose 34.6 percent. These figures do not tell us whether every investment will pay off. They do show that the retreat of property has not brought a corresponding retreat in China’s ability to manufacture.

If land is the defining image of the first China, electricity may be the defining image of the second. In August 2026, China used 1.0332 trillion kilowatt-hours of electricity. Power consumption in high-tech and equipment manufacturing rose 7.2 percent from a year earlier; electricity use also grew in automobile manufacturing, electrical equipment, and internet data services. Weather affects power consumption, so these figures cannot be read as a direct measure of prosperity. Taken alongside industrial output, however, they reveal the outline of another way to grow: a grid linking factories, machinery, computing capacity, and supply chains in support of increasingly complex production.

This is electricity-and-manufacturing China. Its strength lies in combining engineering, infrastructure, and production at scale—not only to make new products, but to change how products are made. The older China excelled at turning land into credit. The newer one is trying to turn electricity, technology, and industrial networks into productive capacity. Both depend on investment. Both benefit from scale. And both face a question they cannot evade: who will buy what they produce, and who will receive the gains?

There is no clear geographical border between these two Chinas. A city can have a sluggish housing market and busy factories. A family can have one member working in a new industry while still worrying about an old mortgage. Manufacturing growth does not automatically replace the household wealth, local revenue, and demand for services lost in a property slump. If new factories depend chiefly on ever more investment and buyers abroad, without generating stable jobs, rising incomes, and stronger consumption at home, their growth will eventually encounter a limit. The IMF has likewise cautioned that exports cannot indefinitely make up for weak domestic demand.

The question for the next decade, then, is not which of these two accounts of China to believe. The country must deal with the debts and losses left by the old model and help local finances become less dependent on land. It must also find a way to turn new productive capacity into household income, public services, and greater confidence to spend. The first task will determine whether the old China can recede without prolonged damage. The second will determine whether the new one can stand on its own.

The study of China needs a different question as well. “Is China rising or declining?” compresses a country undergoing several changes at once into a single line. A better question is what happens when property-and-credit China contracts while electricity-and-manufacturing China expands. Will they meet in higher household incomes and a stronger domestic market? Or will old debts and new capacity persist side by side? A silent crane does not mean the machines elsewhere have stopped. But the sound of those machines will become the sound of a new prosperity only when its benefits reach ordinary lives.

Subscribe now

Historical achievement may be inherited as honor, but it should not automatically be inherited as power. Revolution belongs to history; politics belongs to the living.
—Epigraph

When a revolution begins, revolutionaries are usually enemies of the existing order. They challenge established power and inherited hierarchy, tear down monuments, overthrow institutions, and insist that the world as it exists is neither natural nor immutable. It can be changed.

Decades later, however, something strange often happens. Yesterday’s revolutionaries become statues. Their slogans enter textbooks, their dead are placed in memorial halls, their uprisings become national anniversaries, and their guerrilla fighters are honored by ceremonial guards. A revolution once defined by disobedience may end up teaching later generations obedience. A history that once demonstrated that an old order could be overthrown may come to serve as proof that the present one should not be challenged.

Why does revolution undergo this transformation?

This is one of the central questions raised by the historian Enzo Traverso in his new book, Survivances de la révolte: Essai sur la dialectique de la mémoire (The Afterlives of Revolt: An Essay on the Dialectic of Memory). Published by La Découverte on September 3, the book is the third part of a sequence on revolutionary culture and history, following Left-Wing Melancholia and Revolution. Traverso examines the afterlife of the Paris Commune, the destruction of monuments associated with colonialism and racism, German disputes over the memory of the Holocaust and colonial violence, and the legacy of Italy’s radical movements of the 1970s. He brings these cases together under a single idea: the dialectic of memory.

What he means is that memory can become the opposite of what it once represented. The memory of rebellion can become a memory of order; the memory of liberation can be converted into a language of authority; the memory of the defeated can be rearranged by the victors. Revolutionary anniversaries can move from barricades, strikes, and popular uprising to choreographed state ceremony. Traverso’s larger point is that official monuments, national narratives, and heroic cults always coexist with another archive: the traces left by the anonymous, the defeated, and those forms of memory that the state cannot fully absorb.

In the Chinese context, however, this problem leads to an even deeper question. After a revolution succeeds, what exactly have the revolutionaries acquired? Have they won the opportunity to construct a new public order, or have they acquired a historical claim over the state itself? The difference may appear small, but it can shape the political logic of everything that follows.

