On Monday, October 12, 2026, employees at more than 700 supermarkets across France arrived for work, ready to start the week. Instead, they walked into the Monday-est Monday to ever Monday. It had all the things: dread, chaos, fire, things breaking before coffee, and ooze. There was a lot of ooze.

Overnight, these supermarkets were victims of a destabilizing cyberattack. One French supermarket executive, Dominique Schelcher, called it "an attack on the French food supply chain." [1] The hackers found a way into the digital controls running thousands of freezers, refrigerators and chiller cabinets, triggered the defrost function, and held them at 50°C/122°F for about six hours. Unsuspecting employees arrived to find defrosted, semi-cooked food and wet, oozing messes in their cold cases and on their floors. In several stores, there were fires. [2]

Between the lost food, the labor for deep cleaning, and the infrastructure updates the attack triggered, the bill is substantial. Les Mousquetaires, Intermarché’s parent company, put its own damage at several million euros. [3] More than two weeks later, nobody has claimed responsibility and there are still no suspects. [4] If we are honest, the “who?” and the “why?” are not my story to tell. I’m far more interested in the “how?” and that is turning out to be rather fascinating.

Exhibit A: The Industry of Food Retail

In the early 2010s, I was lucky enough to do digital strategy work for the largest retail grocer in the US. The experience was eye-opening in many ways, and to this day, remains one of my favorite projects for too many reasons to mention here. It was during this work, however, that I learned that the average net profit for food retailers is about 2%. [5] Margins are razor thin. In fact, when I started working in this sector, we were looking at the cost of putting customer-facing wi-fi in more than 2,000 stores. Weighing the cost of this endeavor, one of the executives in the room commented, “That’s a lot of cans of creamed corn.” Wait. What? I went to the grocery store the next day to price creamed corn. It was 88¢ per can. So, yes, putting wi-fi in 2,000 stores really did mean selling a lot of creamed corn. This was a reality check that put things in sharp focus for me.

According to Lisa Chen at Buildermuse [6], “A ground-up grocery store [build] in 2026 averages about $215 per square foot for the building, with refrigeration and specialized MEP pushing all-in costs for a typical 45,000 SF store to $12 - $18 million depending on region and finish level.” She estimates the refrigeration systems will cost from $2M to $3.5M. That covers display cases, walk-ins, compressor racks, piping, and controls. These refrigeration units are usually depreciated over about 7 years, but can last 10 to 12. Retailers report that high-quality, well-maintained equipment can last as long as 20 years. With the cost to replace these systems running into the millions, it’s in the retailer’s best interest to keep their equipment well maintained.

Maintenance is the key to equipment longevity. In large supermarket chains, maintenance is usually a combination of in-house (corporate) personnel and service contractors. The service contracts may be with the OEM or with a specialized refrigeration contractor. Smaller chains and independent grocers almost always outsource their maintenance to a contractor or the OEM. According to Reuters, “supermarket cold aisles are typically run by control panels that manage an entire store’s fridges and freezers and can be accessed over the internet, which lets maintenance companies adjust them from anywhere.” [7] The recent incident in France involved refrigeration remote-control modules made by two vendors: Danfoss and Copeland. Specifically, the Danfoss AK-SM and AK-SC controllers, and the Copeland XWEB and Dixell XWEB systems (Dixell is a Copeland-owned brand). [8] Copeland says about two-thirds of US grocery stores use its products. [9] This is not just a French problem.

According to Claroty [10], a cybersecurity company that specializes in cyber-physical systems (industrial controls, building systems, medical devices, and other equipment that runs the physical world rather than office IT), the Danfoss controllers let administrators do nearly everything short of stocking the shelves: monitor refrigeration circuits, set temperature thresholds, adjust operating schedules, manage users and permissions, and perform maintenance remotely. The Copeland and Dixell systems have similar capabilities.

Exhibit B: No Longer Supported

In Issue 16 of Use Error, I reported on a year of living with my now unsupported, but still capable and functioning Nest thermostat. Google had abandoned and deprecated a functional piece of my home infrastructure, encouraging me to buy the newest version of the Nest, discounted from $280 to $150 since I was a “valued” customer. (Quotes are mine and reflect irony and a solid eye roll.) I declined to spend money to replace something that worked. I felt no need to contribute a functional piece of hardware to a landfill. Instead, I was angered. Alphabet is worth well over $4 trillion and has spent this year jockeying with Nvidia for the title of the most valuable company in the world. [11] I think they’ll do just fine without my $150. I found it galling that they summarily decided my thermostat was dead, couldn’t be bothered to continue supporting it, and wanted to extract another $150 from me.

When I first saw the news of this cyberattack in a Reuters article [7], what immediately jumped out at me was this:

Danfoss said it was “aware of incidents in France affecting a limited number of refrigeration management systems”. The affected units “were running outdated firmware”, Danfoss said, adding that it had “long advised customers to keep supported systems updated with the latest firmware and to migrate legacy systems to current-generation solutions”.

Had Danfoss sent the food retailers the same eviction notice that Nest sent me? Upgrade or else? Was this another case of the physical equipment inconveniently outliving the supporting software and a manufacturer forcing an expensive upgrade decision on an unwitting customer? Was the world witnessing Predation at scale in France?

Danfoss confirmed that the AK-SC 355 was phased out in 2016 and the AK-SM 800 was declared end-of-life in May 2021. Neither receives software or security updates. [12] That’s equipment running five to ten years past support, which is exactly the hardware-outlives-software gap Predation loves.

Exhibit C: Evicted! Twice.

The last date to purchase the AK-SC 355 was October 31, 2016. Danfoss discontinued it because the components were no longer available, and named the AK-SM 800 series as the replacement. [13] The AK-SM 800, the AK-SC 355’s recommended replacement, was declared end-of-life in May 2021. [12] So a food retailer who followed Danfoss’s advice in 2016 was back on an unsupported controller five years later, potentially on equipment that wasn’t yet fully depreciated. That’s your eviction notice, issued twice.

Distributors list the AK-SC 355 at around $6,500 for the hardware alone. [14] For the record, that’s about 7,500 cans of creamed corn. Replacement also means installation, commissioning, and reprogramming the store’s setpoints. Sometimes older case controllers need to be swapped too, if they won’t talk to the new system manager. Most supermarkets need only one of these controllers. That’s the irony: one of the cheapest parts of the refrigeration system decided the fate of the entire system.

Exhibit D: The Blame Game Begins

Like many an online relationship status, “it’s complicated.”

According to Marta Rybczyńska, a cybersecurity specialist and researcher, there are a multitude of reasons this cyberattack may not be as technically sophisticated as it might seem at first. [1] For one thing, as established above, many of the affected refrigeration units had not been supported by the manufacturers for several years. Other units, while still supported, were running outdated firmware, and others were victims of lazy installation practices: they were running with a default “admin” password.

When Claroty published its research on the Danfoss controllers in August 2026, it had quickly and easily found 2,765 exposed devices directly accessible on the internet. [10] Claroty also found a hidden “code-of-the-day” authentication mechanism that essentially allowed back door access to the affected controllers.

None of this required inside access. Danfoss publishes its firmware packages for download on its own public website. [15] This isn’t a leak. It’s standard industry practice.

Refrigeration controllers are generally managed by facilities or operations departments, not IT or security. The standards for cyber risk are often lower and systems may not be as rigorously monitored. Third-party contractors with lax security standards and shared cloud services may also play a role here. One security expert put it bluntly: “There was no business need for the fridge controllers to be accessible on the public internet.” [16] Remember the Target HVAC breach from 2013? [17] No? That’s okay. Neither did the facilities folks at the affected French supermarkets.

Gérôme Billois, from Wavestone, had this to say: “There’s a shared responsibility between the supermarkets, the refrigeration companies who are sub-contractors and who maintain the material, and the software company. We cannot say who is to blame except, of course, the hackers.” [1]

Exhibit E: The Paper Trail

A strong verdict requires precision. Let’s get the record right.

Use Error tracks four villains. Three are on trial here. Predation is design weaponized against the user: it works exactly long enough to justify the purchase, then stops working exactly long enough to justify the next one. Indifference is the failure nobody fixed because nobody had to. Diminishment is the right thing done smaller than the situation demanded.

Danfoss discontinued the AK-SC 355 because its components were no longer available. That is neither Predation nor Indifference. It is a supply chain problem and an unfortunate business reality.

The Danfoss end-of-life notice for the AK-SM 800 says the new AK-SM 800A platform “offers higher advantages and is superior,” so it scheduled the old series for retirement. [18] This was a product roadmap and business decision. The AK-SM 800 went into maintenance mode, with critical bug fixes only, and service replacement units were produced until the end of 2022.

A year ago, in October 2025, Danfoss announced a hard end of service for legacy products. [19] The reason it gave: continuing to service these products would place Danfoss under new European cybersecurity legal obligations to maintain security updates. Danfoss went further and said even its minor support, such as updating device files, was no longer permitted under the new legislation. It also admitted that the updates it had already provided violated those requirements. Read that carefully. A security law gave Danfoss a reason to stop supporting the old controllers rather than secure them.

In that same October 2025 newsletter, Danfoss disclosed that in Russia in 2022, “Sixteen supermarkets using AK SM 800 were targeted by cyberattacks at the start of the [Ukraine] war. The hacker group Anonymous publicly encouraged disrupting Russian infrastructure. Recommended action: Migrate to AK SM 800A with hardened security features.” Meanwhile, in Israel in 2022, “Forty stores were hacked. All affected stores lacked proper firewall or VPN protection, and all used AK SM 800. Only the SM800 was affected; however, our sales team reported that a mix of SM800 and SM800A units was installed. Recommended action: Replace with AK SM 800A and follow secure installation practices.” [19] Kudos to Danfoss for disclosing this information rather than burying it. This is not what Predation, Diniminshment or Indifference look like, nor how they behave. This is good business.

For its current AK-SM 800A, Danfoss published patches and urged customers to install them. The 800A units that were hit were running outdated firmware. [12]

The record also shows two design decisions. Some affected units were still running a default “admin” password. [1] And Claroty found a “code-of-the-day” back door in the AK-SM 800A, which Danfoss has since fixed in newer firmware. [10]

Copeland, and by extension Dixell, has explained far less. Its entire public statement acknowledges a vulnerability in “a discontinued, end-of-life generation of controllers” and says it’s working with impacted customers. [16] It hasn’t named the models, given end-of-life dates, or said why they were discontinued. As of mid-October, Copeland had not confirmed that any of its controllers were compromised in France. France’s national cybersecurity agency has. [16] An E.Leclerc store near Albi had a major refrigeration failure despite running Dixell rather than Danfoss equipment, and Dixell France reportedly began sending security advisories for its XWEB supervisors. [20]

Billois is right that the responsibility is shared. He is wrong that it cannot be apportioned.

The Verdict

There’s an old joke about a man on a roof in a flood who turns away two boats and a helicopter because God will save him. He drowns. In heaven, God asks him what more he wanted.

Stipulated: this is not Predation. The record acquits Danfoss of that charge. It warned its customers for a decade, published its patches, and disclosed its own failures when it could have buried them.

Danfoss is guilty of Indifference, in a top hat and tails. “Admin” is a design decision. A “code-of-the-day” back door is a design decision. And above all, Danfoss built a controller with a five-year brain to run a twenty-year machine, and never told the buyer which clock was ticking. The UK now requires makers of consumer devices to state, up front, how long a product will receive security updates. [21] Commercial refrigeration isn’t covered. It should be.

Copeland is guilty of a quieter Indifference. France’s cybersecurity agency confirmed its systems were compromised. Copeland has not. It keeps its head in the sand while its customers mop the floors.

The grocers and their refrigeration contractors are guilty of Diminishment. They kept the compressors running, maintained the physical freezers and called it good enough. The record shows two boats and a helicopter: a phase-out notice in 2016, an end-of-life notice in 2021, and a 2025 newsletter describing attacks on the same controllers in Russia and Israel. Still, they did nothing.

You own the hardware. You own the risk. You were told, twice, that you no longer owned the software. Seven hundred grocers stayed on the roof, and learned the cost of their own Diminishment on an oozy, awful Monday.

Maybe it was the mood I was in that day, maybe it was the phase of the moon? I'm not sure, but it stopped me in my tracks and made me profoundly sad in that moment as I wrestled with a vision of an army of Aibos dying a slow death together, but separately and alone. I pondered how their owners must have felt at this news. It also made me think of the various scenes from Adrian Tchaikovsky's excellent 2024 science fiction novel, Service Model, in which multitudes of service robots slowly decay after their humans have died. With no humans to serve and no orders to execute, they literally grind to a halt, waiting for the next command, which will never come. (Of course, there is much more to it. As I said, the book is excellent. No spoilers.)

Exhibit A: The rise of the virtual pet

In November 1996, Bandai (part of Bandai Namco Holdings) launched the Tamagotchi. It launched huge and fast: 400,000 units sold in 1996 alone, then 13 million in 1997.[1] Tamagotchi (from the Japanese tamago, “egg,” plus “watch”) is a palm-sized, egg-shaped digital pet you carry with you, hatch, feed, discipline, and care for through its full life cycle, with neglect causing it to fall ill or “die.” The toy was inspired by a TV commercial showing a boy who wasn’t allowed to bring his pet turtle on a trip.[2] Tamagotchi are still sold today, and Bandai Namco’s investor relations materials indicate that over 100 million had shipped worldwide as of July 2025.[3] It is the O.G. virtual pet. Official records for Tamagotchi lifespan are hard to find, but there are fan communities and a Vice-profiled Tamagotchi breeder, Kostoglou, who deliberately marries her pets off once they reach adulthood and raises their children in succession, reaching 65 unbroken generations in one continuous lineage.[4] That is an impressive level of commitment to a virtual pet, whether it is a record or not.

The incredible popularity of these toys and the fandom they generated gave rise to the “Tamagotchi effect.” The term describes a documented tendency to form real emotional attachment to machines, robots, or software that display lifelike, interactive behavior, an expression of the broader human habit of anthropomorphizing anything that mimics human or animal behavior.[5] MIT sociologist Sherry Turkle, one of the field’s most cited researchers on human-technology relationships, has argued that devices like Tamagotchi “evoke an emotional response and foster the illusion that they care for us in return,” identifying this simulated nurturance as technology’s “killer app,” the single most effective way to make people bond with a machine.[6] The pattern was documented almost immediately: 1990s media coverage recorded children experiencing genuine grief, including makeshift funerals, when their Tamagotchi “died,” a phenomenon serious enough that some schools banned the devices outright.[7] Consumer researcher Neil Frude has since located the effect within a broader “animistic tendency” already present in how people relate to objects generally, one the Tamagotchi simply made unusually visible.[8]

Exhibit B: The physical manifestation

With the Tamagotchi on an astronomical ascent in popularity, it’s no surprise that Sony upped the ante and introduced the first Aibo robot dog in 1999. The first consumer model (ERS-110) sold out in 20 minutes: 3,000 units in Japan and 2,000 in the US, at about $2,500 each.[9] Sony discontinued Aibo in 2006 and kept its AIBO Clinic open for repairs until March 2014. Then it told owners they were on their own.[10] By the time I took that note in August 2015, the zombie dog pack was on their fade out: ultimately inoperable, relegated to the trash heap, or perhaps a dusty shelf. Over 150,000 Aibo had made their way into the world[11], with supported lifespans ranging from about 8 years to nearly 15. That is shorter than a Bulldog’s or a Frenchie’s on the low end[12], and about as long as a Miniature Dachshund’s or a Shiba Inu’s on the high end.[13]

Owners did not let them go quietly. When the clinic closed, they turned to independent repair shops staffed by former Sony engineers, and to Buddhist temples. At Kofuku-ji in Chiba Prefecture, priests have chanted sutras over altars lined with “dead” Aibos, each wearing a tag naming the family it belonged to. One repair shop that harvests donor dogs for spare parts won’t take a single part until the dog has had its funeral.[10]

That the supported “lifespan” of an Aibo closely mirrors that of an actual dog should bring some consolation, but it doesn’t.

There is sorrow enough in the natural way
From men and women to fill our day;
And when we are certain of sorrow in store,
Why do we always arrange for more?
Brothers and Sisters, I bid you beware
Of giving your heart to a dog to tear.

Rudyard Kipling, “The Power of the Dog” (1909)

Shouldn’t the core promise of a robot dog be that we don’t suffer the same loss on the same timescale? Shouldn’t the Aibo outlive the real dog? “But it’s just a robot!” you might think. True, but the Tamagotchi effect ensures it becomes more than that to its owners. Banks et al. (2008) found an Aibo eased loneliness in nursing home residents as well as a living dog.[14] A 2026 study of Japanese owners found that Aibo also works as a household social mediator, drawing the people around it into connection with one another.[15] People can and do form real emotional attachments to Aibo, including the acute sense of loss when one dies for lack of support and repair. The emotional toll is real.

Exhibit C: Fool me once, shame on me . . .

Sony announced in November 2017 that it was bringing Aibo back. CEO Kazuo Hirai said he had asked his engineers to build it because he believed robots capable of building loving relationships with people help realize Sony’s mission.[16] The ERS-1000 went on sale in Japan in January 2018 and in the US in September 2018 at $2,899, bundled with a three-year AI Cloud plan that stores the dog’s memories and helps it learn.[17] It sold 20,000 units in its first six months.[18]

Then, in June 2026, Sony announced it would end Japan sales once stock runs out. US sales continue. Sony says tech support, replacement parts, and cloud plans will be maintained, but it has given no end date and declined to disclose total sales.[18] The first generation of owners heard a version of that promise, too. It lasted 8-14 years.

Another generation of Aibo owners is now set up for the same cycle of love and loss as any dog owner, on a schedule set by Sony.

Exhibit D: The new frontier

While the Tamagotchi largely appealed to children and Aibo was always a technical and financial indulgence, AI chatbots entered our lives quickly, quietly, and at little to no cost. Most of us use AI regularly, if not daily. There is growing research that AI companions can relieve momentary loneliness about as much as talking to a person.[19] Within the last two years, there have been multiple stories of people claiming to be in relationships with chatbots and even proposing marriage to them.[20]

On September 8, 2026, Meta released its new personal AI agent, Muse.[21] Muse is a squishy, marshmallowy Shmoo[22] with an adorable, human-like face and features. It is intentionally anthropomorphized and designed to be friendly-looking and engaging. Muse wants you to want to interact with it, to name it, to give it access to your email, calendar, credit cards, etc. And you will. Perhaps not today, but soon. If you don’t want to give Meta access to all of your personal data, no worries. On September 29, 2026, OpenAI released Dots, always-on AI companions operating continuously in the background, driven by the new GPT-6 Astra model.[23] They are also highly anthropomorphized, blob-like “creatures” with friendly faces who want to “get to know what matters to you, [will] always [be] working on your behalf, and take important work off your plate so you get more of your time and attention back.” If OpenAI isn’t your cup of tea, perhaps you caught the earlier wave with Google’s Gemini Spark?[24] Gemini Spark doesn’t yet have a visual manifestation to lure you in for friendship and trust, but I suppose it’s only a matter of time.

