Die as Microsoft, or live to become the IBM you overthrew.

6 min readJust now

–

Press enter or click to view image in full size

I have a weird fascination with early-stage Bill Gates, after his mother convinced a pal of hers — chairman of IBM’s board of directors — to give her son the contract to provide the operating system for the new IBM PC. Gates and his pal Paul Allen tricked another programmer into selling them the rights to DOS, which they sold to IBM, setting Microsoft on the path to be one of the most profitable businesses in human history.

IBM could have made its own OS, of course. They were just afraid to, because they’d just narrowly squeaked out of a 12-year antitrust war with the Department of Justice (evocatively memorialized as “Antitrust’s Vietnam”):

https://pluralistic.net/2022/10/02/the-true-genius-of-tech-leaders/

The US government traumatized IBM so badly that they turned over their crown jewels to these two prep-school kids, who scammed a pal out of his operating system for $50k and made billions from it. Despite owing his business to IBM (or perhaps because of this fact), Gates routinely mocked IBM as a lumbering dinosaur that was headed for history’s scrapheap. He was particularly scornful of IBM’s software development methodology, which, to be fair, was pretty terrible: IBM paid programmers by the line of code. Gates called this “the race to build the world’s heaviest airplane.”

After all, judging software by lines of code is a terrible idea. To the extent that “number of lines of code” has any correlation with software quality, reliability or performance, it has a negative correlation. While it’s certainly possible to write software with too few lines of code (e.g. when instructions are stacked on a single line, obfuscating its functionality and making it hard to maintain), it’s far more common for programmers to use too many steps to solve a problem. The ideal software is just right: verbose enough to be legible to future maintainers, streamlined enough to omit redundancies.

This is broadly true of many products, and not just airplanes. Office memos should be long enough to be clear, but no longer. Home insulation should be sufficient to maintain the internal temperature, but no more.

Ironically, enterprise tech companies’ bread and butter is selling exactly this kind of qualitative measurements for bosses who want an easy, numeric way to decide which of their workers to fire, and leading the pack is Microsoft, whose flagship Office365 lets bosses assess their workers’ performance on meaningless metrics like how many words they type, ranking each worker against other workers within the division, with rival divisions and within rival firms. Yes, Microsoft actually boasts to companies about the fact that if you use their products, they will gather sensitive data about how your workers perform individually and as a team, and share than information with your competitors!

https://pluralistic.net/2020/11/25/the-peoples-amazon/#clippys-revenge

But while tech companies employed programmers to develop this kind of bossware to be used on other companies’ employees, they were loathe to apply them to their own workers. For one thing, it’s just a very stupid way to manage a workforce, as Bill Gates himself would be the first to tell you (candidly, provided he wasn’t trying to sell you an enterprise Office 365 license). For another, tech workers wouldn’t stand for it. After all, these were the “princes of labor,” each adding a million dollars or more to their boss’s bottom line, and in such scarce supply that a coder could quit a job after the morning scrum and have a new one by the pre-dinner pickleball break:

https://pluralistic.net/2025/04/27/some-animals/#are-more-equal-than-others

Tech workers mistook the fear this dynamic instilled in their bosses for respect. They thought the reason their bosses gave them free massage therapists and kombucha on tap and a gourmet cafeteria was that their bosses liked them. After all, these bosses were all techies. A coder wasn’t a worker, they were a temporarily embarrassed founder. That’s why Zuck and Sergey tuned into those engineering town hall meetings and tolerated being pelted with impertinent questions about the company’s technology and business strategy.

Actually, tech bosses didn’t like tech workers. They didn’t see them as peers. They saw them workers. Problem workers, at that. Problems to be solved.

And wouldn’t you know it, supply caught up with demand and tech companies instituted a program of mass layoffs. When Google laid off 12,000 workers (just before a $80b stock buyback that would have paid their wages for 27 years), they calmed investors by claiming that they weren’t doing this because business was bad — they were just correcting some pandemic-era overhiring. But Google didn’t just fire junior programmers — they targeted some of their most senior (and thus mouthiest and highest-paid) techies for the chop.

Today, Sergey and Zuck no longer attend engineering meetings (“Not a good use of my time” -M. Zuckerberg). Tech workers are getting laid off at the rate of naughts. And none of these bastards can shut up about how many programmers they plan on replacing with AI:

https://pluralistic.net/2025/08/05/ex-princes-of-labor/#hyper-criti-hype

And wouldn’t you know it, the shitty monitoring and ranking technology that programmers made to be used on other workers is finally being used on them:

https://www.theregister.com/2025/10/10/microsoft_copilot_viva_insights/

Naturally, the excuse is monitoring AI usage. Microsoft — along with all the other AI-peddling tech companies — keep claiming that their workers adore using AI to write software, but somehow, also have to monitor workers so they can figure out which ones to fire because they’re not using AI enough:

https://www.itpro.com/software/development/microsoft-claims-ai-is-augmenting-developers-rather-than-replacing-them

This is the “shitty technology adoption curve” in action. When you have a terrible, destructive technology, you can’t just deploy it on privileged people who get taken seriously in policy circles. You start with people at the bottom of the privilege gradient: prisoners, mental patients, asylum-seekers. Then, you work your way up the curve — kids, gig workers, blue collar workers, pink collar workers. Eventually, it comes for all of us:

https://pluralistic.net/2021/02/24/gwb-rumsfeld-monsters/#bossware

As Ed Zitron writes, tech hasn’t had a big, successful product (on the scale of, say, the browser or the smartphone) in more than a decade. Tech companies have seemingly run out of new trillion-dollar industries to spawn. Tech bosses are pulling out all the stops to make their companies seem as dynamic and profitable as they were in tech’s heyday.

Firing workers and blaming it on AI lets tech bosses transform a story that would freak out investors (“Our business is flagging and we had to fire a bunch of valuable techies”) into one that will shake loose fresh billions in capital (“Our AI product is so powerful it let us fire a zillion workers!”).

And for tech bosses, mass layoffs offer another, critical advantage: pauperizing those princes of labor, so that they can shed their company gyms and luxury commuter busses, cut wages and benefits, and generally reset the working expectations of the tech workers who sit behind a keyboard to match the expectations of tech workers who assemble iPhones, drive delivery vans, and pack boxes in warehouses.

For tech workers who currently don’t have a pee bottle or a suicide net at their job-site, it’s long past time to get over this founder-in-waiting bullshit and get organized. Recognize that you’re a worker, and that workers’ only real source of power isn’t ephemeral scarcity, it’s durable solidarity:

https://techworkerscoalition.org/