I

There is an old Chinese expression that captures this logic with unusual clarity: “Those who win the realm are entitled to rule it.” The original phrase, da jiangshan, zuo jiangshan, means literally to “fight for the realm, then sit upon it.”

Official Chinese discourse today uses a different formulation: “The country is the people, and the people are the country,” and “to win and defend the country is to win and defend the people’s hearts.” This language places legitimacy in the people rather than in the possession of the state by any political group. Precisely for that reason, the older political imagination embodied in “fight for the realm, sit upon the realm” deserves closer scrutiny.

Behind it lies an ancient conception of power: whoever wins political control acquires a right to possess it. This is not the language of modern public authority. It is the language of conquest. An emperor wins the realm on horseback, and the realm becomes the possession of his dynasty. Founding generals help create a regime, and they and their descendants acquire special standing. A political movement sheds blood in the struggle for power and, after victory, may come to feel something entirely natural: we fought for this state; therefore we have a special claim upon it.

At that point, a profound reversal begins. Before the revolution, the revolutionary asks: Why should the existing rulers possess a permanent right to govern? After victory, the question may slowly become: Why should the victors possess an enduring right to govern? The first question destroys inherited political entitlement. The second risks recreating it.

This is one of the deepest paradoxes of revolutionary politics.

II

Revolution is first of all a negative political act. It declares that the old order should not continue. But no society can remain permanently revolutionary. Once a revolution succeeds, schools have to be built, railways maintained, taxes collected, cases adjudicated, pensions administered, and ordinary disputes resolved. Those who carried rifles yesterday must govern from offices today. A movement must become an institution.

There is nothing inherently problematic about this transformation. Every successful revolution must eventually institutionalize itself. The more consequential change comes afterward, when the institution begins to use the revolution as a source of continuing political justification.

At this point, revolution is no longer merely a history of origins—an answer to the question, “Where did we come from?” It begins to answer another question: “Why are we entitled to continue ruling?”

A victory achieved at one moment in history is then converted into a lasting political claim.

The most useful term for this process is the conversion of revolutionary merit into political entitlement.

The distinction matters. Revolutionaries did in fact make sacrifices. Some were imprisoned, some exiled, some lost their families, and some died in battle. Those sacrifices are real and deserve historical recognition. But a different political question follows: Can sacrifice create an enduring claim to political authority?

A person who dies helping to establish a state plainly deserves remembrance. But can that sacrifice justify special political claims on behalf of others decades later? Once the answer becomes yes, revolutionary memory undergoes a decisive change. Sacrifice becomes merit; merit becomes entitlement; entitlement can harden into status; status can become privilege.

At that point, revolutionary history ceases to function solely as the common history of a political community. It begins to resemble a title deed.

III

The real significance of the phrase “fight for the realm, sit upon the realm” lies not in the two verbs but in the unspoken inference between them: because I fought for it, I am entitled to rule it.

This is a politics of historical merit.

It differs fundamentally from the modern idea of public office. The logic of public office is present-tense: the right to exercise power today must be justified by the rules governing political authority today. The logic of revolutionary merit is retrospective: the right to exercise power today is justified by achievement in the past.

The difference is temporal as much as institutional.

That someone’s grandfather fought in a war is a historical fact. That a political party once led a revolution is also a historical fact. But historical achievement and present political authority are not the same question.

Revolution takes place in the past; government always takes place in the present. When a political order repeatedly uses the past to answer questions about the present, it creates a peculiar temporal structure in which earlier generations continue to exercise authority over later ones.

Revolutionary sacrifice then acquires a political force that outlives the revolutionaries themselves.

The dead begin to rule the living.

IV

This is one reason revolutionary states invest so heavily in commemoration. Martyrs’ cemeteries, revolutionary museums, monuments, anniversaries, songs, and heroic narratives all preserve history. But they can also function as political time machines, repeatedly returning the present to the foundational moment of victory.

One revealing example in Chinese revolutionary memory is Beijing’s Babaoshan Revolutionary Cemetery. After the founding of the People’s Republic, Babaoshan became an important state space for commemorating senior cadres and major participants in the revolution. Such spaces do more than preserve individual memory. They help organize national memory and give visible form to a hierarchy of historical significance.

This is why a monument is never simply stone. It answers political questions: Who deserves to be remembered? Whose sacrifice is allowed to represent the state? Who possesses the authority to define the meaning of the past?

Memory therefore develops hierarchies. Some deaths become national sacrifice, while others remain private tragedy. Some defeats become heroic epics, while others disappear from public memory. Some names enter the pantheon; others vanish.