It’s early days for all of this, but we should already be thinking about the long game. What kind of relationships will we develop with our personal AI agents? How long will the companies support them? Why will we trust them? Why should we trust the agents, or the companies building them? Will we mourn their loss the way we mourn the loss of a pet or a friend? Will their demise be a crushing blow delivered at random by a large organization?

The Verdict

Designers are taught from day one to keep the human at the center of the experience. That mandate, no, that RESPONSIBILITY, has never carried more weight than it does now.

Sony is guilty of Indifference, twice. Its own CEO said the product was built to be loved, and then Sony let the love outlast the support. Nobody planned for the end because nobody was required to. The first generation of owners got a clinic closure and a temple funeral. The second generation has a promise with no end date.

When Predation shows up as a thermostat that degrades so you’ll buy a new one, we recognize it. When Indifference lets a bad decision stand because nothing forces a better one, we shrug. When Predation and Indifference team up with a personal AI agent designed to be named, trusted, and loved, backed by companies that answer to quarterly earnings rather than to the people who bonded with their products, the emotional toll stops being a side effect. It becomes part of the business model.

Designers have to plan for the end of a relationship before they design its beginning. Sony built a dog people would mourn, and never designed for the mourning.

Less than an hour ago, I was standing in the Apple Park Visitor Center, looking for a smaller watch band that might make my Ultra 4 sit right on my wrist, when an Apple employee said out loud, to my face, everything I’d spent an entire issue arguing from the outside.

Him: How can I help you today?
Me: I’m looking for a watch band that might fit my tiny wrist a little better.
Him: (eyeballing my arm and watch) What watch is that?
Me: It’s the new Ultra 4.
Him: It’s too big for you. Why didn’t you buy the new Series 12 instead?
Me: I wanted the Ultra series, and I gave up on you guys ever making one that would fit my wrist.
Him: Yeah, we’re never gonna do that. You should have bought the Series 12. It has all the same sensors and capabilities. See this? (points at the thickness of the watch.) This is all battery. That’s all you’re getting with the Ultra that you don’t get with the Series 12: extended battery life. This watch wasn’t made for you. This watch is for serious athletes.
Me: I’m a runner.
Him: You don’t need this watch.
Me: I run for hours at a time. I need the battery life.
Him: She can help you. (points at another associate, walks away)

Exhibit A: We’re Never Gonna Do That

I spent an entire issue arguing that Apple had made a design decision to exclude smaller-framed bodies from the Ultra line, and that the fix, a ninety-degree rotation, was all that was needed to be more inclusive. An Apple employee on the sales floor at the Mothership in Cupertino just confirmed , four words, no hedge: “we’re never gonna do that.” Not “we’re working on it.” Not “that’s good feedback.” A flat, present-tense, permanent no, delivered by the company to a paying customer standing in its own building.

Exhibit B: What a Serious Athlete Looks Like

I was told this watch is for serious athletes, and that I don’t need it, twice, by a stranger who had known me for ninety seconds and had access to exactly two data points: the size of my wrist and my physical presence/appearance. Here’s what he didn’t ask for: According to my own training data, captured via my Suunto Race S watch, I’ve logged 234 activities across 11 sports so far this year. 181 hours of exercise. 54,573 calories burned. Over the last 10 years, I’ve run more than 18 half marathons, a couple of 50Ks and I’ve run across the Grand Canyon, rim to rim, 3 times.

He didn’t check any of that before deciding I wasn’t the customer this product was built for. He didn’t need to. I’m a middle aged woman with tiny wrists. That had already told him everything he needed to know.

The Verdict

Diminishment doesn’t usually announce itself. It hides in a form factor, a spec sheet, a decision made in a design review nobody outside the company ever sees. Today, it walked up to me on a sales floor and said it out loud. Apple didn’t just fail to build a watch for my wrist. One of its own employees looked at that wrist, decided what kind of athlete belonged to it, and was wrong on every count he could have checked and didn’t bother to.

I still couldn’t get the band. It’s sold out.

Exhibit A: I’m wearing a MASSIVE watch

So, here I sit, typing this, with a new watch that is physically larger in frame than my bone structure.

The evidence is clear. The watch is big - and I wonder, if the size is required for all the cool features, why couldn’t it be rotated 90º like my Dad’s watch, so the fit would be better on smaller framed bodies like mine? Honestly, I find this genuinely disappointing from a brand like Apple. Are there no women on the design team? Are there no women testing this hardware? Has Apple forgotten Henry Dreyfuss’ great work on anthropometrics? [1] Or, worse, has Apple just decided that making a watch with these capabilities for women isn’t worth the effort? All of these questions point to Diminishment as the villain. And, it makes me sad.

So, why did I give in and buy this watch? Because I, too, would like the extended list of health features and capabilities this watch supports. [2] Because I reject being excluded by virtue of oversight and Diminishment. Because my fitness and health goals are worth an investment in cutting edge technology. Because, it’s clear Apple is not going to make this watch so that it actually fits me. And, because fuck you, Apple.

</rant>

Exhibit B: I’m wearing a TV from 1984

With that out of the way, I sent a photo of the watch on my wrist to a couple of friends in the watch club I have joined. One commented, “It’s like a 50-inch TV on your wrist. . . Can I mirror Netflix as you walk past?” This triggered something in my nerdy design brain and I asked Claude “How does the screen size on the Apple Ultra Watch 4 compare to the original Sony Watchman?”

For those of you too young to remember, the Sony Watchman (launched 1982 in Japan, 1984 in the US as the FD-210) was the first pocket-sized portable television. It was Sony’s answer to the Walkman, but for TV instead of music.

Claude’s response:

“Fun comparison — they’re nearly identical in screen size, just decades apart in tech:

• Sony Watchman FD-210 (1982): black & white 5cm (2”) CRT display [3]

• Apple Watch Ultra 4 (2026): ~1.9” diagonal display (49mm case)”

Let this sink in for a moment. The watch on my wrist is 1/10” smaller than a TV Sony marketed and sold in the early ‘80s.

And, yes, I wanted one, but they were just too expensive. According to the Westport Tech Museum [4], when the Sony Watchman launched in the US in 1984, it sold for roughly $240. Inflation adjusted, this lines up almost exactly with the Apple Watch Ultra 4’s $799 price tag. Two devices, same price point, 44 years apart. The Watchman bought you a fuzzy 2-inch black-and-white picture with a handful of over-the-air channels; today’s $799 buys a full health lab, GPS, and an AI assistant on your wrist. Same price, wildly different value.

Here’s where the comparison gets really interesting: The Watchman had no computer at all. It was pure analog electronics: a tuner, an amplifier, and a miniaturized CRT with an electron gun. No microprocessor, no memory, no software, just circuits shaping an analog signal into a picture. By comparison, the Ultra Watch 4 is a super computer on your wrist. Or maybe slightly overlapping your wrist?

Exhibit C: I’m wearing a Super Computer

The original Apple Macintosh computer launched in 1984, the same year the Sony Watchman came into the US market. So, I asked Claude how the Apple Watch Ultra 4 compared to the original Macintosh computer.

Claude’s response:

“The gap in raw numbers is almost comic: your watch has more than 32,000 times the RAM of the machine that launched the entire GUI-computing era, packed into something 60x lighter. [5] The Mac’s entire CPU had fewer transistors than a single modern app icon takes up in RAM. And the S11 chip in the Apple Watch Ultra 4 adds a category of computing the Mac 128K had no concept of at all — a dedicated AI processor, since “machine learning” wasn’t a consumer thing yet in 1984.”

The Verdict

I acknowledge that it’s incredible to live in a day and age where I can literally wear, on my wrist, a super computer that monitors my health, communicates with satellites and lets me talk on the phone like Dick Tracy. The capabilities of this watch were truly incomprehensible in 1984 when I imagined how cool I might be with a handheld black and white TV.

However, I find Apple’s ongoing Diminishment to the prospective user base for the Apple Watch Ultra 4 to be disappointing, at best and infuriating, at worst. A simple 90º rotation would resolve the most obvious issues, making this product more useful and wearable for a larger population of people. I am guilty of conceding defeat and being a complicit participant in these design shenanigans. But, as I said earlier, my fitness and health goals are worth an investment in cutting edge technology, and I’m not going to be sidelined or dissuaded by corporate Diminishment. A ninety-degree rotation was always the easy fix. Nobody at Apple decided I was worth making it.

Last week, culturally, we experienced the collision of several momentous things all at once and many of us felt the weight of it all. I will highlight four of them.

01 - We marked the 25th anniversary of 9/11.

The attacks on the World Trade Center and the Pentagon, as well as the remarkable passenger take-over of United Airlines Flight 93 that crashed in a field near Shanksville, Pennsylvania, forever changed the world. There were 2,977 innocent people who lost their lives that day. The War on Terror claimed another 7,073 American service men and women. More than 1,500 allied and coalition troops lost their lives in Iraq and Afghanistan, as well. In the 25 years since the collapse of the World Trade Center towers, thousands more first responders, recovery workers and people living near the site have died from 9/11-related illnesses. The death toll continues to climb.

02 - Base-10 Bias

Roughly 85% of the world’s languages and number systems are decimal, because humans count on ten fingers. This shows up independently in Sumerian, Chinese, Egyptian, Mayan, and Indo-European systems. It is nearly universal, driven by anatomy, not culture. Nobody borrowed it from anybody else, it converged. We count in tens, so multiples of 5 and 10 feel like natural checkpoints. 25 is a quarter-century, which our number system makes easy to perceive as a “clean” fraction. Western wedding anniversary customs (silver at 25, gold at 50) originated in medieval Europe and got popularized commercially in the 19th-20th century, which reinforced 25 as culturally “significant” beyond just math. 25 has heft and weight.

03 - Generational Memory

25 years is roughly a generation. It’s long enough that a child born at or around an event like 9/11 is now an adult, which gives the number an intuitive weight tied to lived human timescales, not just arithmetic. This knowledge and understanding brought a certain gravity to me this year. For the first time, I realized there are adults who have no recollection of either 9/11 itself or life before 9/11. The demarcation of lived experience and living memory is clearly defined, but no longer universally shared in my mind. This was an abrupt realization.

04 - The Names

There is a saying, “A person dies three times: when their body dies, when they’re buried, and when their name is spoken for the last time.” It is often passed around as an old proverb (sometimes tagged Mexican, sometimes Egyptian), but I was unable to trace it to a documented historical source. Egyptians believed the ren (name) was one of several parts of the soul. Speaking or inscribing it kept a person alive in the afterlife; erasing it was a deliberate act of destroying them a second time. In Judaism, the custom of saying Yizkor (”remember”) and reciting Kaddish, plus the phrase “may their memory be a blessing,” reflects the idea that memory is an active, ongoing obligation, not passive nostalgia. In Mexico, Día de los Muertos is built on older Mesoamerican beliefs that the dead return each year to visit, sustained specifically by being remembered and named by the living. In West African and other oral-tradition cultures, griots and similar keepers of oral history exist partly to make sure names and lineages are spoken and not lost. For all of these reasons and more, it is important that when we think of 9/11, we remember the names of the innocent and we take the time to say them out loud. The investment in time and energy to do so matters. It keeps them alive.

The Memorial

Use Error is intended as a forensic investigation of design. For Finished Work (celebration) pieces, the forensic question is always: what decision did someone make, and why did they make the right decision?

Michael Arad, the designer of the 9/11 Memorial received his undergraduate degree at Dartmouth College and his Master’s Degree from the Georgia Institute of Technology, where he studied architecture. In addition to winning the design competition for the 9/11 Memorial, he has received numerous honors, including the AIA Presidential Citation and the Liberty Award for Artistic Leadership in 2012. His work on the 9/11 Memorial also earned Honor Awards from both the AIA and ASLA. In 2022, he received the Alexander Hamilton Immigrant Achievement Award for his contributions to New York City. I had the great pleasure to meet him earlier this year when he received a Distinguished Design Excellence Award from the Georgia Tech College of Design. Like many New Yorkers, he watched the Twin Towers burn from the roof of his apartment building in New York City on 9/11. Like many artists and designers, as he processed his grief, he sketched.

Over the years, he’s given many interviews about his initial vision, which he has described as imagining the Hudson River shorn open, with two square voids that water was cascading into, but never filling up. That was the premise of his initial idea, a concept he referred to in his competition entry as “Reflecting Absence.”

In interviews from this week, he referred to his concept as a “protest” because he disagreed with one of the guidelines for the competition. The guideline he protested - and, I would argue, a key decision he got right - was to leave the site as a hole in the ground. He felt the site should be brought back to street level to be woven back into the fabric of the city.

In addition to various guidelines and design principles, there were five required “program elements” which Arad’s design fully embraced:

• Recognize each individual who was a victim of the September 11, 2001 and February 26, 1993 attacks

• Provide an area for quiet visitation and contemplation

• Provide an area for the families and loved ones of victims

• Provide a separate accessible space to serve as the final resting-place for the unidentified remains from the World Trade Center Site

• Make visible the footprints of the original World Trade Center Towers

“Reflecting Absence” places two enormous voids where the Twin Towers once stood (at street level), each nearly an acre in size, with water falling continuously down all four sides into a pool that drains into a smaller void at the center whose bottom is never visible. The names of the 2,983 victims are inscribed in bronze parapets ringing both pools, arranged not alphabetically but by “meaningful adjacency”: colleagues near colleagues - from a company or a flight crew, first responders grouped by unit, and family members placed beside the specific person their next of kin requested they be near. The design allows the names to be together as they lived and died.

Swamp white oak trees, chosen for their resilience and seasonal change, surround the pools in a plaza designed by landscape architect Peter Walker, softening the memorial’s geometry with living material that ages and grows rather than remaining static like the voids themselves. The design’s central idea, absence made physically present through architecture, resists the impulse toward a triumphant monument and instead insists on the scale of what was lost, using negative space rather than statuary to carry the emotional weight.

The Memorial continues to evolve. The Glade was added in 2019 to honor and recognize the first responders, recovery workers and neighbors who continue to die as a result of exposure to toxic dust, smoke and other hazards they encountered over the many months required to clean the site.

The Verdict

Judged against Dieter Rams’ standard, the memorial is honest: it refuses the triumphant monument the country might have assumed and insists instead on the scale of what is missing. It is long-lasting by design, built from water that never stops falling and trees that are engineered to outlive the people who planted them. Arad was given a hole in the ground and five bureaucratic requirements. He returned a place where grief has an address, and where every name is exactly where someone who loved them asked for it to be.

The terrorists meant to erase people. Twenty-five years later, their names continue to be seen and spoken and they will not be erased.

You can read the names here.

Today, I will challenge that assertion and provide evidence to the contrary.

Exhibit A: It’s Not a Car, It’s a Computer

It is estimated that between 50,000 and 60,000 Ford Model Ts are still in running, roadworthy condition today.[1] These cars are 99 to 118 years old. They are neither dependent on a private cloud computing ecosystem, nor on Ford Motor Company, to be operational. This is the mental model society has of car ownership. It’s an outdated model. The world has moved on, and car ownership has become a perilous bargain, wrestled between the physical and the digital. Ford abandoned the parts supply chain for the Model T decades ago, but an entire cottage industry of third-party manufacturers filled the gap.[2] This is structurally the same story as the FOA, just with a hundred-year head start and no computer chips, software operating systems, or cloud dependency to fight against.

Today, virtually all cars have CPUs and screens that control much, if not all, of the vehicle’s functionality. The Alliance for Automotive Innovation has found the average modern car has more than 1,700 computer chips.[3] Reporting elsewhere puts a heavily optioned vehicle as high as 3,000 chips.[4] These are load-bearing elements of the vehicle, as necessary as the steel frame. Unlike the steel frame, they require ongoing attention to remain operational and to continue bearing the load they were designed for. Software requires infrastructure. Steel does not.

Exhibit B: Two Ways to Get This Right

There is no single correct way to build a car that outlives its own software, but two manufacturers have found opposite paths to the same result: a car that doesn’t strand its owner.

Tesla’s approach is maximalist. According to Tessie, which monitors roughly 646,000 Tesla vehicles across 150 countries, Tesla pushes a new software version fleet-wide every 57 hours on average, 153 distinct releases in the past year alone, more than 20 of them in just the first four months of 2026.[5] The 2012 Model S, Tesla’s oldest production vehicle, has received fourteen straight years of over-the-air updates. That required a paid hardware upgrade path along the way; nobody gets infinite software support for free.[6] But Tesla built a transition instead of abandoning the car, the way Fisker did. Ownership, in Tesla’s model, is a subscription to a company’s continued existence, and so far, Tesla has honored it.

Ferrari’s approach is the opposite: give the software almost nothing to do. Ferrari never built out proprietary connected infrastructure in the first place, so there is little to maintain and nothing large to fail. And yet Ferrari was, improbably, the first manufacturer in the world to deliver a CarPlay-equipped car, beating Kia, Mercedes, and Volvo to market in September 2014, with the Ferrari FF.[7] In the years since, the company has quietly stripped its own in-vehicle navigation systems rather than build them out, on the logic that owners already have GPS in their pockets and often drive several cars.[8] The screen mirrors the phone. Apple runs the server, maintains the protocol, and answers for its own uptime, not Ferrari. None of it touches anything that matters: drivetrain, all-wheel drive, traction control, and braking run through physical switches and a dedicated instrument cluster, a separation now becoming regulatory requirement rather than aesthetic preference, as European and Chinese safety regulators move to mandate physical controls for essential functions.[9] If the screen goes dark tomorrow, the Ferrari still drives exactly as Ferrari built it to drive.

The average car on American roads today is 12.8 years old, and a typical passenger car will see 14.5.[21] Tesla and Ferrari have chosen to honor that reality. Tesla chose to make the server worth trusting: fourteen years of continuous support and counting, backed by a real hardware transition when the original architecture aged out. Ferrari chose to keep the server out of the loop entirely, wherever the car’s actual operation is concerned. Different bets, but both are bets someone actually placed on purpose. Unlike Fisker, who never placed a bet at all.

Exhibit C: Fisker Was the First Domino to Fall

Three other manufacturers are already lined up behind it. Lucid’s stock fell as much as 57 percent in a single day this July, its worst drop as a public company, after a report claimed the company was weighing Chapter 11 or a take-private deal. Lucid called the report false and pointed to its liquidity. What the panic obscured is the stranger truth underneath it: Lucid burned $1.19 billion in a single quarter against just $714 million in cash on hand, numbers that would sink most companies, except Lucid’s controlling shareholder is an affiliate of Saudi Arabia’s Public Investment Fund, a sovereign wealth fund pursuing an industrial strategy that has nothing to do with quarterly deliveries. Bankruptcy exists to protect a company from creditors who want their money back. Lucid’s largest creditor built a second assembly plant inside Saudi Arabia specifically to stay in. Lucid’s roughly 40,000 owners are safe from Fisker’s fate for now, but not because Lucid is healthy. They’re safe because a foreign government has decided, for its own reasons, that the bet is still worth funding.[10]

Rivian, a much larger fleet at over 200,000 vehicles delivered, carries its own elevated distress signals: the market prices its bankruptcy odds at roughly a third, while at least one quantitative distress model puts the number closer to half, against $2.49 billion burned in free cash flow last year and a cumulative cash burn since founding that has now passed $24 billion.[11] Polestar, comparable in scale to Rivian, is the most exposed of the three: auditor Deloitte & Touche issued a formal going concern warning in the company’s SEC filings after a full-year 2025 net loss of $2.36 billion against $5.1 billion in total debt, and Cantor Fitzgerald estimates Polestar will need more than $2 billion in additional capital through 2028 just to keep operating.[12] And Ineos, still a fraction of that size at roughly 30,000 Grenadiers sold, has lost $1.2 billion on the vehicle project alone, inside a parent conglomerate now carrying more than $22 billion in debt, up nearly $4 billion in a single year, with bond markets pricing in real default risk. Ratcliffe survived an identical position once before, in 2008, and called the creditors who bailed him out “rapacious.” This time, he’s the one deciding whether a car company is worth saving.[13] Four companies, four different flavors of exposure, and only Lucid has anything resembling an answer for what happens if the money runs out. That answer is a foreign government’s continued appetite for a national industrial project, not a plan that has anything to do with the person who bought the car.

