There was a nurse in Scottsdale who trusted a blood test result to a machine that never really worked. She made a treatment decision based on a technological lie sold as disruptive innovation. That's not rhetorical exaggeration to open this piece — it's literally what happened at Theranos, a company built on the same mantra that today covers office walls in motivational posters everywhere: move fast, break things. The problem is that when the thing that breaks is a cancer test result, "breaking" stops being an abstract agile engineering concept and starts being real people making the wrong medical decisions. That phrase was born in a context far more harmless than most people imagine. Mark Zuckerberg coined the slogan in Facebook's early years, back when the company was essentially a college toy that had turned into a phenomenon, and the logic behind it made sense at that specific moment: better to ship an imperfect feature and learn from real usage than spend months in an approval committee debating hypotheticals. For a news feed that occasionally lagged, fine. Nobody dies because the "like" button took an extra second to load. What nobody warned the rest of Silicon Valley — and later the whole world that copied Silicon Valley without understanding the context — was that the phrase only held up within pretty specific limits. Fintech startups adopted the same slogan to handle other people's money. Digital health companies adopted it to handle diagnoses. Self-driving car makers and, more recently, generative AI labs adopted the same rush to ship systems that make decisions about résumés, loans, and court sentences. The scale of the damage changed completely, but the philosophy kept getting repeated as if it were universal. ## <br>Facebook itself gave up on it, and nobody paid attention In 2014, at a developer conference, Zuckerberg publicly announced the slogan's replacement with "move fast with stable infra." It was a quiet, almost embarrassed admission that the original version had become unsustainable for a company that was no longer a Harvard dorm room but a platform used by over a billion people. There's a certain irony to it: the company that invented the mantra was the first to realize it didn't scale, and yet the rest of the industry kept wearing the shirt for another decade, as if the correction had never happened. That says something about how slogans outlive the context that produced them. They become identity, they become corporate culture printed on a laptop sticker, and they stop getting reevaluated. And Theranos wasn't the only one to pay the extreme price of that mindset. The 737 MAX crashes, though rooted in a very different engineering culture than a startup's, were repeatedly described by investigators as the result of pressure to speed up certifications and cut validations that, at a slower pace, likely would have caught the MCAS system flaw before it killed people mid-flight. The parallel is uncomfortable because it shows that the "ship first, fix later" logic isn't exclusive to some app startup — it seeps into any place where market pressure combines with the belief that speed equals competence. There's also a cost that never makes headlines: the silent accumulation of technical debt and human debt. Every feature shipped too fast, without the testing it should have had, becomes a problem someone will have to solve later — usually with interest. Entire engineering teams now live under an endless sprint regime, each one promising to deliver faster than the last, and the burnout that generates rarely factors into anyone's praise of a product's shipping speed. Users pay a price too, in a more diffuse way: they become permanent, unpaid testers of products that should have been tested before ever reaching them. ## <br>The bill came due in the form of regulation, and it isn't cheap Maybe the most ironic effect of all is this: the philosophy that promised radical freedom, zero bureaucracy, asking no one's permission, ended up producing exactly the opposite. Every scandal driven by haste — Cambridge Analytica, massive data breaches, discriminatory credit algorithms shipped without audit — fed the pressure for laws like GDPR in Europe, and today fuels the regulatory debate around artificial intelligence in the United States and the EU alike. Companies that once asked no one for permission now have entire legal departments dedicated to complying with rules born precisely because the industry, collectively, proved over and over that it couldn't regulate itself. Short-term haste turned into long-term external control. It's not an exaggeration to say much of the bureaucracy that now frustrates startup founders was built, brick by brick, by the excesses of that very culture. That said, it would be dishonest to pretend the philosophy brought nothing good. The culture of rapid iteration, of shipping an MVP instead of a perfect plan, of launching and learning from real usage instead of imagining scenarios in a conference room, let genuinely useful products reach the market in a fraction of the time they'd take under a traditional corporate model. Much of the internet we use today, including tools that solve real problems for ordinary people, only exists because someone decided to ship something imperfect instead of waiting for a perfection that never arrives. The problem was never speed itself — it was the refusal to admit that speed has a price, and that price changes depending on what's actually at stake. What seems to be emerging now, somewhat hastily and somewhat out of necessity, is a more mature version of this conversation. Feature flags that let teams test changes with a small slice of users before rolling out to everyone. Red teams hired specifically to try to break a system before it reaches the public. Discussions about responsible design that, a decade ago, would have sounded like a waste of time in a startup meeting and today occupy entire executive roles inside tech companies. Nobody's saying we need to go back to moving slowly. The question that remains, still without an easy answer, is what exactly deserves to be broken along the way — and who pays the bill when the answer turns out wrong.