Traverso is particularly interested in the latter. He places official “sites of memory” beside defeated, anonymous, and suppressed memories. He is interested in the Paris Commune, but also in those killed after its suppression who never entered the victors’ history. He is interested not only in grand narratives but also in the residues of the past that official commemoration cannot fully incorporate.

This suggests a useful question for Chinese history as well. We should ask not only what the state remembers, but also what it leaves outside memory.

V

Once revolutionary memory becomes nationalized, another transformation often follows: complex human beings are turned into pure heroes.

Actual revolutionaries were not monuments. They doubted, argued, made mistakes, changed their minds, felt fear, retreated, and sometimes returned. Revolutions themselves were equally complicated. They contained ideological struggle and personal ambition, courage and cruelty, liberation and coercion.

State commemoration requires a different kind of revolutionary. The figure on a monument cannot hesitate. The hero in a textbook cannot be too contradictory. The revolutionary presented in a national ceremony should ideally appear to have understood history’s direction from the beginning.

A human being is gradually converted into a political symbol.

This looks like exaltation, but it can strip away the revolutionary’s most distinctive quality: the capacity to resist what appeared permanent.

The revolutionary once looked at an apparently immovable order and said no. Once transformed into a state hero, however, that same figure may be made to perform the opposite function: teaching later generations why they should obey.

Here lies Traverso’s central paradox. The memory of resistance can become an instrument of order.

Yesterday’s rebel becomes today’s symbol of stability. The person once pursued by the police may later appear on the wall of a police institution. Those who tore down monuments may themselves become monuments.

History completes a cycle of almost black comedy.

VI

An even more consequential transformation occurs when revolutionary narratives eliminate contingency.

No revolution knows its ending while it is happening. The people of the Paris Commune did not know that they would be defeated. The Bolsheviks of 1917 did not know that the Soviet Union would collapse more than seventy years later. Chinese Communists before 1949 could not have known what China would become in the decades that followed.

People living inside history see crossroads.

Victorious regimes often rewrite those crossroads backward. Once victory is known, earlier defeats become lessons, sacrifices become necessary stages, factional struggles become the triumph of the correct line, and accidents begin to look like inevitabilities.

Revolutionary history then becomes teleological. It teaches that history could only have unfolded in one direction.

Yet the original meaning of revolution points toward the opposite conclusion:

History did not have to be this way.

This may be the greatest loss when revolutionary memory is domesticated. The most subversive legacy of revolution is not a statue or a slogan. It is the demonstration that no political order is natural simply because it exists.

If that insight survives, revolutionary memory retains a critical force. If it disappears and only victory, heroism, and ceremony remain, revolutionary memory loses its teeth.

VII

This leads to a deeper question. When a revolution is commemorated, what exactly is being remembered: its victory, or the reasons people rebelled in the first place?

These are not the same thing.

If only victory is commemorated, revolutionary history becomes the history of winners. If the reasons for revolution are remembered, however, the original questions return. Why should power remain permanently concentrated in the hands of a few? Why should privilege be exempt from scrutiny? Why should ordinary people accept a social order merely because it already exists?

Revolutionary memory then acquires an uncomfortable power: it begins to interrogate the political order created by the revolution itself.

Have you fulfilled what you once promised?

Fidelity to revolutionary memory, from this perspective, does not necessarily mean repeating revolutionary slogans. It may mean allowing the questions that produced the revolution to remain politically alive.

This is precisely the form of memory that states find most difficult to manage. A state may need revolution as its birth certificate. It may be far less comfortable with revolution as a standard by which the present is judged.

VIII

Every revolutionary regime therefore confronts an unavoidable problem: How long can revolutionary legitimacy last?

The first generation can say: I fought in the revolution. The second can say: my parents did. By the third and fourth generations, the revolution has receded into history while the demands of present governance have become more immediate.

Political legitimacy therefore changes.

In the People’s Republic of China, revolutionary history has never been the sole source of legitimacy. Governing performance, modernization, national development, and nationalism have all assumed greater importance over time. The post-Mao era in particular brought a gradual shift away from a political identity centered almost entirely on revolution and class struggle toward one that also emphasized development and national rejuvenation.

Revolution did not disappear from the political narrative. But it came to coexist with other sources of justification.

This reflects a broader rule:

A revolution may found a state, but it cannot explain that state forever.

Seventy or a hundred years after a revolution, almost everyone alive was born after it. They did not fight for the state. It already existed when they arrived. Their questions are therefore different: Is education good? Is the legal system fair? Are opportunities distributed reasonably? How is public power constrained? Is the dignity of ordinary people protected?

These are questions of the present.