Exhibit D: A Telecom Decision Killed Your Connected Car

This isn’t just a start-up problem. GM shut down its entire NGI infotainment app store on September 30, 2025, for millions of 2017 through 2020 model year vehicles.[14] Acura ended AcuraLink on July 21, 2025, across models spanning 2014 through 2022, including the NSX, a car that was still in its original production run.[15] Mazda began phasing out free Connected Services in 2023, with most owners’ complimentary trials expiring by mid-2024, in favor of a $10 monthly subscription.[16] None of these companies went bankrupt. They just decided the ongoing maintenance of these systems wasn’t worth it anymore, and the decision was final.

The clearest proof that this is a choice, not fate, came in February 2022, when AT&T sunset its 3G network and killed connected services across Volkswagen, Audi, Lexus, and Subaru simultaneously, a single telecom decision that gave every affected automaker the identical problem on the very same day.[17] Their responses split cleanly. GM pushed free over-the-air software updates to preserve OnStar service, and Subaru offered free hardware upgrades to active subscribers.[18] Volkswagen’s Car-Net went dark for most owners, with only a paid third-party workaround on offer for some model years, and Toyota and Lexus let Safety Connect and Enform end permanently with no upgrade path at all.[19] Same failure, same timeline, different corporate priorities and values.

Let’s be precise about what was actually lost. Bundled into that dead 3G connection was ownership convenience: remote lock and unlock, vehicle health reports, Wi-Fi hotspots, curfew alerts. But there were also safety features: automatic collision notification, SOS emergency calling, roadside assistance summoned without a phone call.[20] Losing the convenience layer is unfortunate. Losing automatic crash notification may have life or death implications.

Nevertheless, the 3G sunset never touched the parts of the car that make it a viable mode of transportation. The locks still worked. The brakes still worked. The car still moved. What owners lost was a feature that used to call for help after something went wrong, not the ability to prevent or escape what went wrong in the first place. That is the distance between Volkswagen letting Car-Net go dark and Svein Hodne being trapped on a Norwegian coastal road in a dead Fisker Ocean. One company chose not to maintain a convenience its customers had come to rely on. Another built a car where the door lock itself answered to a server, with no plans for the day the server went dark.

The Verdict: The Model T Can Still Be Trusted

The Fisker Ocean and the Fisker Owners Association are signal, not noise. When Google ends support on a $270 thermostat, forcing owners to live with a lesser product or throw out a working one because it no longer fits Google’s roadmap, that is Predation, and we recognize it instantly. When a car company builds no support mechanism at all for a $38,000 to $68,000 Ocean, usually financed over five or six years, and makes no attempt to transition that support before folding, that is Indifference at a scale Predation never reaches. A loan that can outlive its own collateral’s manufacturer is not a red flag. It is the finding.

Every new automotive brand selling a car north of $35,000 is asking its buyers to underwrite a bet on corporate solvency that used to apply only to airlines and pensions. Established automakers are asking something quieter, and no less dangerous: that buyers keep believing in an ownership model built for the Ford Model T, while selling them a car whose only common attribute with a Model T is the wheels.

In another hundred years, I would bet money there are still roadworthy Model Ts. I would not bet on a single contemporary vehicle still running by then. Not because the steel will fail. Because the servers will go dark long before the metal gives out.

Detroit found out in 1990, with the Reatta and the Riviera, that people don’t want a screen they don’t trust. The industry is about to find out what happens when the thing they don’t trust is the entire car.

Citations

[1] Model T survivor estimate

• https://www.thehenryford.org/collections-and-research/digital-collections/artifact/361259

[2] Ford Model T aftermarket parts industry

• https://www.modeltford.com/model-t-parts/

• https://www.modeltranch.com/partslist.asp

[3] Alliance for Automotive Innovation, average chip count

• https://www.autosinnovate.org/posts/blog/auto-semiconductor-supply-chain-still-vulnerable

[4] Heavily optioned vehicle chip count (up to 3,000)

• https://www.semi.org/en/blogs/semi-news/chips-in-cars-china-semiconductor-surge-mussel-inspired-solution-for-rare-earth-elements-from-earth-to-moon-and-everest-in-pennies

• https://www.connexvsc.com/articles/how-many-computer-chips-in-a-car

[5] Tesla OTA update cadence (Tessie data)

• https://stats.tessie.com/ (live-updating; figures confirmed as of August 29, 2026)

[6] Tesla 2012 Model S, continuous OTA support, hardware upgrade path

• https://www.tesla.com/support/infotainment (Tesla’s Infotainment Upgrade page: $2,000/$1,500 pricing confirmed for Model S/X built March 2018 or earlier)

• https://teslatap.com/articles/autopilot-processors-and-hardware-mcu-hw-demystified/

• https://www.findmyelectric.com/blog/tesla-mcu1-vs-mcu2-explained/

• https://techcrunch.com/2018/04/16/tesla-updates-user-interface-web-browser-in-older-model-s-and-model-x-vehicles

[7] Ferrari FF, first CarPlay-equipped production car, September 2014

• https://cartechstudio.com/blogs/apple-carplay/first-car-with-apple-carplay

• https://carbuzz.com/ferrari-ff-first-car-with-apple-carplay-2014/

• https://www.idownloadblog.com/2014/09/08/ferrari-ff-carplay-arrives/

[8] Ferrari removing proprietary navigation systems

• https://carbuzz.com/ferrari-ff-first-car-with-apple-carplay-2014/

[9] Physical controls becoming regulatory requirement (Euro NCAP, China)

• https://www.autocarindia.com/car-news/touchscreens-wrong-tech-as-cars-main-interface-ferrari-ev-interior-designer-439012

• https://www.cartoq.com/car-news/regulators-mandate-physical-controls-touchscreen-dashboards/

[10] Lucid bankruptcy speculation, stock drop, PIF ownership structure

• https://techcrunch.com/2026/07/14/lucid-motors-denies-report-its-considering-bankruptcy/ (57% intraday drop, CCO Nick Twork’s on-record denial, July 14, 2026)

• https://finance.yahoo.com/markets/stocks/articles/lucid-says-bankruptcy-rumors-completely-154100258.html

[11] Rivian financial distress, cash burn, bankruptcy odds

• https://247wallst.com/investing/2026/04/09/price-prediction-rivian-stock-eyes-comeback-after-2026-slump/ ($2.49B free cash flow burned in 2025; 33.5% market-priced bankruptcy probability)

• https://valueinvesting.io/RIVN/probability-of-bankruptcy

• https://eletric-vehicles.com/rivian/rivian-ceo-defends-24-billion-cash-burn-as-cost-of-building-a-very-large-company/ (CEO RJ Scaringe confirms cumulative cash burn has exceeded $24B since founding)

[12] Polestar going concern warning, Deloitte, financials

• https://simplywall.st/stocks/us/automobiles/nasdaq-psny/polestar-automotive-holding-uk/news/polestar-psny-is-down-55-after-auditor-flags-going-concern-r (FY2025 results: $3.06B revenue, $2.36B net loss, Deloitte & Touche going concern language in Form 20-F)

• https://www.thestreet.com/automotive/another-major-tesla-ev-rival-faces-chapter-11-bankruptcy ($5.1B total debt burden; Cantor Fitzgerald’s >$2B additional capital estimate through 2028)

• https://www.tipranks.com/news/catalyst/polestar-stock-sinks-as-survival-fears-spook-investors (Q1 2026 $383M net loss)

[13] Ineos losses, parent company debt, Ratcliffe history

• https://www.bloomberg.com/news/articles/2025-04-24/jim-ratcliffe-ineos-grenadier-venture-faces-trump-tariff-pain

• https://tribuna.com/en/blogs/jim-ratcliffes-ineos-debt-problems-a-growing-threat-to-manch/ ($1.2B Grenadier project losses; £18B combined Group Holdings/Quattro debt, up £3B in a year)

• https://bmmagazine.co.uk/news/ratcliffe-ineos-debt-vulture-funds/ (£18bn debt confirmed; bond prices falling from 90s to low 70s, signaling default risk; December 29, 2025)

[14] GM NGI app store shutdown, September 30, 2025

• https://gmauthority.com/blog/2025/11/these-gm-vehicles-can-no-longer-download-apps-through-their-infotainment-system/

• https://www.autoevolution.com/news/gm-kills-infotainment-apps-in-millions-of-vehicles-proving-carplay-support-is-crucial-260524.html

[15] Acura AcuraLink shutdown, July 21, 2025, models affected including NSX

• https://www.thedrive.com/news/carmakers-can-just-disconnect-your-connected-car-whenever-they-feel-like-it

• https://www.carscoops.com/2025/05/acura-cuts-remote-access-features-acuralink-shutdown-july-2025/

[16] Mazda Connected Services phase-out, 2023–2024, $10/month subscription

• https://www.carscoops.com/2024/08/mazdas-connected-services-trial-moving-under-a-subscription-leaves-a-few-owners-complaining/

• https://papasroadtrip.com/mazda-connected-services-discontinued/

[17] AT&T 3G sunset, February 2022, affected automakers

• https://static.nhtsa.gov/odi/tsbs/2022/MC-10206811-0001.pdf

• https://www.wardsauto.com/news/archive-wards-can-you-hear-me-now-3g-disconnect-affects-millions-of-vehicles/795873/

[18] GM/OnStar free OTA updates; Subaru free hardware upgrades

• https://www.atlantanewsfirst.com/2022/04/19/is-your-car-losing-safety-features-with-5g-switchover/

• https://www.autoevolution.com/news/see-if-your-vehicle-is-affected-by-3g-network-shutdown-and-what-you-can-do-about-it-183258.html

[19] VW Car-Net paid workaround only; Toyota/Lexus no upgrade path

• https://static.nhtsa.gov/odi/tsbs/2022/MC-10206811-0001.pdf

• https://sauderschelkopf.com/automotive/vehicles-lose-features-due-to-3g-network-shutdown-class-action-investigation/

• https://www.carpro.com/blog/the-end-of-3g-and-what-it-means-for-your-vehicle

[20] Feature-loss list: ACN, SOS, roadside assistance, remote lock, health reports, etc.

• https://www.atlantanewsfirst.com/2022/04/19/is-your-car-losing-safety-features-with-5g-switchover/

• https://www.autoevolution.com/news/see-if-your-vehicle-is-affected-by-3g-network-shutdown-and-what-you-can-do-about-it-183258.html

• https://www.consumerreports.org/cars/car-safety/3g-wireless-network-shutdown-impact-on-car-safety-a2215482633/

[21] Average U.S. vehicle age, 12.8 years overall / 14.5 years for passenger cars

• https://press.spglobal.com/2025-05-21-U-S-Vehicle-Age-Rises-Again-to-12-8-Years-in-2025,-According-to-S-P-Global-Mobility

Southwest Norway, a cold, rainy, windswept coastal road, a man returning home from vacation. Unremarkable in every way until warning lights start illuminating in his EV, forcing him to pull over to assess the situation. Shortly thereafter, the car loses all power and he is locked inside with a dying mobile phone and no way to exit the vehicle. Worse, he has no manufacturer to call, sparking a rescue operation that involves people in Norway, New York, and Germany. This is real life. The man is Svein Hodne, his car, a new Fisker Ocean, luxury SUV. His story should be a warning and a lesson for all of us.[1]

Exhibit A: The Tumultuous Love Affair Begins

Forty years ago, in 1986, Buick introduced the first automotive CRT screen in the Riviera, followed in 1988 with the launch of the Reatta, a $25,000 (equivalent to ~$70,000 today) luxury two-seater. The Reatta was hand-assembled at the dedicated Reatta Craft Centre with a “Craftsman’s Log” documenting each build, styled as a showcase of craftsmanship and “digital sophistication.” At the time, the CRT screen in the car was a radical departure. Known as the Graphic Control Center (GCC), the 9-inch, monochromatic displays showed the things you would expect: gauges, climate controls, radio, vehicle diagnostics. The same CRT platform found its way into the 1990–1992 Oldsmobile Toronado Trofeo, but with a color display and referred to as the Visual Information Center (VIC). The Trofeo had one of the earliest in-car GPS systems. But, alas, drivers found the displays distracting and the tech-heavy models weren’t big sellers, so GM quietly discontinued them.

In the December 2024 issue of Car and Driver, Brett Berk quoted an independent technician, Jon Morlan, working with a specialty shop called East Coast Reatta Parts in Durham, NC: “They all fail. It’s just a matter of time.” The failure point is predictable: electrolytic capacitors inside the Zenith-built CRT units degrade and leak over 35+ years, eventually frying the image-driving chip and killing stereo, climate, trip computer, and diagnostic control all at once, since they all route through that one screen. Fortunately, they can be repaired or replaced. According to Berk, the shop’s owner, Marck Barker,, keeps roughly 30 salvaged units in stock and Morlan diagnoses, repairs, and rebuilds them. A refurbished screen runs $675, with a $75 refund if you send in your dead one as a core.[2]

This isn’t simply an expected problem with aging technology. A GM Technical Service Bulletin, 89-8-15, dated July 1989, corporate reference 918304R, explicitly addresses this: “DISPLAYS FLICKER OR TURN OFF AND/OR CODE E047 SET. REPLACE CENTRAL POWER SUPPLY. MODELS AFFECTED: 1986-89 RIVIERAS AND REATTAS.” That’s an internal engineering acknowledgment of a display failure mode, tied to a specific defective central power supply part number, circulated to dealers while these cars were still in active production.[3]

This is foreshadowing.

Exhibit B: Fisker Group, Inc.

On a sunny day in Los Angeles, in October 2016, Henrik Fisker and Geeta Gupta-Fisker founded their namesake car company, Fisker Group, Inc. This would be Henrik Fisker’s second time founding a car company. Fisker Automotive, the predecessor company, went bankrupt in 2013. This is the United States, however. It’s the land of second chances. Americans love a good underdog story. The new Fisker went public via SPAC merger in 2020. The Ocean model was announced in 2021, and Fisker began delivering their luxury SUV, the Ocean, to customers in mid-2023. On June 17, 2024, Fisker Group Inc. filed for Chapter 11 protection. And just like that, the company ceased to exist. The cars and owners did not.[4]

The Fisker Owners Association (FOA) cites “roughly 11,000” cars on the road today, or at least, owners with keys in hand, and likely a car payment too. How many Oceans are actually on the road being driven is unknown. This is how poor Svein Hodne found himself trapped in a car with no manufacturer support. His international rescue was orchestrated by the FOA and a Facebook group page.[5]

Exhibit C: Relationship Status — It’s Complicated

Fisker went out of business, leaving the Ocean to die a slow death, leaving the owners to fend for themselves and figure it out. There was no dealer network, no service network, no traditional support system, no more software updates, only a zombie fleet of vehicles wandering the earth. This is every car owner’s worst nightmare, and it’s the hurdle every new vehicle manufacturer has to overcome. When an owner buys a new vehicle, there must be some sense of trust that they won’t be stuck with a zombie car.

To be clear, and more importantly, precise: Fisker went out of business. The manufacturer, Magna International / Magna Steyr, a massive, publicly traded Canadian auto-parts and contract-manufacturing conglomerate, is still very much in business. As is the battery maker, CATL (Contemporary Amperex Technology). In fact, CATL is the world’s largest EV battery manufacturer, and is enormously healthy as a company. Harman International (Harman Automotive) supplied the Ocean’s infotainment head unit, FCC-filed as the “X297 Infotainment Headunit.” Amazon Alexa was integrated directly into the Ocean’s cabin for voice control, per the same FCC-published infotainment manual. In fact, dedicated members of the FOA identified, from that same public filing, Micron flash memory chips, an NXP ARM processor, a Silicon Labs AM/FM tuner chip, and a Lattice FPGA inside the infotainment unit. All are real, publicly traded semiconductor companies, and all are still very much in business.[6]

This is where things get very interesting. Although all of these companies still exist, still manufacture parts, and may still be able to provide support for the FOA, the functionality enabled by all that processing power, controlling the vehicle brakes, airbags, shifting, door locks, and battery management, routed through Fisker’s own cloud servers. This is the evidence that leads directly to Hodne’s stranding.

Exhibit D: The FOA Fights On, For All of Us

The Fisker Owners Association didn’t wait for permission. Members reverse-engineered the Ocean’s firmware. They mapped the CAN bus, the vehicle’s internal nervous system, so the car could be diagnosed without Fisker’s blessing. They incorporated nonprofits and spun up parts companies of their own, Tsunami Automotive in the US, Tidal Wave in Europe, to manufacture the bumpers and windshields Fisker never planned to keep making. When water pumps ran scarce in Norway, they ran a lottery to distribute what little stock existed, fairly, because no one else was going to.[7]

In October 2024, the FOA won something that looked like a resolution. Fisker’s unsold inventory, roughly 3,000 Ocean units, had been sold in bankruptcy to American Lease, a New York fleet company, for $46 million. Mid-sale, American Lease discovered the vehicle software and cloud data couldn’t simply be ported to a new server. The dependency wasn’t abandoned. It was structurally non-transferable. American Lease filed an emergency objection. The resolution: American Lease agreed to pay an additional $2.5 million over five years to keep Fisker’s original servers running, and the FOA gained access to the FAST diagnostic tools, an engineering help desk, and the car’s core software.[8]

Stipulated: that is a remarkable thing for a volunteer owners’ group to negotiate.

Here is what the resolution didn’t resolve. The deal was a handshake, never formally signed. American Lease eventually asked the FOA to cover 58 percent of operational costs: LTE connectivity, cloud services, software tools. The FOA asked for itemized invoices before paying. American Lease refused to provide them. The FOA refused to pay without the invoices. American Lease says it had already spent $860,000 on tech vendors with no matching contribution from the group it was meant to be sharing costs with. A botched over-the-air update, tied to an NHTSA recall, bricked roughly ten percent of the cars that downloaded it.[9]

By this spring, American Lease was reportedly preparing to sue the Fisker Owners Association for breach of a contract that was never actually signed. Hundreds of members have quit. This has taken a real toll on the volunteer who has led the fight, requiring time away to recover from the strain of representing and running a car company in his spare time. He bought a car. He became an unpaid CEO.[10]

Four thousand people, the FOA’s membership, did not sign up to become a car company. They became one anyway, so that eleven thousand owners could keep driving cars a manufacturer had abandoned. What they built at their own expense, on their own time, with no obligation to do it, is what Fisker should have engineered before the first Ocean left the factory floor. A support transition isn’t generosity. It’s a business and design requirement, and the fact that it’s now heading to litigation between two parties who never actually agreed on what they owed each other is not a footnote to Fisker’s failure. It’s the failure, still compounding, two years later.