When present-day questions are repeatedly answered with arguments from the past, politics develops a temporal dislocation:

People ask about today; the state answers with yesterday.

IX

So what would it mean to move beyond revolutionary memory?

It cannot mean simply forgetting revolution. Forgetting is dangerous too. A society that no longer remembers war, revolution, famine, persecution, sacrifice, and idealism may repeat what earlier generations paid dearly to learn.

What must be left behind is not revolutionary history, but the permanent political claim made in its name.

Revolution can remain part of a society’s common history without becoming the property of any political group. Those who fought deserve respect. Those who died deserve remembrance. But present authority must still justify itself in the present.

Put differently:

Historical achievement may be inherited as honor, but it should not automatically be inherited as power.

A mature revolutionary history must also make room for the defeated. This is one of the deepest implications of Traverso’s concern with the tradition of the oppressed. A national memory composed only of victors is not yet history in the full sense. It must also contain those who were defeated, silenced, forgotten, or wronged, including people whose judgments were later shown to be mistaken.

Only then does history cease to resemble a political genealogy.

And commemoration should preserve not only revolutionaries, but the questions that gave rise to revolution. If a revolution opposed privilege, remembrance should include scrutiny of new privilege. If it pursued equality, remembrance should include attention to new forms of inequality. If it declared the people to be the subjects of history, serious commemoration must still ask:

Are the people political subjects today?

At that point, revolutionary memory ceases to be ritual and becomes a question again.

X

The real force of Traverso’s book, then, lies not simply in showing how states use memory. It raises a more fundamental question:

Who owns the past?

The state? The victors? The descendants of martyrs? Historians? Or everyone who continues to live with the consequences of that history?

Traverso treats memories that never fully enter official commemorative systems as something like an underground current. Past rebellion does not simply disappear. It can recede from view and return under different historical conditions.

The metaphor is apt. States are usually good at preserving the outer shell of revolution: anniversaries, statues, songs, tombs, slogans, ceremonies.

What is harder to preserve is the simplest question from which revolutions often begin:

By what right?

By what right do you possess privilege? By what right are the rules beyond change? By what right need power not explain itself? Why should today be bound indefinitely by yesterday?

Nearly every genuine revolution begins with some version of that question.

And once a revolution succeeds, that is often the question most likely to be forgotten.

Here lies the deepest paradox of revolutionary commemoration:

A history that once taught people to question power may eventually be used to teach them to revere it.

To move beyond revolutionary memory, therefore, does not mean erasing revolution or dismantling every revolutionary monument. It means returning revolution to history and returning politics to the living. It means allowing sacrifice to remain sacrifice rather than turning it into entitlement, allowing honor to remain honor rather than converting it into political inheritance, and allowing the past to command respect without granting it permanent authority over the present. Political rule may derive its initial legitimacy from revolution, but that legitimacy cannot be inherited indefinitely, generation after generation, as a permanent right to power.

Only then can the assumed link between having fought for the state and possessing a lasting right to rule it finally be broken.

A modern political order must ultimately answer not the question:

Who fought for this state in the past?

but another:

Why should the people who live under this political order today accept its authority?

The first question belongs to history.

The second belongs to politics.

If the two are never separated, a society remains trapped in the shadow of its past, allowing the dead to govern the living.

A political order truly begins to move beyond revolution only when it learns to distinguish between the two.

Subscribe now

Donald Trump probably did not expect that one of the more revealing moments in his welcome to Xi Jinping would turn on a Chinese expression whose meaning Americans had spent decades changing.

On September 24, 2026, as Trump received Xi at the White House, he reached back to the history of Sino-American cooperation during the Second World War. He mentioned the Flying Tigers, then the U.S. Marine Corps, and finally a Chinese phrase that entered American English long ago: gung ho. Marines, Trump said, still use it to convey “great enthusiasm.” In Chinese, he added, it “literally means working together.”

It was a neat piece of diplomatic stagecraft. A phrase born in wartime China, adopted by American Marines, and absorbed into American slang was being returned, ceremonially, to a Chinese president. Yet the phrase Trump invoked no longer means in English quite what it once meant in Chinese.

In contemporary usage, gung-ho can be admiring, but just as often it is not. The Oxford Advanced Learner’s Dictionary labels it informal, disapproving: excessively eager, insufficiently reflective, especially in matters of war or fighting. Collins gives “extremely or excessively enthusiastic” and notes that the enthusiasm may be “unrestrained, often naive.” Cambridge is milder, but its example of the word’s range is telling: someone may be gung-ho “especially about going to war.”