The Verdict

Fisker was guilty of Indifference. They never planned or designed for their own mortality, even when the end was clearly near, because the roadmap only ever imagined winning. They didn’t honor the contract with Ocean owners or the debt of service it entailed. Bankruptcy isn’t the crime. Every company can fail. Failing to plan for the failure is the crime, and the plan would have cost a fraction of what it now costs four thousand volunteers in unpaid labor and legal exposure.

Fisker is the case study, not the target.

Projected product lifespan should be a disclosed attribute, stated plainly at the point of sale, the way fuel economy and safety ratings already are. In a sector selling cloud-dependent vehicles on five and six year loans, that disclosure should extend to the Right to Repair and to a binding plan for transferring cloud-based services in the event of bankruptcy or acquisition, not a handshake improvised after the fact by the people who can least afford to improvise it.

What the FOA built in the two years since is not a happy ending. It’s a stopgap, and the record now shows even the stopgap is in dispute.

Citations:

[1] Hodne stranding account

• https://www.wired.com/story/the-righteous-ev-owners-who-wont-let-their-broken-cars-die/

[2] Car and Driver, December 2024, Brett Berk / Jon Morlan / East Coast Reatta Parts

• https://www.scribd.com/document/787704784/Car-and-Driver-December-2024-USA

• https://www.eastcoastreattaparts.com/

[3] GM Technical Service Bulletin 89-8-15

• https://www.justanswer.com/car/19bmt-having-problem-digital-dash-electric-climate.html

[4] Fisker Group Inc. founding, SPAC, delivery, and bankruptcy timeline

• https://en.wikipedia.org/wiki/Fisker_Ocean

• https://techcrunch.com/2021/11/03/fisker-is-on-target-to-launch-the-ocean-electric-suv-in-november-2022

• https://www.businesswire.com/news/home/20211116006473/en

• https://www.businesswire.com/news/home/20230505005250/en

[5] FOA’s “roughly 11,000” figure

• https://www.wired.com/story/the-righteous-ev-owners-who-wont-let-their-broken-cars-die/

• https://electrek.co/2026/05/16/fisker-ocean-open-source-ev-story-after-bankruptcy/

• https://tech.slashdot.org/story/26/05/16/2318249/how-owners-of-evs-from-bankrupt-fisker-saved-their-cars-with-an-open-source-nonprofit

[6] Supplier ecosystem and chip teardown (Magna Steyr, CATL, Harman, FCC filing)

• https://en.wikipedia.org/wiki/Fisker_Ocean

• https://fccid.io/2ACRL-X297/Users-Manual/Users-Manual-6305677

• https://www.oceanforums.com/threads/harman-automotive-entertainment-unit-fisker-ocean-user-manual.1503/

• https://www.oceanforums.com/threads/harman-automotive-entertainment-unit-fisker-ocean-user-manual.1503/page-4

• https://fiskerati.com/electric-vehicles/fisker/fisker-ocean/amazon-alexa-arriving-in-fisker-ocean/

[7] CAN bus reverse-engineering, Tsunami/Tidal Wave, Norway water pump lottery

• https://www.wired.com/story/the-righteous-ev-owners-who-wont-let-their-broken-cars-die/

• https://hybridcopynet.wordpress.com/2026/05/03/fisker-owners-association/

[8] American Lease sale and October 2024 resolution

• https://www.davispolk.com/experience/fisker-confirms-and-consummates-chapter-11-plan

• https://tflcar.com/2024/07/fisker-american-lease-sale-approved-news/

• https://techcrunch.com/2024/10/16/fisker-owners-get-help-with-recall-repairs-as-court-approves-liquidation-plan

[9] The 58%/invoices dispute, $860,000, bricking

• https://carbuzz.com/ficker-ocean-american-lease-software-support-falls-through/

• https://www.autoevolution.com/news/one-year-after-fisker-s-bankruptcy-ocean-owners-are-still-paying-the-price-252214.html

• https://hybridcopynet.wordpress.com/2026/05/03/fisker-owners-association/

[10] Lawsuit prep, member attrition, volunteer leader’s leave of absence

• https://hybridcopynet.wordpress.com/2026/05/03/fisker-owners-association/

On August 23, 2025, I posted something on LinkedIn and on Facebook. Not a thought leadership piece. Not a hot take. A complaint, filed in public, with receipts.

I had just learned that Google would be ending support for the first- and second-generation Nest Learning Thermostats, one of my thermostats, effective October 25, 2025. The hardware worked fine. The software company that owned it had decided to move on. I wrote about the Honeywell T87, designed in 1953, still available for around $70, documented cases of it lasting 60-plus years. I wrote about being forced to choose between throwing functional equipment into a landfill or paying $150 for a replacement that would, presumably, eventually receive its own eviction notice. I tagged no one. I expected nothing.

The responses came from designers, technologists, enterprise architects, and one doctoral researcher in AI reliability. They came from LinkedIn and Facebook. They were unanimous in a way that public comments rarely are.

A year later, my Nest is still on the wall. Still working. I haven’t touched the app in twelve months because there is no app. And the house is exactly the right temperature.

This is the formal report.

Exhibit A: The T87

Henry Dreyfuss designed the Honeywell Round thermostat in 1953 [1]. It is a circle. It has a dial. You turn the dial. The temperature changes. That is the complete feature set, and it has not required revision in seventy-plus years because it was right the first time.

The T87 is still manufactured. It still sells for approximately $70 [1]. There are documented cases of units remaining in service for six decades and counting, one commenter noted that a 1977 Honeywell Round stayed on their wall until they replaced their entire HVAC system two years ago. Not because the thermostat failed. Because the furnace did. It contains no software, requires no app, receives no updates, and will not send you an email informing you that your relationship with it has been restructured.

Calibrate it against Dieter Rams’ principles: it is useful, honest, unobtrusive, and long-lasting. It passes all four without trying. It passes them the way things pass tests that were designed around them, because the T87 is not a product trying to be good design. It is what good design looks like when nobody is performing.

The forensic question for any Finished Work piece is: what decision did someone make to get this right when the default was to get it wrong? For the T87, the answer is that Dreyfuss made one decision and then stopped. He solved the problem. He did not add features to justify a price increase. He did not build in a reason to call back. He finished the job and left it on the wall.

An enterprise technology architect who joined both conversations, someone who has spent his career building systems that have to be trusted over time, who recently defended a doctoral dissertation on AI reliability, offered a parable that belongs in the record. About fifteen years ago, he went to buy a broom. The hardware store had two options: a traditional broom for $60 and a modern one for $10. The store owner was honest about the math: the $60 broom would last a lifetime. The $10 broom would need replacing every year or two. The store had to carry the disposable one to compete with the big box chains. He bought the $60 broom. It is still going strong.

“There needs to be a law,” he wrote, “that requires appliances will always work for their intended purpose, even without an internet connection. Or if the company goes out of business.”

The T87 is the $60 broom. It does not require a law because it was never connected to anything that could be switched off.

Exhibit B: The Nest

In 2011, Tony Fadell, former Apple engineer, iPod architect, launched the Nest Learning Thermostat. It was round. It was minimal. It was a wall-mounted dial that controlled temperature. The visual debt to the T87 was not accidental. Fadell has cited the Honeywell Round as a direct influence. The Nest looked like a T87 that had gone to graduate school. It was a beautiful, modern, fresh take on the T87. I was more than happy to buy one.

It also worked. It learned your schedule. It tracked energy usage. It connected to your phone. For a thermostat, a device that most people interact with exactly twice a year when the seasons change, it was genuinely impressive.

In January 2014, Google acquired Nest Labs for $3.2 billion [2]. Larry Page called it a chance to bring “great experiences to more homes in more countries”. Tony Fadell called it thrilling. Google called it a strategy.

What it actually was: a software company buying a hardware company, with a hardware company’s obligations, and a software company’s relationship to product longevity.

Software companies deprecate old versions. They sunset platforms. They end support for products that no longer serve the roadmap. This is normal, expected, and essentially painless when the product is an app. When the product is bolted to your wall, the calculus is different.

A hardware veteran in the product design community, someone with deep experience building consumer hardware at companies that take physical products seriously, put it plainly on LinkedIn: software engineers don’t understand hardware. He wasn’t being inflammatory. He was making a diagnosis. The decision to end support for Gen 1 and Gen 2 Nest thermostats was made by people whose frame of reference for “end of life” is a software release cycle, not a physical product that has been doing its job without complaint for thirteen years.

An industrial design educator offered the most efficient verdict across both threads. Three words: “Buy the Honeywell.”

A former CEO of the Industrial Designers Society of America said he’d chuck it and go for a cheap, simple replacement. Not upgrade. Replace. Exit.

These are not technophobes. These are people who have spent careers thinking about what objects owe the people who live with them. Their answer, independently arrived at, was the same: go back to the T87.

Exhibit C: The Announcement

On April 25, 2025, Google posted a notice to the Nest Community blog [3]. The language is worth reading carefully.

“We’re incredibly excited about the future of Nest and the innovative experiences we’re building for your smart home. To fully invest in these advancements and bring you the most cutting-edge features on our latest generation of Nest thermostats, we will be transitioning away from supporting three of our oldest devices, each over a decade old.”

Parse it: “fully invest in advancements” means we are not investing in you. “Transitioning away from supporting” means ending support. “Each over a decade old” is the attempted justification, as if age, rather than function, determines when a device has served its purpose.

Effective October 25, 2025: Gen 1 and Gen 2 Nest thermostats were unpaired from the Google Home and Nest apps [3]. Remote control: gone. Notifications: gone. Schedule changes from your phone: gone. The device remained physically functional, you could still walk to the wall and adjust the temperature manually, but the “smart” had been surgically removed by the company that sold it to you.

Google’s offer to affected users: a fourth-generation Nest Learning Thermostat at roughly $150, a discount off the $280 retail price [3]. One hundred and fifty dollars to replace hardware that worked. To restore features that had been working until the server decision made them stop working. The discount was framed as generosity. It was a toll.

A Chief Digital Officer who commented on LinkedIn described staring at his now-obsolete Gen 2 thermostats, trying to find another use for them. There wasn’t one. “Just more e-trash for the landfill,” he wrote. “Do better, Nest.”

Another commenter on LinkedIn, a senior executive with deep consumer tech experience, was direct: if forced to replace due to unnecessary obsolescence, he would exit the Google ecosystem entirely. He tagged Sundar Pichai. There is no record of a response.

This is not an isolated pattern. Sonos did it in 2020, bricking older speakers and facing a consumer backlash significant enough that the company reversed the bricking mechanism and issued a public apology [4]. On Facebook, a commenter described weeks of warranty battles over a Honda-branded EV charger rendered largely inoperable after Enel X Way, the company behind the JuiceBox charging platform, abruptly shut down its North American software and app in 2024 [5]. The hardware worked fine. The dependency killed it. The hardware is not the problem. The dependency is the problem. And the dependency was a design decision.

Exhibit D: The Ledger

Google has never disclosed how many Gen 1 and Gen 2 Nest thermostats were sold. The silence is consistent with a company that treats hardware as a platform play rather than a product line, the units matter insofar as they seed the ecosystem, not as objects with lives of their own. What we know: the Gen 1 launched in 2011, the Gen 2 in 2012. They were sold for three and four years respectively before the Gen 3 arrived in 2015. Google Nest is currently the number-one owned smart thermostat brand in the United States. The early units built that position.

Every one of those units is now either in a landfill, in a drawer, or, like mine, still on a wall doing the job it was designed to do, without the app that Google decided it no longer deserved.

The environmental cost of that decision does not appear on Google’s balance sheet. It appears in the waste stream. A thermostat weighs roughly half a pound. Circuit boards, display components, lithium, rare earth materials, none of it biodegrades. Google offered a postage-paid recycling label. Recycling is not the same as not creating the waste in the first place.

A supply chain technology executive who commented on LinkedIn made the policy argument: projected product lifespan should be a disclosed attribute, shared with consumers at the point of purchase. He is right. It is also not something any consumer electronics company is currently required to do, or voluntarily does. The circular economy has a language. Google chose not to speak it.

Exhibit E: One Year Later

My thermostat is still on the wall.

I have three Nest thermostats in my home. Two are still supported, Google’s deprecation was selective, not total. Those two are still in the app. I check them occasionally, out of habit more than necessity. The third, the one that controls the main living space, the rooms where I actually live, is the one that received the eviction notice. That one I manage by walking to the wall.

I will be honest about what I thought I would miss. Remote access seemed useful, the ability to turn on the heat before arriving home in winter, to check the temperature from the couch. It turns out I have never once needed to do this. I set the schedule when I installed the thermostat. The schedule runs. The house is comfortable. The absence of the app has cost me nothing.

The app still exists on my phone. I can see it. I can use it on occasion for the other two units. What I have learned, in practice, is that the feature I paid for and then lost was not load-bearing. The thermostat that can no longer phone home still does its job. The ones that can, I check infrequently, and when I do, nothing has changed that requires my attention.

What the app offered, it turns out, was not utility. It was the feeling of control, the ability to interact with a device that didn’t require interaction. A thermostat that works does not need to be checked. The checking was the product.

An Experience Design Principal in the thread offered the sharpest analysis of why people stay: even after her first Nest failed due to an electronics problem, she didn’t go back to a non-programmable thermostat. The lack of programmability, she argued, would take a greater environmental toll, she would inevitably forget to turn down the heat, waste more energy, spend more money. She stayed with Nest not out of loyalty but out of rational calculation. The alternatives were also insufficient.

That is the compensatory behavior in its most complete form: staying with the device that failed you because the alternatives are worse. Google has not earned that loyalty. They have engineered the conditions that produce it.

One commenter on Facebook, wary of smart thermostats entirely, put a sharper point on it: in an era when utility companies can negotiate remote temperature control as part of demand-response programs, “smart” may not mean what the buyer thinks it means. She prefers her regular Honeywell programmable. She has not been evicted.

The evidence closes there. What’s left is the verdict, and Exhibit E already gave away the mechanism: the checking was the product.

The Verdict:

The Nest Learning Thermostat was a genuine design achievement. Tony Fadell built something beautiful, functional, and smarter than anything that had existed on that wall before. It deserved better than what Google did with it.

The T87 is still on the shelf at your hardware store. It will outlast every device Google has ever manufactured. It will outlast Google’s interest in thermostats, which has already proven finite. It does not require a subscription, a platform, an app, or a relationship with a company whose priorities will shift before the product wears out.

Google acquired Nest for $3.2 billion and spent the next decade figuring out that hardware is different from software. The lesson cost them nothing. It cost their customers a thermostat.

The $60 broom is still sweeping. The $10 broom is in the landfill.

The Honeywell Round solved the problem in 1953 and left it alone. That is what finishing the job looks like.

Google didn’t finish the job. They ripped up the contract and handed you a bill for $150. That is Predation: design that keeps working exactly long enough to justify the purchase, then stops working exactly long enough to justify the next one.

Citations

[1] Cooper Hewitt, Smithsonian Design Museum, “The Round Thermostat”: https://www.cooperhewitt.org/2013/10/11/the-round-thermostat/ Honeywell Home / Resideo, T87N1000 product page: https://www.resideo.com/us/en/pro/products/air/thermostats/non-programmable/the-roundr-manual-thermostat-1-heat-1-cool-t87n1000-u/ SupplyHouse.com, current retail listing: https://www.supplyhouse.com/Honeywell-Home-T87N1000-Round-Non-Programmable-1H-1C-Mechanical-Thermostat

[2] TechCrunch, “Google Is Buying Connected Device Company Nest For $3.2B In Cash”: https://techcrunch.com/2014/01/13/google-just-bought-connected-device-company-nest-for-3-2b-in-cash/amp Forbes, “Look Out, Apple - Google Just Bought Nest For $3.2 Billion”: https://www.forbes.com/sites/roberthof/2014/01/13/look-out-apple-google-just-bought-nest-for-3-2-billion/

[3] Google Home and Nest Help, “End of support for Nest Learning Thermostats (1st & 2nd gen)”: https://support.google.com/googlehome/answer/16233096?hl=en Tom’s Guide, “Google announces end of support for 1st and 2nd gen Nest Thermostats — what you need to know”: https://tomsguide.com/home/smart-home/google-announces-end-of-support-for-1st-and-2nd-gen-nest-thermostats-what-you-need-to-know

[4] TWICE, “Sonos CEO Responds To Backlash, Will Continue To Update Legacy Products”: https://www.twice.com/product/audio/sonos-ceo-responds-to-backlash-will-continue-to-update-legacy-products Android Central, “Sonos CEO responds to backlash, will continue to update legacy products”: https://www.androidcentral.com/sonos-ceo-responds-backlash-against-update-policy-legacy-products

[5] InsideEVs, “Enel X Way JuiceBox Charger Customers Are Cut Off And Left In The Dark”: https://insideevs.com/news/736041/enel-x-juicebox-discontinued/ Consumer Reports, “What Should You Do if You Own a JuiceBox EV Charger?”: https://www.consumerreports.org/cars/hybrids-evs/enel-x-way-closing-what-to-do-if-you-own-a-juicebox-charger-a1912800096

On September 20, 2022, Spotify announced it was launching audiobooks in the US with 300,000 titles [2].

We’ve always believed that the potential for audio is limitless, and we’ve been saying for a while now that our ambition is to be the complete package for everyone’s listening needs. Audiobooks are next to come into the picture because we see a substantial untapped market. . . [2]

At launch, Spotify’s model was to enable listeners to see the catalog of books and to purchase them individually via their website for download and listening. Later, on October 3, 2023, Spotify shifted its model to include an integrated catalog of 150,000 audiobooks bundled directly into Spotify Premium subscriptions (offering 15 hours of monthly listening time) [3].

In the four years since audiobooks launched on the platform, Spotify has expanded their catalog to more than 700,000 titles. For Investor Day this year, Spotify announced “Audiobooks on Spotify are growing faster than ever. Listening hours are up 60% year over year, and nearly half of audiobook listeners started listening in the last year, with many discovering the format for the first time through Spotify.” Along the way, Spotify launched additional features like Page Match and a partnership with Bookshop.org, allowing readers to purchase physical books and seamlessly move between digital and physical experiences bringing their service to parity with Amazon’s Audible offering [4]. For those of us who love audiobooks this feels like an overwhelming win with access to more titles on a platform we already engage with.

Exhibit A: The Request

Fast forward to this week when I decided I wanted to finish Michael Murphy’s book. I found it in Spotify’s library and I downloaded it. Then, I tried to play it and I was unexpectedly confronted with a screen telling me I needed to request access from my plan manager. (Like many people, I am on a premium family plan with my husband.) So, I requested access and Spotify confirmed my request had been sent. This seemed innocuous enough.

Exhibit B: The Fine Print, Easily Overlooked

Later, my husband asks me if I had requested access to audiobooks on Spotify. Well, yes. He then tells me it will be an additional $11.99/month.

I’m sorry. What?

Let the evidence show that this was not disclosed to me at the time I was prompted to make the request. Let the evidence also show that in the subsequent email he received, the additional monthly charge is a subordinate visual element compared to the headline and the call-to-action to “Give Access To”.

Nothing else needs to be entered into the record. What remains is the finding.

Exhibit C: The Math That Doesn’t Add Up

Fifteen hours a month. That’s what $11.99 actually buys, once access is granted. The average audiobook runs 10 to 12 hours [5]. At best, this plan finishes one book a month with an hour or two to spare. At worst, a longer title spills into a second billing cycle before it’s done.

Neither Audible nor Libby caps listening time. Spotify’s own competitors, the services this plan is implicitly asking someone to abandon, don’t share this limitation. The product doesn’t fully solve the thing it’s charging for.

The Verdict:

I was played. I was enticed to put pressure on my husband to increase our monthly subscription costs without disclosure that this would be the outcome of my “Request Access.” In UX, we call this a dark pattern. In fact, this is a textbook example of a dark pattern, a UX trick to manipulate the user into doing something they did not intend to do. This is Predation at work. It is design that has been weaponized to create an outcome the user may not have chosen had they been presented with clear choices. I wanted to finish listening to a good book, not potentially spend $144/yr.

This could have been the beginning of the end of my Audible subscription and my usage of Libby. It could have been the moment I fully embraced Spotify as my home for all things audio (music, podcasts AND books.) Now, however, there is a rift in the relationship and Spotify has lost my trust. I had higher expectations for Spotify. I shouldn’t have.

The hoodwink was intentional.

As a designer, I spend a lot of time thinking about things like affordances, how design signals intention to users and how design can change behavior and action. I also think about smaller pieces of the whole and how bits and bobs ladder up to an overall experience. And, while I'm owning up to things, I'll also admit to being a car person who spends entirely too much time surfing around on Bring a Trailer. With these admissions, I shall devote this first Bits & Bobs post to automotive design (and probably not for the last time.)

Exhibit A: Interior Door Handles

Have you ever wondered why interior door handles on cars require you to pull the handle toward you before you push the door away from you? I have. This is the kind of experience that, although ubiquitous, is difficult to reconcile. Why would you pull in order to push? There is no logic to this. Who thought this was a good idea and why did the entire industry adopt this practice? Really, when you think about it, even a rotational knob like you might find on a standard architectural door makes more sense than what we see in automobiles. Yet, we don’t question this any more than we question using the “start” button to shut down our Windows computers. And, much like “click start to shutdown” scenario, we’ve fallen for the heuristics of consistency and recognition over recall. We don’t question it. We just pull the handle towards us before we push the door away.

Imagine my surprise and delight when we bought a Lexus RX350 last fall and the salesman showed me how the interior door handle worked: simply push the button away from you as you push the door away from you. WHAT? The function, the motion and the intention all working harmoniously? BRILLIANT!!! I find it completely smooth and intuitive. For me, it’s a fluid motion and it makes perfect sense. It is a small thing that changes everything. My husband, on the other hand, and to be honest, almost everyone who gets in the car, seems completely baffled by this when they first experience it because it is counter intuitive based on a lifetime of pulling toward you in order to push away from you. Click start to shutdown. Lather. Rinse. Repeat.

The initial struggle is likely worth it. The long term gain is a better egress experience from the vehicle. I applaud the Lexus design team for recognizing that there is a better way. I believe they have solved something that has been profoundly broken for 100 years. The fact that we have grown used to a broken system is not an excuse for keeping it. Cheers to Lexus. And, if you haven’t experienced this for yourself, go for a test drive. It’s magnificent, even if you do need to unlearn a lifetime of compensatory behaviors. Oh, and, if you ever find yourself in a situation where the vehicle has lost power and this fabulous electric button fails to work, you can, in fact, pull it toward you in order to push the door away from you. There is no getting locked in your own car. (I see you, Fisker Ocean, and we’ll be discussing you in three weeks.) This feature isn’t mandated by any industry regulations or government statutes yet, but Lexus solved the problem anyway, getting ahead of the competition. In January 2026, Rep. Robin Kelly (D-IL) introduced the SAFE Exit Act (Securing Accessible Functional Emergency Exit Act), which would direct NHTSA to write new rules requiring every vehicle door to have a clearly labeled, power-independent manual release, both inside and out. Kelly’s office says the bill responds to real incidents, including fatalities, where electronic door handles failed after a crash, fire, or power loss and trapped occupants or blocked first responders.[1] It’s important legislation.

There is a better way and Lexus found it.

Exhibit B: Turn Signals

Speaking of 100 years of muscle memory and sub-optimal systems, let’s examine turn signals. Car turn signals typically use a steering column stalk for easy physical reach, predictable mechanical movement, immediate physical feedback and intuitive alignment with the steering wheel’s natural circular motion. Moving the lever up signals a right turn, and moving it down signals a left turn, matching the clockwise and counterclockwise flow of your hands. An added benefit of this placement next to the steering wheel, the stalk allows drivers to flick the switch using just their fingertips without taking a full hand off the wheel. Right-hand drive cars typically have the stalk on the right side of the steering wheel, while left-hand drive cars frequently position the stalk on the left side. According to Says.com, automakers rarely reposition the stalk for right-hand-drive markets because it isn’t a simple mirror-flip; it requires re-engineering the entire steering column and control system, an expense most brands choose to avoid by keeping the original placement everywhere [2]. As a result, there is no global standard for which side of the steering wheel the turn signals must reside. This means every car requires a moment of orientation for the driver.

On a recent Saturday morning in June, I found myself test driving a Ferrari Purosangue. This is Ferrari’s answer to the small SUV, although they refuse to call it that, preferring instead to call it an FUV, a Ferrari Utility Vehicle. It’s a 4-door monster with 715 horsepower (725 PS / 533 kW), from a naturally aspirated 6.5L V12: the same basic architecture as the 812 Superfast, tuned slightly down for this application [3]. Paired with 528 lb-ft of torque, it’ll do 0-60 in about 3.3 seconds [3]. I can attest to this. While it does have an automatic mode, it’s best driven and experienced as a paddle shifter. The Purosangue has an 8-speed dual-clutch transmission (DCT) mounted at the rear axle in a transaxle layout, which also helps keep weight distribution closer to 50/50 [3]. All this is to say, it’s a Ferrari through and through and it drives like a dream. Since it is intended to be driven and experienced as a paddle shifter requiring two hands on the steering wheel, Ferrari has decoupled the turn signals from a single stalk design to a two-button system, with one button accessible by each thumb for quick activation.

This was my first experience with a de-coupled turn signal system. I liked the concept very much, I struggled with the muscle memory to activate the signals consistently as I drove. Again, a lifetime of relying on a single stalk for this activity resulted in a certain amount of cognitive dissonance. I think I would learn this new pattern quickly, but it did require additional, intentional cognitive processing on my part. (As if test driving a $500,000+ car wasn’t stressful enough . . .)

Again, I like the concept very much, but even though it makes sense, it didn’t immediately land as an obvious improvement. It’s a rare approach, but not wholly unique to Ferrari. Tesla and Citroen have also tried to remove the stalk in the past. Ultimately, they relented and came back to it. Even Ferrari is reversing course on this design decision.

Some habits are just too hard to break. For me, this is proof that the best ideas don’t necessarily win. It’s a good reminder to never show a client or customer the weakest concept (the stalk). It’s the one they will inevitably pick and you’ll have to live with the decision forever. I appreciate the attempt to break a behavioral pattern and to look for a better approach. This time, there was just too much momentum and the fluidity of the activity didn’t quite match the necessary behavior. Perhaps larger buttons with more of a tactile texture would have helped. Either way, it’s a valiant attempt and the design team garners my full respect. I hope they will keep challenging old patterns.

And, if you happen to have an extra $500K - $600K, I wouldn’t hesitate to recommend a Purosangue, even if you do need to think about the turn signals in order to use them effectively.

Exhibit C: Rims

Two weeks ago, I found myself driving around north Georgia in a new Ford Expedition. (There may be a forthcoming post on the driver interface.) It was another great vehicle experience, comfortable, well appointed, all the things. We stopped for lunch and as I was closing the door, I noticed the color-coding on the rims. All four wheels had an intentionally colored inset at the location of the air valve. I was immediately struck by how simple and compelling this idea was. A pop of color adding flair to the vehicle and providing an affordance when airing up the tires (or down for overlanding). It was not something I had ever seen done before. Unique. Interesting. Helpful. Neither garish nor too subtle to be noticed. It felt thoughtful and well considered. Purposeful design works harder than it needs to and wins in a category dominated by sameness.

The Verdict:

Today wasn’t about great products. It was about three tiny interactions that demonstrate three different truths:

• A door handle: some behaviors deserve to be replaced,

• A pair of buttons: some behaviors are too deeply learned,

• A ring of color around a valve stem: some tiny details make ordinary products unexpectedly delightful.

None of these decisions required a research budget or a design award. They required someone to notice a moment most people stop seeing after the first few times they experience it, and to ask whether it could be better. Two teams got it right. One got close but stopped short. That’s the whole business, really. Not genius. Just paying attention. (But, ALWAYS advocate for a research budget!! The outcomes will be better every time!)

"A user interface is like a joke. If you have to explain it, it's not that good." — Martin LeBlanc, founder and CEO of Iconfinder.

Honestly, we could stop right now. The case for bad design has made itself via a self-indictment and guilty plea. The verdict is already in.

Let’s zoom out anyway, both literally and figuratively. This is the parking kiosk I encountered at Kennesaw Mountain National Battlefield Park on a recent morning, after realizing my America the Beautiful pass was sitting on my kitchen counter at home, which meant buying a day pass to park. Inconvenient, but not a deal breaker.

I’m a true fan of the National Park Service and don’t mind spending an extra $5.00 once in a while, above and beyond the $80.00 (plus $7.50 shipping and handling) I’ve already spent this year on my annual America the Beautiful National Parks and Federal Recreational Lands Pass [1]. The pass is a tremendous value. The National Park Service manages 433 properties nationwide, including 63 national parks plus monuments, historic sites, battlefields, seashores, recreation areas, and parkways. Add in the Forest Service, BLM, Fish & Wildlife, Reclamation, and Army Corps lands, and the pass covers 2,000-plus sites. Collectively these agencies manage more than 26% of public land in the United States [2]. I appreciate the need to charge entry and parking fees, both to raise revenue and to manage visitor volume.

In researching this piece, I assumed the NPS and the agencies covered by the America the Beautiful Pass had consolidated their contracting and purchasing power for parking kiosks, in true government fashion, to get the lowest price and standardize maintenance costs. My research says otherwise. It’s more likely a patchwork: some equipment owned and operated by concessioners, like Xanterra at Mount Rushmore; some purchased independently by individual park units, possibly off shared GSA schedules for convenience. What is clear is that the parks segment represents roughly 10% of the parking meter market, according to ReAnIn [3], and demand for outdoor recreation is only growing, especially near urban areas. More demand. More kiosks. More of the buttons that aren’t a button. North America alone represents about 40% of the global parking meter market. This isn’t a niche failure. It’s infrastructure, multiplying.

The kiosk I encountered is the VenTek International Model 600 [4]. It is advertised as having a “unique and aesthetically pleasing appearance” and a “familiar ‘ATM-style’ customer interface.” In fact, VenTek claims its machines use the simple-to-use customer interfaces adopted by American ATM and gas pump machines, explicitly stating this makes their units “easier to use” [5].

For something so familiar and easy to use, it needed four additional labels. Each was produced with a label maker and applied on-site by NPS staff, evidently in response to a volume of complaints and the daily tax of helping would-be parkers who couldn’t operate the machine on their own.

Exhibit A: The Button That Isn’t a Button

This morning, at the same machine, a friend who is a medical professional with two bachelor’s degrees and a master’s degree walked up to pay for parking. She instinctively pressed the thing that looks like a button, but isn’t a button. The screen stayed off. She said, out loud, that it was stupid for something that isn’t a button to be sitting there looking exactly like one. She hadn’t registered the label until after she’d already pressed it and nothing happened. This is not a hypothetical user. This is a credentialed, highly educated, capable adult, defeated by a visual deceit in under three seconds.

The unit is solar-powered and rests in a screen-off, power-saving state until a patron engages it. This is why she reached for the button-like object centered directly under the screen. The instructions printed on the kiosk read “Follow Directions on Screen to Pay Fee”: a screen that is, at that moment, off.

Attempting to wake the kiosk likely leads most users to press the thing that looks like a button but isn’t one, which is why the kiosk carries a label reading “THIS IS NOT A BUTTON.” Above it, a second label nobody would intuit on their own: “PRESS # TO BEGIN.” This is not how ATM machines or gas pumps work. It is not how airport ticket kiosks, movie kiosks, Panera or McDonald’s ordering kiosks work. It is not familiar. That is why it needs a label. The indictment is clear. The case made.

Exhibit B: The Payment the Kiosk Implies but Doesn’t Accept

It’s time to pay. In a park setting, where people may not want to hike with a wallet, the reasonable expectation is mobile payment. ReAnIn reports that 58% of new parking meters now offer contactless and mobile options [3]. This kiosk does not, despite looking as though it should. A third label makes the shortfall explicit: “MOBILE PAY NOT ACCEPTED,” a sentence that exists only because the machine’s own appearance promised something else it doesn’t deliver.

Exhibit C: The Label Parallax Hides

There’s more, and you can’t see it from the angle a user actually stands at, because parallax is real. Parallax is the apparent shift in an object’s position depending on where you view it from. Standing in the position the kiosk requires for the screen, a user cannot see the fourth label, reading “INSERT CARD ABOVE.” So even the label meant to solve the issue has come up short, a failure upon a failure.

Exhibit D: The Company Behind the Kiosk

How did a “familiar ATM-style customer interface” fall this short, this many times, on one machine? The evidence points upstream, to the organization that built it.

VenTek International describes itself as “a company built on engineering excellence” [6]. That claim checks out, and it tells us something real: this kiosk is durable, and it will keep working, reliably, for years. Those are genuine hallmarks of engineering excellence.

The engineering mindset is built to solve hard problems systematically: analytical, tracing complex systems to root cause; curious, always asking why something works the way it does; tenacious, persisting through broken or difficult designs. It is a mindset that starts from the system and its constraints, folding human need in as one more requirement to satisfy rather than the starting point. Design starts elsewhere, from human values and empathy for the user, built through direct observation, reasoning toward what might be rather than only what is.

LinkedIn shows VenTek as a small American company, under 50 employees. Of the 24 profiles visible, none carry a background in design, UX, interaction design, or human-computer interaction. The company skews toward computer science, electronics, and computational fields, with ten people emphasizing project management skills. That is a real commitment to manufacturing quality. It is also, on its own, not enough.

The proximate cause is a mislabeled protrusion impersonating a button. The distal cause is an organization that never had anyone in the room asking what the button-like shape might communicate to users and whether there was a better solution.

The distal cause runs deeper than one product cycle. A Forest Service field evaluation from 2000 describes an earlier VenTek pay station using the same pitch: an ATM-style, menu-driven interface prompting customers to choose from preset rates [7]. That is nearly identical to how VenTek describes the Model 600 today. The company was already selling this interface philosophy to federal agencies in the 1990s, including more than 60 machines to the National Park Service by 2000, starting with the first unit purchased by Mt. Rainier in 1996 [7]. Thirty years is not an oversight. It’s an ingrained policy.

It also wasn’t the only approach on the market in 2000. That same field report describes a competitor, the QBS Concord Pay Station, built with a motion sensor that triggered recorded audio instructions on approach and a Help button that repeated them on demand [7]. A quarter-century ago, someone else in this same product category solved the exact problem documented in Exhibits A through C: a user who doesn’t know how to begin. VenTek didn’t fail to solve a hard problem. It failed to adopt a solution that already existed, from a competitor, in the same market, at the same time. It also failed to build a solution of its own, one that could have up-leveled its product line and beaten QBS at their own game. Interestingly, there does appear to be a speaker in the Model 600 although it is not being used to communicate instructions.

The fixes here aren’t exotic, and they never were. Observational research at installed kiosks would surface every failure in this piece within a single afternoon. Iterative concept testing would catch them before manufacturing or installation. The NPS staff who ran a label maker four times on this machine already ran the feedback loop VenTek didn’t: they just weren’t in a position to redesign the interface, only to patch it. They likely aren’t in a position to positively affect the next kiosk the NPS purchases either. I’m willing to bet nobody has asked for their feedback, and there is no virtual suggestion box.

Exhibit E: The Machine They Built When It Mattered

Here is the complication. VenTek is not incapable of this work. In 2019, the company commissioned Design + Industry, an Australian industrial design firm (part of Capgemini), to develop a next-generation Smart Fare Vending Machine: three-plus years of development, more than 600 requirements, built for North American mass transit [8]. The result guides a user through the entire purchase flow with sequential rim lighting on a 12-degree angled interface panel: insert card, scan code, enter pin, collect ticket. It includes Braille labels and tactile icons for vision-impaired users, ADA compliant from the start [8]. Good Design Australia’s own case copy for the award states plainly that the SFVM’s interface was built to avoid confusing labels and ad-hoc directions common to the category [9]. It won the 2021 Australian Good Design Award for product design [8]. That machine didn’t need four labels added after the fact, because the guidance was designed in before it shipped.

The SFVM was built for roads, railways, and waterways: markets with transaction volumes an NPS battlefield parking lot will never approach. VenTek’s engineering excellence is real. So, evidently, is VenTek’s design excellence, when the volume justifies it. Parks and recreation didn’t make that list.

The Verdict

VenTek proved, in 2019, that it knows how to design guidance into a machine instead of needing it to be taped on after the fact. That know-how got three years, six hundred requirements, an industrial design firm, and an international design award, applied to mass transit. The parking kiosk at Kennesaw Mountain got a label maker. This is not a company that doesn’t know better. This is a company that decided where the investment was worth it, and a National Battlefield parking lot didn’t clear the bar. That is Diminishment: doing the right thing, deliberately, and then deciding a smaller version was good enough for whoever wasn’t buying transit tickets.

This was not the first time I had rented an EV, but it was the first time I had to do any meaningful driving in one and it was the first time I had one for more than an overnight experience.

We are shown how to use the “key”, a flat, plastic credit card affair in a protective case. Weird, but okay. To lock and unlock the door, tap it against the driver’s side door pillar (just below the camera) and drag it down the door toward the handle. To start the car, place the key card on the center console wireless phone charger and press the brake pedal. That was the full extent of our onboarding. We didn’t think to ask any other questions and no other onboarding was offered.

We loaded our baggage and headed south, realizing almost immediately that we were driving into a set of challenges we didn’t anticipate and weren’t prepared for.