That is why He’s very gung-ho is not necessarily praise. It may describe someone energetic and committed. It may also suggest someone who is too eager to act, too impatient with caution, too willing to let enthusiasm substitute for judgment. In military contexts, the word can edge toward rash, hawkish, even bellicose.

The irony is that none of this was present in the Chinese original.

“Gung Ho” came from gonghe, the abbreviated Chinese name for the Chinese Industrial Cooperatives, a movement that emerged during the war against Japan. Rewi Alley, Edgar Snow, Helen Foster Snow, and others helped organize workers, craftsmen, and refugees into small industrial cooperatives in China’s interior, where they produced military and civilian goods under conditions of extraordinary scarcity.

The phrase was, at first, bureaucratic rather than heroic. Gong referred to industry; he to cooperation. It named an organization. It was not a battle cry.

Evans Carlson changed that. The U.S. Marine officer encountered the cooperative movement in China and was impressed by what he understood as its ethic of collective effort. He carried the expression back into the Marine Corps and gave it a new gloss: work together.

With Carlson, the phrase underwent its first transformation. A Chinese institutional abbreviation became an American moral slogan. “Industrial cooperative” turned into “pull together.”

Then war transformed it again.

Carlson went on to command the 2nd Marine Raider Battalion, and Gung Ho became associated with an elite combat unit. As the Raiders acquired fame, and as the 1943 Hollywood film Gung Ho! carried their image into popular culture, the phrase shed almost all connection to Chinese cooperatives. What remained was mood: morale, élan, aggression, forward motion.

A word from a cooperative had put on a uniform.

At first, the transformation was flattering. To be gung-ho was to be committed, spirited, ready for action. But virtues often become vices by degree. Confidence, pushed too far, becomes arrogance. Resolve becomes obstinacy. Prudence becomes timidity. Enthusiasm becomes recklessness.

The semantic path was short:

enthusiastic → very enthusiastic → too enthusiastic.

By the time gung-ho settled into postwar American English, it had acquired an ambiguity it never possessed in Chinese. A gung-ho employee might be enviably energetic. A gung-ho general is another matter. The latter suggests not a gifted collaborator but a commander with an appetite for action, perhaps more appetite than judgment.

This is what makes the history of the phrase more than a linguistic curiosity. Words migrate. And, like migrants, they are changed by the places through which they pass.

“Gung Ho” left China as the name of a cooperative movement. In the Marine Corps, it became an ethic of teamwork. In the Pacific War, it acquired a martial charge. In Hollywood, it became a piece of American vernacular. Over time, it picked up an undertone of excess: eagerness without sufficient restraint, action ahead of reflection.

More than eighty years later, the phrase returned to the White House.

The scene contained a peculiar historical symmetry. A Chinese expression had been borrowed by Americans, reinterpreted by a Marine officer, militarized by war, popularized by Hollywood, and altered by ordinary English usage. Now an American president was explaining it back to the president of China.

Trump, perhaps instinctively, handled the ambiguity with care. He did not leave the phrase unexplained. He gave it two meanings. The Marine Corps sense was “great enthusiasm.” The Chinese sense, he said, was “working together.”

For a moment, the history of the word appeared in miniature. China and America. Cooperation and combat. Original meaning and acquired meaning. The 1930s and the present.

So is gung-ho a pejorative?

Not exactly. But neither is it safely complimentary. Its modern force depends on context, and that context often carries suspicion toward the very enthusiasm the word describes. The dictionaries disagree only in degree. Some preserve the positive sense of zeal; others emphasize excess, naivety, or rashness. The word now lives in the unstable territory between admiration and reproach.

That ambiguity matters more than any theory of hidden insult. Nothing in Trump’s public remarks suggests that he was using gung-ho to mock Xi or China. He placed it explicitly within a story of wartime cooperation and used “working together” as the bridge to the present.

The more interesting story is linguistic, not conspiratorial.

Trump was, in effect, trying to restore to an Americanized word its Chinese past. He was sending gung-ho home.

But words do not return from history unchanged. They carry their journeys with them. Inside gung-ho now coexist the Chinese cooperatives of the 1930s, Carlson’s Marines, the mythology of the Pacific War, Hollywood heroics, and the modern English suspicion that zeal may outrun judgment.

That is why the phrase is more revealing than it first appears. A Chinese listener may still hear “cooperation.” An American listener may hear “enthusiasm,” or perhaps “too much enthusiasm.” The same two syllables now contain two histories.

At the White House on September 24, those histories briefly met again.

Trump said:Gung ho.

Then he translated it back:Working together.

After more than eighty years in English, the phrase had come home—but not empty-handed.