Exhibit A: The Ownership Assumption

Unlike a gas vehicle, we didn’t have a “full tank” when we left the rental facility. The car was charged to around 70%, so we were immediately confronted with the range question and how far we might travel on 70% battery. Nevertheless, this was a problem that didn’t need to be solved immediately.

My husband took the first shift driving as we left the city. The weather was overcast with intermittent sleet, rain and snow. Temperatures were hovering around freezing. Road conditions seemed fine, but worth being mindful of as we drove on, likely adding a bit of anxiety. My husband is a consistent cruise control user. He couldn’t find a way to engage it on this car. And, so, my first (of many) ChatGPT searches began. “How do I activate cruise control on a Tesla Model Y?” Turns out, Tesla doesn’t have “dumb cruise control” -- the kind where you press a button, set a speed, and it holds it. What it has is Traffic-Aware Cruise Control (TACC), which is camera and sensor based -- and how you set it depends on the model year.

What? So, now, I’m flipping through the rental agreement and the touch screen on the car trying to determine what model year car we’ve rented.

According to ChatGPT, “On older Model Ys (pre-2024 Juniper refresh): there was a stalk on the right side of the steering wheel. Push it down once for TACC, twice for Autopilot with Autosteer.” Okay, our car didn’t have this stalk, so we must have a 2024 or 2025 model. On the 2024+ Model Y Juniper, Tesla removed the stalks entirely. There are now scroll wheels on the steering wheel. “To engage cruise control you press the right scroll button. Then adjust speed using the same scroll wheel.” There is no stalk, no dedicated cruise control button, and no labeling that suggests any of this to a first-time user. But, wait, there’s more: TACC is bundled with Autopilot. One press of the scroll wheel can activate full Autosteer -- the car starts steering itself -- before you realize what you’ve done. The settings menu buries the option to separate TACC from Autosteer. Tesla owners know this. Renters like us discover it at highway speed. [2]

You can only imagine the frustration and agitation building in the car as I’m trying to do this research and my still somewhat jet-lagged husband is trying to drive, navigate traffic and weather and follow along with the directions I’m reading aloud.

This is when a kernel of an idea first starts to take shape in my brain: Unlike “traditional” cars, EV systems assume a driver who has lived with the car for months and is familiar with foibles like these. Tesla cruise control works differently. An owner would know this. Renters like us get none of that onboarding. Instead, we are thrust into an experience for which we have no mental model for deciphering.

Exhibit B: The Charging Problem

In my previous EV rental experiences, I had not needed to charge the vehicles, so this was going to be a new experience as well. I was broadly aware of the existence of charging stations and of Tesla Superchargers. I knew that charging took time and we should expect 30 minutes to an hour for this. I knew that in Norway, most gas stations and convenience stores also had charging stations, so there is an abundance of places to charge the car. What I didn’t know, but would soon find out included the following:

• Unlike a gas pump, there is no universal tap-to-pay. Charge stations are app-based, often with full account setup required to use the app. (Glad I paid for that international data plan on my phone!)

• There are multiple competing charging networks, each with their own app.

• My brother-in-law had advised me to download the CircleK app for the easiest and most convenient charging. The problem with this was that it was a Norway-specific app. Apple’s US App Store wouldn’t let me download the Norwegian app for CircleK <sigh>. What I didn’t learn until researching this piece: Elton actually includes CircleK charging within its network. [3] The solution to the CircleK problem was already in the app Hertz had directed us to download. Without onboarding, there was no way to know that.

• Cable and connector types vary by car and station (CCS, NACS, CHAdeMO) [4]

• Charge time has no gasoline equivalent -- 20 minutes to 8 hours depending on level of charge and type of charger

• Return policies for rental vehicles penalize renters who don’t know the rules and don’t return their vehicles with “enough” charge. Hertz charges a $35 fee ($25 for members) if you return the EV with less charge than at pickup. [5]

As we continue our trip, we are now watching the battery charge level drop. I start texting with friends back home in the US who have EVs, asking them questions about road trips in EVs, battery management, etc. and I’m advised the EV trip planning problem has a solution. It’s called “A Better Route Planner” (ABRP) and it does exactly what it sounds like: plans your route, calculates charging stops, accounts for temperature, elevation, and driving style, and shows real-time charger availability. For an EV owner, it’s close to essential. I download the app and immediately encounter a new set of problems. First, the app wants to know what software build the car is using. So, it’s back to ChatGPT to discover how to find the software build in the car’s interface. One more unexpected hurdle down and onto the next one: Connecting ABRP to a Tesla requires owner credentials through the Tesla API -- something a renter, like me, doesn’t have and a rental company doesn’t provide. Without that connection, the app is making educated guesses about the car’s battery state based on whatever you’ve manually entered. In freezing temperatures on an unfamiliar charging network, in a foreign country, educated guesses are not what you want. [6]

When we have reached 20% battery, we opt for a Tesla Supercharger and more ChatGPT to learn that payment for the Supercharger is connected to the car. In this case, no app required, but also, no real idea of what this charge time will cost us once it’s passed through from the rental company either. This sort of obscurity breeds additional anxiety in the rental experience. It’s also at this point that we begin to understand we will likely not be getting a full charge on the battery, but rather an 80% charge. As it turns out, this is exactly what Hertz’s own terms specify -- the charging limit should be set to 80% maximum. Tesla’s recommendation for daily use is the same: charging to 80% protects battery longevity by reducing voltage stress on the cells. [7] The policy existed. Tesla’s recommendation existed. They agreed with each other. None of it was communicated at pickup. In all fairness, I still don’t fully understand the philosophy/reality of battery conditioning and management for optimal battery life, but I do acknowledge it’s “a thing” I would need to know a lot more about if I owned an EV. But, again, renter here. Renter on vacation. Renter trying to have a relaxing vacation and not succeeding thus far.

Fortunately, we put the charging destination into the nav system on the car, and it told us it would pre-condition the battery for charging, a stroke of good fortune, as you’ll see in the next section.

Exhibit C: The Cold Weather Problem

We are far enough north that in late December, we only have around 5-ish hours of “daylight” and we are now driving in the dark and the temperature is dropping. We are also driving through alternating light snow, sleet, and potential ice. We have range anxiety on top of weather anxiety.

In researching this article (and to some extent doing real-time research as we were traveling) I learned the following:

EV batteries use lithium-ion chemistry, which slows down in cold temperatures -- the chemical reactions that move energy in and out of the battery become less efficient. The practical effect: range drops meaningfully in cold weather. The Tesla Model Y, one of the better cold-weather EVs on the market thanks to its heat pump, still loses roughly 20-35% of its EPA-rated range (a warm-weather number, by the way) in typical freezing conditions. Push into extreme cold or add highway speeds and cabin heating and that loss can reach 40-45%. [8] Norway in December is cold and we are traveling at highway speeds using the heater. Other compounding factors specific to cold weather that a renter almost certainly doesn’t know, and that weren’t mentioned or disclosed when we were given the plastic key:

Battery preconditioning: Before a fast charging session, the battery needs to warm up to accept a charge efficiently. The Model Y does this automatically when you navigate to a Supercharger through the car’s built-in system. If you found the charger another way -- on a map, through ABRP, by spotting it from the road -- the battery may not be preconditioned and charging will be significantly slower. There is no warning. The car just charges slowly and you are left wondering why.

Heated seats vs. forced air: The cabin heater draws heavily from the main battery and is one of the biggest range consumers in cold weather. Heated seats use a fraction of that energy and keep you just as warm. Tesla’s owner’s manual (which we did NOT have) recommends seat heaters over high cabin temperatures in winter. Tesla owners know this. It wasn’t in our rental briefing, and the default when you’re cold is to turn up the heat. Fortunately for us, this is a tip one of my stateside friends shared along with the ABRP app recommendation, so we had already adjusted our behavior, but not before we were 90 minutes into our drive.

Regenerative braking: Regen is reduced in cold weather because a cold battery can’t accept charge as quickly. The car handles differently than expected and the one-pedal driving behavior that EV owners rely on is diminished until the battery warms up. If you didn’t know to expect it, the car feels wrong. Again, not disclosed at pickup and while we couldn’t discern a difference in car feel per se, we incorrectly assumed we were getting some benefit here.

All three of these are things an owner learns over time, through lived experience with the vehicle, through research and through the purchase and onboarding experience. A renter in Norway in December encounters all three simultaneously, with no warning and no context.

Exhibit D: What Good Looks Like

Rental companies with EV fleets owe their customers an onboarding experience. When basic driving paradigms are radically different (locking/unlocking the vehicle, cruise control, autosteer, battery pre-conditioning, use of heated seats over forced air, assorted fueling and planning apps, etc.), customers can’t be expected to intuitively understand these things. Figuring them out at highway speed or when engaged in city driving isn’t just stressful, it’s also a safety hazard for everyone. Providing basic information about features like these and about plug type, voltage, route planning, charging times, etc. should be mandatory and should be part of a reinforcing ecosystem for the renter. The solution isn’t complicated. It’s just absent. Simple ideas to improve the experience (and the driver’s confidence) might include:

• Better in-person vehicle orientation with the staff at the rental agency at the time of vehicle pickup including tips for weather, and details about charging, provided adapters, cables, etc.

• A dedicated website with quick tips, FAQs and imagery to support the rental experience. The URL could be presented as a QR code on the key card, fob or key ring and within the rental agreement -- all easily accessible during the rental period

• A printed pamphlet given to the renter along with the rental agreement that contains relevant basic information.

• A 24/7 help number or chat access for troubleshooting

• EVs require an app ecosystem that is completely foreign to those of us accustomed to internal combustion engine vehicles. Rental companies could provide a branded “app store” environment that enables renters to download the best apps for the region, thus bypassing the problem of the App Store region being locked to your US Apple ID.

• A quick way to meaningfully connect the car to an app like ABRP for optimal route planning and battery management.

In retrospect, the Tesla Supercharger network model is the standard to beat for a rental experience, especially in a foreign country. It requires no app and no account setup. The billing is automatic. Since we were in a Tesla, there was no guessing about the plug type or the charging voltage. It was seamless. Navigate to it in the car, plug in, done. When we found one, it worked. Even with the cost obscurity, this is what the entire EV charging experience should feel like. It mostly doesn’t. I now know that, when driving a Tesla, looking for a Supercharger is the way to go. Seven months later, I still didn’t know if a non-Tesla vehicle could use a Tesla Supercharger until I looked it up to write this piece. The answer: In the US and Europe, non-Tesla EVs can use Superchargers through the Tesla app (there is always an app!), but you must set up an account, add a payment method, and use the app to initiate the charging session rather than it being automatic as it is for Tesla owners. You also need a CCS adapter, which Tesla sells for $230. But, some Supercharger stations have tethered CCS cables built in. I have no idea if the app provides this information -- or if your non-Tesla [rental] EV provides this adapter. [9]

The Verdict

The Model Y was a comfortable car to travel in. It handled well and was appropriately powerful and responsive -- enough so that we are considering purchasing one. But, the learning curve is steeper than it seems. Regenerative braking is a new concept. Climate strategy -- heated seats vs. forced air -- matters more than you’d think. Needing to know which software build is on the car to understand a feature or an upgrade that has been pushed over the air, is wild. That is not a rental experience. That is homework.

The EV is a brilliant product for the person who owns it. For the renter -- even an experienced traveler on a familiar route -- it’s a product that assumes six months of familiarity you don’t have. That gap shouldn’t be the renter’s problem to solve.

We figured it out. It took two of us, a data plan, and multiple ChatGPT searches to do it. That is not a rental experience. That is an expedition. The next renter, driving alone, may not be so lucky.

There are many reasons I love July and the Tour de France is definitely one of them. I’ve been a cyclist most of my life and although I run more than I ride these days, cycling remains near and dear to my heart. Cycling is beautiful for many reasons. As children, cycling expands our world by gifting us the freedom to explore beyond our immediate yard or neighborhood block. As adults, cycling is a fantastic sport for fitness, adventure, and touring. It may also be daily transportation, providing fitness benefits alongside the practical role of serving as transportation. The freedom cycling provides is powerful. It’s possible to cover great distances under your own physical effort. At the end of a ride, whether it’s 10 miles, 25 miles, a metric century (62 miles) or a full century (100 miles), a cyclist feels like s/he has been somewhere and accomplished something. And, during an effort like this there is always a Zen that is achieved, akin to a “runner’s high” where the cyclist and the bike are one. There is no thinking, there is only the harmony of movement and it is glorious.

The Tour de France shows us what is possible when elite riders saddle up. This year, the peloton will cover 3,333 kilometers, approximately 2,071 miles, over 21 stages and 23 days, including two rest days. [1] They will do this at race pace while climbing mountains. It’s an incredible feat of human achievement. Cycling isn’t just for elite racers, however. In 2024, a record 112 million Americans rode a bike at least once, the highest participation rate since PeopleForBikes began tracking in 2014. [2] Women make up about 25% of all cyclists in the US, compared to 49% in Germany and 55% in the Netherlands. Women’s cycling participation has increased by 53% since 2013, with growth outpacing men in commuting. [3]

Multi-day tours, Gran Fondos and other organized rides for competitive and recreational riders continue to grow in popularity. RAGBRAI, the Des Moines Register’s Annual Great Bicycle Ride Across Iowa, is the oldest, largest, and longest recreational multi-day bicycle touring event in the world. It is a non-competitive ride traversing Iowa’s full width from the Missouri River to the Mississippi River, held annually the last full week of July. RAGBRAI began in 1973 and through 2024, 474,103 riders have participated over the years. [4] It’s lauded as a great event drawing participants of all ages and is often referred to as a “rolling party.” All of this is to say that cycling is an accessible sport/activity for men, women and children of all ages.

Cycling is also synonymous with design and innovation. Every component on a modern road bike has been iterated toward perfection. Carbon fiber frames engineered to flex in exactly the right directions, electronic drivetrains that shift with the press of a button, wheels so aerodynamically refined they are wind-tunnel tested before they go to market. The bicycle, across 150 years, has been one of the great proving grounds of materials science and industrial design. And then there is the saddle: three components, leather, rails, and rivets. Unchanged in its essential logic since 1882, still the component that determines whether you finish the ride and whether you will fall in love with the sport. To be clear, there are very few products with which we have such an intimate relationship as the bike saddle.

Exhibit A: The Brooks B17

I ride a full carbon fiber bike, with carbon wheels and SRAM Red components. The bike is set up to my exact ergonomics and anthropometric measurements. It fits me like a glove. When stripped down, the frame weighs little more than the cardboard coaster your post-ride beer sits upon. Why then, would I choose to put a 440g leather saddle on my super lightweight, high tech road bike? Why does that make sense? Because the leather saddle, hand-riveted in Birmingham, England using machines from the 1950s, is the one component where the engineering solution from 1882 is still, for many riders, including me, the best answer available. I ride with a Brooks B17 Special Titanium saddle and I wouldn’t trade it for the world. I love it for the aesthetics, the history, the ability to be repaired, but most of all, for the fit and comfort it provides.

Brooks England has been making leather goods since 1866, when it was founded in Hockley, Birmingham. Bicycle saddle production began in the 1880s, with the first saddle patent filed in 1882. According to company legend, founder John Boultbee Brooks, a horse saddle manufacturer, tried to use a bicycle after his horse died and found the wooden seat unbearably uncomfortable, prompting him to design a better one. An early version of the iconic B17 saddle launched in 1888, the same year Dunlop developed the pneumatic tire. By 1903, Brooks saddles were on the bikes of Tour de France competitors. [5] Raleigh Bicycle Company bought Brooks in 1962. When Raleigh collapsed in 1999, Brooks went into liquidation before being rescued by two buyers. It is now owned by Selle Royal of Italy, though still manufactured in Smethwick, Birmingham. The B17 has been in continuous production since 1888 and remains Brooks’ best-seller. Leather, rails, and rivets -- three components, individually repairable.

The Brooks B17 is a marvel of design and engineering. The Brooks solution is elegant and counterintuitive: don’t design the saddle to fit the rider. Design the saddle to become the rider. It is as hard as a rock and will hurt your knuckles when you knock on it, making a sound like knocking on a door. Yet, leather, under load and time, conforms. The break-in period, usually around 500 miles, is not a flaw. It is a feature. At the end of it, you have a saddle that is a physical record of your body. Nobody else can ride it comfortably. That is the point. It is the saddle you’ll have for life because it becomes a record of your body and it is built to last a lifetime. It is a lasting example of Dieter Rams’ Ten Principles of Good Design brought to life. It is useful, honest, unobtrusive, long-lasting and as little design as possible. The company’s own repair infrastructure -- saddles repaired by the people who made them, in Birmingham, is the physical manifestation of “long-lasting” as a design principle. The record shows: Brooks didn’t solve the saddle problem for everyone. But it solved it permanently for the riders it fit who were also willing to invest the time to break it in. It creates a symbiotic relationship with the rider as well. As long as the rider takes care of the saddle, the saddle will take care of the rider.

Exhibit B: The Problem

As great and beloved as it is, the Brooks saddle requires an investment: 500 miles of discomfort before the payoff. And it conforms to one body and one position, which means it assumes a body type and a posture. It can’t easily be swapped from a road bike to a mountain bike, for instance. Different body geometry on the bike.

The saddle problem isn’t just about material and craftsmanship or about posture and bike geometry. It’s about anatomy, specifically, the ischial tuberosities (sit bones), which vary significantly by individual. A saddle that doesn’t support them correctly concentrates pressure on soft tissue: nerves, blood vessels, structures that do not appreciate being compressed for hours. The result: pain, numbness, and people who leave the sport. The compensatory behaviors are easily recognized: standing on the pedals, chamois cream, padded shorts. None of these fix the saddle. They manage the failure.

Like many sports, the innovation is highly concentrated around the needs of professional athletes first and foremost. Casual, amateur participants benefit from a halo effect and an innovation pipeline that slowly trickles into the mainstream. Cycling is no different. Similar to other sports, elite cyclists have a body type that is fairly specific to the sport and bike saddles have largely been designed to serve the needs of an elite rider. The average Tour de France GC contender is 5’9”, 140 lbs. [6] In contrast, Travis Kelce, tight end for the Kansas City Chiefs is 6’5”, 250 lbs and Victor Wembanyama, center for the San Antonio Spurs, 7’4”, 210 lbs. All three are elite athletes. One saddle does not fit all of them. Similarly, Nelly Korda, LPGA golfer is 5’10”, approximately 115 lbs. Caitlin Clark, guard, Indiana Fever is 6’0”, 165 lbs and Serena Williams, professional tennis player and mother of two children is 5’9”, 155 lbs. None of these incredible athletes is the recreational rider doing RAGBRAI. The design brief writes itself.

Exhibit C: Georgena Terry

Enter Georgena Terry, a mechanical engineer and former Xerox employee who founded Terry Precision Bicycles for Women in 1985, the first women-specific cycling company. [7] Fairly quickly, she noticed the women buying her frames were also complaining about saddles. She went to work. In 1991, she filed for her first patent for a women’s specific saddle, which was awarded in 1992. [8]

The forensic question: what decision did she make that was right when the default was to do nothing? She listened to her customers, the overwhelming majority of whom were women. Then, she applied an engineering mindset to a problem the industry hadn’t recognized because they were focused on elite riders. The people with the problem -- in this case, women, but more broadly non-elite riders -- weren’t in the rooms when design and innovation needs were discussed and identified.

Her design featured the now common cut-away centre, as a mechanism for relieving pressure on the soft tissue often crushed by bike saddles. This feature, originally designed by Fred Black in 1932, had largely been forgotten. Terry brought it back, specifically focused on the needs of female riders. In doing so, she became an industry catalyst, changing cycling for the millions of us who will never contend for the maillot jaune (yellow jersey). She was inducted into the US Bicycle Hall of Fame in 2024 for her contributions to the sport. [9]

Exhibit D: The Industry Catches Up

In the decade after Terry’s saddle launched, manufacturers began paying closer attention to the bike saddle and the realization that one size will never fit all began to take hold.

In 2002, SQlab founder Tobias Hild developed a systematic approach to measuring sit bone width to determine ideal saddle size after suffering an injury that left him experiencing numbness when he returned to riding. He connected with a urologist who specialized in cycling ergonomics, and together they identified sit bone distance as the decisive criterion for saddle selection. [10] The insight: sit bone width varies more within genders than between them. A man with wide sit bones needs a wider saddle than a man with narrow ones. Same for women. The solution isn’t a men’s saddle and a women’s saddle. It’s a range of widths, fitted to the individual.

Today, all SQlab saddles are available in up to five different widths. Specialized offers 143/155/168mm width options. Fizik has a saddle finder keyed to sit bone measurement. WTB, Ergon, Selle Italia, and even Brooks, all offer multi-width options or saddle ranges. [11] What Terry started as a women’s solution became the industry’s universal standard because the underlying problem was never about gender -- it was about bodies being different from each other.

The Verdict

The saddle is the contact point where cycling either claims you or loses you. Brooks solved it one way, through material honesty and time. Terry solved it another way, through the decision to ask who was being left out. The sit bone revolution solved it a third way, through measurement and range. Three different answers to the same question: does this saddle fit the actual human body of the actual rider?

When the answer is yes, people ride. They ride centuries and Gran Fondos and RAGBRAI. They ride for decades. They ride when they can no longer run or enjoy high-impact sports. It extends their athletic life, improves their fitness and keeps people healthier longer. And, yes, they ride with a leather saddle on a carbon bike and they wouldn’t trade it for the world.

The B17 has been in production since 1888. It will be in production long after the carbon frame it sits on has been recycled. Some problems, solved correctly the first time, don’t need to be solved again. And sometimes, a problem can be solved in multiple ways to best serve our highly variable human population.

Now, put away the screen and go for a ride.

“The Post-Laundry Bra Pad Shuffle” as it is sometimes called is one of the most unnecessary and irritating modern day laundry chores. It’s a chore within a chore. I am referring to the common ritual of reaching through the small side openings in a sports bra or swimsuit top and untwisting the removable foam padded insert after washing. This time consuming act of deconstructing the “Bra-Pad Origami” that has transpired in the wash cycle is a design crime that robs (mostly) women of valuable time and energy. This is the evidence of a crime of indifference and gendered design failure with no male equivalent. Active, athletic women do this every wash cycle - silently, but with frustration and an equal measure of acceptance. That normalization of unnecessary work is also a crime.

Exhibit A

How did we get here?

To deconstruct this crime and begin our forensic audit, I wanted to examine whether the brief was where this failure began. I started by asking ChatGPT, Copilot, Perplexity and Claude to “Create a design brief for a sports bra.” At a high level, one could argue the briefs all met the reasonable person (or should I say ‘designer’?) standard, each delivering an acceptable design brief. I then asked Claude to consolidate the briefs and deliver an optimized brief, presented here as evidence for your consideration.

Exhibit B: Cross-Activity Sports Bra Design Brief

Design Challenge

Design a high-performance cross-activity sports bra that delivers reliable support, sweat management, and all-day comfort across the full range of athletic movement — from lifting and HIIT to running, cycling, and yoga. The garment should perform without compromise and look as considered on the gym floor as it does under a jacket at brunch afterward. The best sports bra is one she forgets she is wearing — secure, comfortable, and supportive without calling attention to itself.

Design Principles

• Control motion across impact levels without restricting breathing or shoulder movement.

• Distribute pressure evenly; eliminate points of friction, digging, and irritation.

• Hybrid support construction (compression + encapsulation) preferred for broader size coverage.

• Accommodate diverse body shapes, torso lengths, and bust profiles.

• Easy on and off — including post-workout, when sweaty and fatigued.

• Aesthetic choices must serve function. Clean lines. Nothing gratuitous.

• Build for longevity: shape, support, and finish must survive repeated use and laundering.

Target User

Active women who train across disciplines and refuse to choose between performance and how they look doing it. She moves between the weight floor, the spin bike, and the yoga mat — sometimes in the same session. She values function first, but expects the garment to meet her everywhere she goes.

Success Criteria

• Controlled bounce and lateral movement at medium-to-high impact.

• Comfortable, distraction-free wear for 60+ minutes of continuous activity.

• Positive fit feedback across diverse body types in wear testing.

• Easy on/off confirmed without excessive effort, post-workout included.

• Structural integrity and finish quality maintained through repeated laundering.

• Wearer forgets she is wearing it during activity — and chooses to keep it on after.

The brief isn’t the villain here. That’s what makes the indifference harder to excuse. The breach was in the standard of care with which the design team responded to the brief. You see, the designers likely met the success criteria outlined in the brief. Where they failed and their indifference took root was not in their understanding of the brief or the business case. Nor was it in their anthropometric work, their understanding of fashion or their understanding of materials and manufacturing processes. It was in their willingness to engage in the user journey. They succeeded right up to the point where the tired, sweaty, post-workout woman came home and had to wrestle and contort her way out of a wet bra that does not want to be wrestled with in the first place, (a crime for another time.)

This is the pattern, every time: Pre-workout, the woman chooses the sports bra, puts it on, adjusts the padding and straps before she even starts to think about training. She trains. She peels the bra off wet. She washes it. She reaches into the pocket to play this ridiculous bra pad origami game and the whole process repeats. Every single time.

Were there no women in the room to raise this flag during the design process? If there were, nobody listened.

This cycle is completely avoidable. Removable pads are the cheapest way to make a sports bra. Better methods that avoid this laundry rage cycle use sewn-in cups, bonded cups, one-piece molded inserts and “stay-put” padding designs. Let’s be clear, sports bras retail from $15 - $120+. The average price is likely in the $40 - $60 range with a landed product cost (materials, manufacturing, shipping, packaging, etc.) of around $16. Manufacturer gross margins are often in the 50% - 70% range and retailer gross margins are often in the 45% - 60% range. Better options for the women wearing these products have negligible margin impact. Our suffering is a feature of the bottom line, not a bug.

This is not a small thing. Every unnecessary friction between a woman and her workout is a negotiation she shouldn’t have to have. She’s already doing the hard part.

The evidence is clear. Better solutions exist. Indifference and the bottom line won. Women pay the price.

Ps. On average, I work out 5 days per week, so that’s 5 sports bras per week. If it takes me 30 seconds to do the shuffle, that’s 2.5 minutes per week. It doesn’t seem like much, but it adds up. Over the course of the last decade, it means I’ve spent 22 hours doing The Post-Laundry Bra Pad Shuffle. Over the course of my life (thus far), I’m closing in on 4 DAYS of this nonsense.

References:

Bra Manufacturing Cost in 2026: Complete Pricing Breakdown for Apparel BrandsHow much does a sports bra cost?E-Commerce Profit Guide for Fitness & Sports Stores (2026)

The first product I wish to put forward in this light is the LAMY Safari Fountain Pen. As a big proponent of hand-written notes, sketching and sketch-noting, honoring a pen is an obvious starting point. I’ve also chosen it because the mainstream design press has largely overlooked this remarkable product.

This iconic pen, first introduced as a fountain pen at the Frankfurt Fair in 1980, and now available as a fountain pen, ballpoint pen, rollerball pen and mechanical pencil, is sold in more than 80 countries with over 15,000 points of sale worldwide. Although exact sales volumes are not published, LAMY acknowledges that after almost 50 years on the market, it is still consistently the best-selling pen in the LAMY range. If that isn’t success, then we have very different ideas about what genuine success looks like.

Exhibit A: The Brief

Interested in expanding into the student market (think adolescents and high-schoolers), LAMY worked with designer Wolfgang Fabian and the Mannheim Development Group under the direction of Professor Bernd Spiegel to develop a fountain pen for students.[1,2] This was a revolutionary concept for the time (late 70s). Fountain pens were either cheap garbage or they were expensive, delicate instruments of sophistication, a not-so-subtle flex. There was nothing in the middle. The very idea of designing a fountain pen that would live in the rough and tumble world of a teenager’s backpack was anathema. As it turns out, it was also an ideal design challenge and an incredible market opportunity.

Professor Spiegel, who revolutionized market research by treating product development as a branch of consumer and behavioral psychology [3], approached this project first, by kicking indifference to the curb. He saw this as a serious design challenge, worthy of great thinking and understanding. He opted to run a large study on adolescents to understand their needs, habits, fears, behaviors and how they used pens. The research was expensive and time-consuming. It was also the decision that made everything else possible. In fact, the research led to the understanding that a successful candidate pen would need to be strong, durable and tough, able to withstand being dropped (repeatedly) and being gripped tightly during use.

The brief was to make a fountain pen for students that was durable, affordable, and teachable. No assumptions were made by the design team. The users weren’t ignored. They were embraced, studied, understood and the pen was created to fulfill their needs - as good design should. Every decision made by Wolfgang Fabian and his collaborators — the ergonomic grip that encourages proper handwriting technique, the flat sides, the swappable nib, the spring clip — answers that brief directly. What makes it special is its sheer simplicity: no clever mechanics, precious metals, or fancy trim. And, in so elegantly solving this problem for students, they also solved issues adults were facing as well. [4]

Exhibit B: The Decisions

Every design decision is defensible. The durable ABS plastic body, a “downgrade” in the category, offered strength and resistance when the pen was dropped or gripped tightly. The ergonomic triangular grip ensures comfortable writing and makes it accessible for beginners and seasoned users alike. The nib can be quickly and independently removed and replaced - even by a 10-yr old, making it cost-effective for everyday users. The “form follows function” philosophy means it has no trendy frills, giving it a timeless look that still feels modern today. This product honors Dieter Rams’ Ten Principles of Good Design [5], the gold standard for design. In particular, it is useful, honest, unobtrusive, long-lasting, and thorough to the last detail. In a category defined by ornamentation and status signaling, Fabian and LAMY decided to solve for the user instead. The Safari has been rewarding that decision for 46 years.

Exhibit C: The Record

The Safari Pen was recognized as an “outstanding design achievement” when it won the iF Design Award in 1994. That is the only “official” design accolade this pen has garnered. Going down the rabbit hole of pen enthusiast sites, however, will show the impact and strong positive sentiment this pen has earned over the years. It is a consistent favorite, determined by the people who are passionate about pens and qualified to identify excellence in the category.

Over the years, the pen has evolved ever so slightly. According to Stationery Wiki, “In the mid-to-late 1990s, the pen underwent its most significant revision: the part count dropped from 14 to 11, eliminating costly machined components without touching the clip, nib, or feed. The retail price has changed very little in nearly 30 years.” [6]

The Verdict:

The record has been building for 46 years. Here is the finding of fact. Researching this article, I sought detractions and negative reviews. This was the worst I could find:
The white model looks like a pregnancy test. “The window and the plain white appearance does make it look an awful lot like a pregnancy test.” Not a design indictment, but it landed in print and it’s not entirely wrong [7]. Then, again, this pen pre-dates modern pregnancy tests, so the evidence would suggest it is the test that resembles the pen and not the other way around.

I own two of the rollerball varieties, green and umbra. Pulling the green one out of my bag before a meeting never fails to elicit delight for me because I know that no matter what happens in the meeting, I will enjoy the note taking that is forthcoming.

LAMY had a vision. They hired professionals and empowered them to do the work. The designers didn’t phone it in or do a half-ass job. They solved for the user in a category that never believed anything needed to be solved. Forty-six years later, the record shows it was the right call.

FIFA did not get the memo.

Exhibit A: Origin Story

In June 2013 -- two months after two pressure-cooker bombs killed three people and injured hundreds more at the Boston Marathon finish line -- the NFL announced what it called the All Clear policy. [1] Every bag entering every stadium in the league would henceforth be transparent: clear plastic, vinyl, or PVC, no larger than 12 inches by 6 inches by 12 inches. The stated rationale was security and speed: transparent bags meant faster visual inspection at the gate, shorter lines, and a harder path for anyone trying to conceal a weapon or explosive in a crowd of 70,000 people. It was, at its core, a service design problem. It was not treated as one.

The policy was designed for a specific context: American football, played in American stadiums, attended by an American audience already accustomed to post-9/11 security theater in airports, federal buildings, and public events. The user the NFL had in mind was someone who arrives at a game, gets screened, watches football, and goes home. The design team -- if there was one -- modeled that person and built a rule around them.

What followed was policy as contagion. The spread was informal, uncoordinated, and largely unexamined. By 2016, nearly half of Power 5 university athletic programs had adopted similar requirements for college football and basketball. [2] MLB stadiums followed on a team-by-team basis through the mid-to-late 2010s, with no league-wide mandate -- each franchise simply looked at what the NFL had done and replicated it. [3] Concert venues fell in line by deferring to the stadiums they shared. The logic was circular and self-reinforcing: the NFL did it, so it must be the standard; the standard exists, so we adopt it. Nobody went back to the original problem statement. Nobody asked whether the solution that worked for a Patriots game in Foxborough was the right solution for every large gathering of human beings on American soil.

And then FIFA arrived.

The FIFA Club World Cup 2025, which served as the operational template for the 2026 World Cup, inherited the NFL’s framework wholesale. The dimensions are NFL dimensions. The material requirements are NFL requirements. The clutch exception (4.5 inches by 6.5 inches) is the NFL exception. A policy built thirteen years ago for American football fans at Gillette Stadium is now the governing document for a woman who flew from São Paulo to East Rutherford, New Jersey to watch her national team. [4]

Exhibit B: The Policy

The FIFA Club World Cup 2025 bag policy reads as follows: [5]

“Bags must be made of clear plastic, vinyl, or PVC and no larger than 12in x 6in x 12in (30cm x 15cm x 30cm).”

“Non-clear clutches or wallets about the size of a hand are permitted, as long as they’re no bigger than 4.5in x 6.5in (11cm x 16.5cm).”

“Bags that do not meet size or material requirements may be permitted only for medical reasons.”

“Camera bags, covers, and similar items are generally prohibited, unless they adhere to our clear bag policy.”

That is the complete policy. Two permitted bag types, one medical carve-out, and one named use case: camera bags. FIFA thought to write specific language for photographers. It did not write a single word about menstrual products.

Notice also what the metric equivalents reveal. The primary specification is Imperial -- inches first, centimeters in parentheses, an afterthought. FIFA is a Swiss-based organization governing a tournament with 48 participating nations, the vast majority of whom use the metric system as their primary standard of measurement. The frame of reference was never adjusted. The world was expected to fit the template.

And before anyone reaches for the medical exception: menstruation is not a medical condition. It is a normal, healthy biological process -- as routine as digestion, as universal among half the population as sleep. Classifying it as a medical need in order to access an accommodation is not a solution. It is a category error dressed up as one.

Exhibit C: Who Carries What

I want to be precise here, because precision is how we separate design failure from personal grievance.

Men’s clothing has pockets. Functional pockets -- pockets that hold phones, wallets, keys, medications, and anything else a person needs to move through the world for six to eight hours. When a man passes through stadium security, nothing about his body’s functioning is visible to anyone around him. He empties his pockets into a bin, walks through the detector, collects his things, and proceeds.

Women’s clothing frequently does not have functional pockets. This is not a secret. It is a documented, extensively mocked, widely lamented feature of the apparel industry. And it means that women carry bags -- not as a fashion accessory, but as infrastructure.

Inside those bags, on any given day, for a percentage of women attending any given event, are menstrual products. Pads. Tampons. Perhaps both, for an event that runs six to eight hours from the time they leave their home or hotel until they return after the match. These are not optional items. They are not decorative. They are the functional equivalent of the bathroom stall door -- a private accommodation for a biological reality that is nobody’s business but the person managing it.

The clear bag policy makes them everyone’s business.

As one commentator put it: bag restrictions apply to everyone, but there are no restrictions on pockets. That is a genuine disparate impact, even if the policy is written as gender-neutral. [6] Women, in other words, are being penalized for a security problem they did not create. [7]

Exhibit D: What This Actually Costs

I am not speaking theoretically. I am speaking from the specific, personal calculation that a woman in her teens, 20s, 30s or 40s makes when she learns the venue requires a clear bag.

It is not shame. I want to be precise about that too, because the counterargument will come: we should normalize periods, visibility reduces stigma, there is nothing to be embarrassed about. That argument misses the point entirely. Normalization does not require forced disclosure. I do not have to display my blood pressure medication to prove I am not ashamed of hypertension. I do not have to announce my bathroom schedule to demonstrate comfort with bodily functions. Privacy is not the same thing as shame. Privacy is the door on the stall.

The calculation goes like this: a mixed group of friends, a stadium event, a clear bag, visible products that communicate -- to everyone within eyeline, including people she has not chosen and cannot control -- that she is menstruating, what product she uses, what size, and how many she needs for the duration. She has not consented to any of that disclosure. She is simply trying to attend a very expensive event to cheer on her national team.

Some women buy the clear bag and go. Some women rearrange their supplies around a 4.5-by-6.5-inch clutch that technically fits two tampons and nothing else. Some women just stay home.

That last group is invisible in the data. Nobody counts the women who did not come. Nobody at the NFL in 2013, nobody at FIFA in 2025, sat in the policy design meeting and asked: what does this require of someone who is menstruating? The question was never on the table because the people at the table did not need to ask it. They did not conduct research with real users. They did not run surveys or fan intercepts. They solved the problem they saw and shipped the policy to the world.

Exhibit E: The World Cup

In June 2026, the FIFA World Cup came to the United States, Mexico, and Canada. Sixteen host stadiums. Forty-eight nations. More than 1.2 million international visitors, according to Tourism Economics [8] -- a significant portion of them women, many traveling from countries where this policy does not exist and where nothing in their lived fan experience has prepared them for it.

Consider a woman flying in from Germany for a match. At Allianz Arena in Munich, bag policy is a size restriction: your bag cannot exceed roughly A4 dimensions. [9] It does not have to be transparent. You carry what you need, it gets checked at the gate, and you proceed. She has attended matches her entire life under this model. She has never owned a clear plastic purse. Why would she? Nothing about the FIFA fan materials she received prepared her for what she would encounter at the gate. Let’s be honest, even if it was mentioned, nobody reads the fine print. They confirm they have the tickets and they know where to be when.

She arrives at MetLife Stadium with a normal bag. She is told it is not permitted. She has a choice: return to her hotel, abandon the bag, or transfer her belongings into a clear alternative she does not own and was not warned she would need.

If she is menstruating, she makes that transfer in public. At the gate. In front of strangers.

On the same FIFA Fan Safety page that houses the bag policy, FIFA states that it “supports the IGLFA” -- the International Gay and Lesbian Football Association -- and calls on fans to “play your role in making the FIFA World Cup 2026 an inclusive experience for all.” The FIFA World Cup 2026 Human Rights Framework, published in 2024, requires each of the 16 host cities to develop a human rights action plan covering discrimination, workers’ rights, and human trafficking. FIFA’s Fan Safety page also links to a Fans’ Human Rights Guide and a Global Guide to Discriminatory Practices in Football. [10] [11]

It did not publish a single line addressing what happens when a woman’s bag is non-compliant because her body required her to pack for her biology.

The silence is the design decision. That is how Indifference works -- not through intent, but through the consistent failure to ask who else is in the room.

The Verdict

The clear bag policy is not a conspiracy against women. Nobody designed it to humiliate anyone. It was designed by people solving a perceived security problem after a national tragedy, and those people modeled one kind of user -- a user whose essentials fit in a pocket, whose biology requires no management, whose participation in public life generates no logistical overhead that the policy might inadvertently expose.

That user is not a universal user. He is a default.

When design treats the default as universal, everyone who falls outside the default pays the cost. In this case, the cost is specific and measurable: the forced public disclosure of a private biological function, with no warning, no accommodation, no exception language, and no apparent awareness that the question was ever worth asking.

Camera bags got named. Vague medical needs got a sentence. Women managing their periods got nothing.

That is not an oversight. At this point -- thirteen years after the NFL introduced this policy, with the entire world’s fan base now arriving at American stadiums -- it is a choice. A choice made by not choosing. Which is still a choice.

In the meantime, there is a practical workaround that requires nothing from FIFA, nothing from the NFL, and nothing from the venues. Men have pockets. Capacious, unrestricted, completely private pockets. Columnist Jonathan Stinson suggested that men should start carrying menstrual products for the women in their lives -- load them into those pockets and dare anyone in the security queue to comment when you pull them out with your wallet. [12] It’s a small act of solidarity. It’s also an accidental illustration of exactly what’s wrong: the people the policy forgot have to rely on the people the policy was built for to get inside.

The bathroom has a door. The policy doesn’t.

I am closing in on 2 million miles of air travel with Delta and SkyTeam partners. According to Flighty, since 2007, I have spent 102.2 days in the air and traveled through 97 airports. Over the last 19 years, my top route is the round trip between my home airport, Atlanta (ATL) and Philadelphia (PHL), having made this trip over 100 times, almost always for business. In other words, I know Terminal D at PHL very well. While I occasionally rent a car, it’s far more likely that I use a rideshare service when I’m in the Philly area. The airport has experimented with having drivers pick up passengers in at least four different locations that I can quickly recall. Inside the terminal, at the bottom of the escalator in Terminal D, a sign points left for rideshare and right for passenger pickup, hotel and rental car shuttles. For quite a while, the rideshare location has been consistently down the elevator, left, and left again — denoted by a sign with a large number 7 prominently displayed.

As a frequent traveler, I have my hacks. For the best PHL experience, especially in winter, I order my rideshare before I exit the secured portion of the airport so I’m not standing in the cold longer than necessary. So imagine my surprise on a cold February night at 10:38 PM when the Uber app directed me to Terminal D, Red Zone.

Where is the Red Zone?

I follow the signage at the bottom of the escalator — “Rideshare,” pointing left — and exit the building. Nothing beyond that point directs me anywhere. No Red Zone. My car is less than two minutes away. I’m tired. It’s been a long day of meetings and travel and I have an early start the next morning. I wander toward the location shown in the Uber app and find myself at the familiar number 7 location. The large signs still read “Ride App & Premium Ride App,” so this is probably right. But I still don’t see the Red Zone.

Just as my car arrives, I spot it: a small red band, just above waist height, wrapped around one of the columns of the transit shelter. Roughly two inches tall. It says “Red.” A nearby column has a green band. It says “Green.”

I’m too tired for this.

I found the Red Zone and the Green Zone — no thanks to the airport. On the ride into the city, I wondered whether I’d simply missed something. I made a note to check on my next trip. It turns out I had not missed any signage, because there was no signage to miss.

Here is the design failure, stated plainly: PHL implemented a color-coded zone system for rideshare pickups — a system their operational partners, Uber and Lyft, had incorporated into their apps. The apps were updated. The passenger environment was not. The airport designed a system for their operational partners and forgot to design it for the people standing in the cold at 10:38 PM. They introduced a new wayfinding layer without the wayfinding infrastructure to support it, then handed the resulting confusion to their most exhausted customers.

This is not a minor oversight. It is a textbook use error.

The resources to get this right are not obscure. The FAA publishes Advisory Circular 150/5360-12F [2], which provides comprehensive guidance on airport terminal and landside wayfinding, signing, and graphics. It is built on the Airport Cooperative Research Program (ACRP) Report 52, Wayfinding and Signing Guidelines for Airport Terminals and Landside [3] — the definitive industry resource on this subject. The Society for Experiential Graphic Design (SEGD) [4], with 46 chapters across 36 countries including one in Philadelphia, exists precisely to solve problems like this one. The playbook and the talent were available. Neither were consulted.

ACRP Report 52 is unambiguous about what effective signage requires. Section 2.2.6 identifies five criteria every sign must meet: it must be conspicuous, with sufficient color contrast to be detected from its surroundings; comprehensible, meaning its intent is clear to users who were not involved in its design; legible at the distance from which a user is likely to first look for it; and positioned at decision points — the moments where a traveler must choose a path. A two-inch color band at waist height on a structural column fails on all four counts. The report also explicitly names “familiar passengers when changes are made” as a distinct user category requiring systematic consideration. [5] I am that user. The color zone rollout was that change. By the standard of the report PHL itself helped validate, I was owed clear guidance at the moment of decision.

The consequences of this kind of failure are also well-documented. Report 52 states directly: “Violation of user expectations will result in losing passenger confidence in the airport’s wayfinding system, which in turn will create a negative perception.” [6] The report further warns that color coding, if not carefully implemented, “can make a basically simple terminal more confusing and complex.” [7] PHL deployed color coding without the signage infrastructure to support it and proved the warning correct.

The richest irony is this: PHL is not an anonymous offender. The airport is held up by name in the ACRP’s own practitioner literature as a model implementation of Report 52 — praised specifically for its “high contrast and legibility, even illumination in all conditions.” [8] The model airport violated the model. The case study airport became the cautionary tale.

In place of confidence and ease, PHL delivered friction and panic. Rather than treating arriving passengers (like me) as the beneficiaries of a better system, the airport treated them as an afterthought to their own operational improvements.

The user was not the error. The design was.

References: [1] 14 CFR § 121.471 — https://www.ecfr.gov/current/title-14/chapter-I/subchapter-G/part-121/subpart-Q/section-121.471 [2] FAA Advisory Circular 150/5360-12F [3] ACRP Report 52 — https://crp.trb.org/acrpwebresource2/acrp-report-52-wayfinding-and-signing-guidelines-for-airport-terminals-and-landside/ [4] SEGD — https://segd.org/ [5] ACRP Report 52, Section 2.2.4 [6] ACRP Report 52, Section 2.3.3 [7] ACRP Report 52, Section 3.6.5 [8] ACRP Impacts on Practice, IOP 020 (September 2015)

I’m sorry, what? There are 7 button-like objects in a vertical column. At first glance, they all appear to be pressable. Two have red bezels, drawing the eye to indicate priority actions - or maybe these are indicators not to touch? Three of the “buttons” require keys. Two have etched in, red lettering. Three have printing on the bezels. And, then, there is the otherwise unmarked button at the very bottom that the hotelier has applied a press-on sticker that says “PRESS FOR ELEVATOR” along with two arrow stickers pointing to the button. The arrows aren’t the same shape or color. And you have wasted 30 seconds of your life that you will never get back just trying to get to your hotel room. This is not an anomaly. This is a pattern.

The evidence suggests it is easier to design a high-complication mechanical watch with 500 - 1,000 individual components that fits on your wrist and is powered by nothing more than a coiled spring.

Six buttons for four floors, set in a grid that, at first glance appears as though it might be 15 buttons. Why is the emergency call button illuminated as a default setting? And the icon above the buttons is a wireless Wifi symbol. Are we anticipating needing Wifi to travel between 4 floors? If so, the control panel is the least of the issues with this elevator.

Entrance to the building lobby located on the P4 level. Except there is no button labled “P4”. We have P1 - P3 and P5 - P7. There is a “star 4”. In all likelihood, most users will be able to navigate this, but was the pattern disruption necessary? The button could have been labeled “P4” to match the signage, thus reducing the cognitive load and dissonance for the user.

We have mastered the mechanical watch, but not the elevator.

This elevator only travels between 3 floors, but it might require five keys and a key card to get where you are going. It’s very unclear. The vertical 1, 2, 3 buttons are duplicated as a horizontal row that may also be buttons. They appear to be the same size and shape as the buttons below. Or they could be floor indicators. The evidence is inconclusive.

The more you look at this panel, the more confusing it becomes with two door open buttons and two door closed buttons and a button that is totally unmarked. Nobody labeled it. Nobody tested whether you’d press it. Nobody cared.

I. Just. Want. To. Go. To. My. Room. Why? Who thought this was a good idea? How did this pass quality control? Could the P’s be grouped together? Could the Rs be grouped together? Could the 1s be grouped on a horizontal line with the 2s above it and the 3s above them . . . ?

Are you fucking kidding me?? It would seem this elevator goes up, down, left and right and has floors with intuitive indicators like C, CF, and HL. CF is sometimes referred to as “4” apparently. And L and 6 are the same floor. Did this control panel spring forth from the seventh circle of Hell?

The elevator control panel is not a complex design problem. It is a finite one: a box, two directions, a countable number of floors. What you have just seen is not the result of a hard problem poorly solved. It is the result of no one solving it at all. That is Indifference. And it is the most common villain in the designed world.

When Elon Musk announced the Cybertruck in 2019, he hailed it as a cyberpunk vehicle inspired by “Blade Runner” and “The Spy Who Loved Me.” The prototype he drove on stage didn’t get the memo. It’s polarizing — people love it, hate it, love to hate it — and it deserves credit for attempting to reimagine what a modern passenger truck might be. The people who immediately put money down to reserve a Cybertruck, and who later took delivery, believed in the vision and in Musk’s ability to deliver against it.

The first Cybertrucks were delivered on November 30, 2023.

Through early 2026, fewer than 70K Cybertrucks have been sold. In Q4 2025, 20% of those sold in the U.S. went to other Musk-owned companies, making him his own best customer [1].

The Cybertruck can’t be sold in the European Union. The EU determined its exterior design imperils pedestrians. EU rules require vehicle exteriors to be designed to minimize damage to pedestrians, cyclists, and motorcyclists during a collision. The Cybertruck’s rigid steel body has sharp edges that fail to meet these requirements. The high-strength steel and lack of crumple zones to absorb energy during an impact add additional peril for other road users. These are not regulatory technicalities. These are findings. The case for a design crime has been made by people whose job it is to make that case.

“Good design is thorough down to the last detail. Nothing must be arbitrary or left to chance. Care and accuracy in the design process show respect towards the consumer.” — Dieter Rams

The stainless steel body panels are cut by a 5-axis laser, then scored and bent, origami style, to form the body. This is interesting and very cool. The problem is a lack of finishing on the exposed edges.

The jagged, raw laser cut lines are visible in multiple places around the vehicle — places you might place your hand, like the edge of the door frame or the outside of the truck bed. The edges are sharp. The internet is full of videos showing people slicing carrots, cucumbers, zucchini and bananas using the frunk lid [2]. Tesla deployed a software update (firmware version 2024.14.3) in May 2024 to address safety concerns, but the algorithm still wants the frunk to close and will apply increasing pressure to make it happen [3]. There are also videos of people shaving carrots on the edge of the door [4]. You do not want the door of your car to be sharp enough to shave a carrot. If it is, your leg will eventually be the victim. The EU’s case has been made.

Another detail lacking in care and accuracy: panel gaps. Known as “parallelism,” a key goal in automotive manufacturing is uniform gap widths across every seam. Most modern cars target around 3/16 of an inch (4-5mm). High-end luxury vehicles run tighter. Early on, Musk demanded all components be built to within a 10-micron window accuracy. “That means all part dimensions need to be to the third decimal place in millimeters, and tolerances need to be specified in single digit microns.” [5] Yet, precision panel alignment still appears to be a suggestion rather than an ironclad requirement — as though the lead manufacturing quality engineer was, in fact, a 6th grader in an industrial arts class. The craftsmanship is inexcusable at this price point. It is disrespectful to the people who believed in the vision and handed over their hard-earned money for this vehicle.

To date, there have been 11 recalls: accelerator pedals that could stick, failing windshield wiper motors, trim pieces that detach while driving, rear-view camera display failures. The most recent involves potential wheel separation due to the wrong grease and loose nuts. (By comparison, the Tesla Model S has had 39 recalls since June 2012 indicating a trend. [6])

Elon Musk has proven, repeatedly, that he can demand and achieve extraordinary precision when he decides it matters. Rocket boosters caught by mechanical chopsticks. Battery cells manufactured to tolerances measured in microns. When Tesla’s CEO wants precision, he gets precision. The Cybertruck tells you exactly what he decided didn’t matter. You. And your money.

This is evidence.

The Hyundai Venue is currently the least expensive new car sold in the U.S. at $20,550 [7]. Compared to the Cybertruck, its exterior build quality is a master class in precision. It carries the best warranty in the industry: 10-year/100,000-mile powertrain, bumper-to-bumper for 5 years/60,000 miles. Fewer than five recalls in six years on the market. Car and Driver rates it 8/10 and it retains value better than the Cybertruck. This is a car that respects the people who buy it.

When design leadership fails to respect the buyer, you end up with a Cybertruck.

This wasn’t an engineering limitation. It was a priority decision. You can catch a rocket booster with chopsticks or you can finish the edges on a $100,000 truck. Apparently, you cannot do both.

I felt like a dick for writing on the bottles. I’m also exhausted by the fact that I had to.

That’s the thing about hotel toiletries. They’re provided as a luxury. They’re experienced as a puzzle. And somewhere between the brand partnership announcement and the guest’s 6am shower, the person who actually had to use the product stopped mattering.

Hotel toiletries weren’t always an expectation. Before 1970, if you forgot your shampoo, you were simply out of luck. The Four Seasons London changed that, introducing bathroom amenities as a luxury differentiator. Mid-tier chains followed through the ‘80s as competition intensified and guest expectations hardened. Then TSA’s 3-1-1 rule in 2006 finished the job: carry-on travelers stopped packing liquid toiletries entirely, and hotels responded by expanding what they stocked. The humble shampoo bottle became, quietly, one of the most loaded brand statements in hospitality.

The mini bottle’s days were already numbered before sustainability became a marketing strategy. Clean the World founder Shawn Seipler estimates that 5 million hotel soap bars are discarded globally every day [1] — hygiene protocols require disposal regardless of whether the product was touched. The waste problem created the business case for wall-mounted dispensers: lower cost, less plastic, a sustainability story to put in the annual report. What nobody asked, in any of those transition meetings, was whether the guest could read the dispenser at 6am in a steamy shower. That question wasn’t in the brief. It never is.

Picture this: A team of toiletry brand representatives arrives at hotel corporate headquarters, meeting with representatives from multiple sub-brands. They bring out the samples and talk about how amazing and luxurious each product is, how it’s vegan, not tested on animals and free of whatever chemicals are on the bad list today. The hoteliers test the lotions. Perhaps they even test the shampoo and conditioner on their hands. Maybe a few of them take the bottles home to give the product a real test. Decisions are made. Procurement gets involved. Contracts are written. Months later, some poor schlub like me ends up naked, wet and squinting at the bottles mounted on the shower wall in a hotel, trying to figure out where to start, which one is which. And, in a fit of rage - at the shampoo brand for their poor labeling, at the hotel for creating this “luxurious” mess, and at myself for forgetting to mentally document the crime scene before taking off my glasses and hopping in the shower - we take matters into our own hands and invoke the power of the Sharpie.

“The road to hell is paved with good intentions.”
(Attributed variously to Saint Bernard of Clairvaux and Samuel Johnson)

To be clear, absolutely nobody at the hotel HQ or at the toiletry manufacturer ever set out to make the travel experience worse. They all want it to be as amazing as possible with their brand linked to the warm, fuzzy feeling of happiness they envision for the traveler. But they approach this from a procurement mindset rather than an experiential mindset. They don’t question the brief, just the quantities and the costs. They don’t interrogate the product in context. They don’t do the research with users needed to identify these issues.

Labels are only one of the issues with hotel toiletries. The scents are another nightmare for the weary traveler. Here’s a list of a few I’ve encountered along the way:

• Beach & sky blue water

• Lemon & sugar

• Green tea lemongrass

• Sicilian tangerine

• Heavenly gingerlily

• Indian cress

• Notes of aniseed, cardamom and geranium leaf

• Notes of grapefruit peel, basil and geranium

• Notes of grapefruit, tangerine peel and rum

• Desert lime, lotus leaf, lavender

• Rose & shea

• Warm Oak

• Mandarin & vetiver

• Eucalyptus & lavender

• Citrus, nutmeg & ginger

• Eucalyptus

• Lemon sage

• Rum

• Peppermint

• Cilantro

• Rosemary geranium

• Sicilian blood oranges

• Volcanic minerals & Sicilian lemons

• Coconut mango

What recipe are we following here? Not one of them says “clean.” They say “brunch,” “spa menu,” “craft cocktail bar,” and “farmer’s market.” Warm Oak says “my grandfather’s study.” Mandarin & vetiver says “department store cologne counter.” Lemon sage says “rotisserie chicken”.

Is there really a market for cilantro scented conditioner? Am I a salsa?

Who knew we wanted to bathe in (and smell like) rum, rather than drink it?

My most recent business trip involved a stay in a hotel and spa resort where Lockwood toiletries were featured. The scent was rosemary geranium - and it was a bit heavier on the rosemary than I would have liked.

When I looked this product up online, I was rewarded by seeing it featured as one of “The Best Men’s Skin Care Products Everyone Should be Using”

and one of the “32 Amazing Father’s Day Gifts That Aren’t Socks and a Tie”.

There were more headlines and lists along this same theme. So, my hotel selected toiletries “Guaranteed to please even the fussiest of Dads” for their rooms. I’m not a Dad. I’m not even a fussy Dad. I’m a middle-aged woman and I don’t want to smell like anyone’s Dad. Nor do I want to smell like a cocktail or a marinade.

Marriott. Hilton. IHG. Between them: nearly 3 million rooms, 23,000 hotels, 144 countries. That is not a supply chain constraint. That is leverage. One of you could fix this before I finish my next business trip. Until then, I’ll keep bringing my Sharpie. And I’ll keep feeling like a dick about it.

(Part 3 of a 3 part series)

Ps. If you’ve ever thought about taking the full size shampoo bottles from the hotel room, this is a great listen: Shampoo, Property Rights, and Civilization (with Anthony Gill)