Toxic Coworkers Don't Kill Teams. Weak Managers Do.

Everyone's arguing about the wrong thing

Some ex-SpaceX engineer posted a thread calling the culture 'toxic.' Gwynne Shotwell responded. Twitter lost its mind. Hot takes everywhere. And every single one of them missed the actual problem.

Here's the thing: 'toxic' is a useless word. It's what people say when they don't want to do the hard work of diagnosing what's actually broken.

I watched this play out in real time

2021. My team at a Series D fintech, about 600 engineers, $180M ARR. We had a principal engineer I'll call Derek. Brilliant. Genuinely brilliant. Also a nightmare to work with. Interrupted people constantly. Dismissed ideas in design docs with one-word comments like 'no' and 'lol.' Made junior engineers cry. Twice, that I know of.

HR called him toxic. Three teammates filed complaints. Leadership did nothing for eight months because Derek owned the payment routing layer and nobody else understood it. Single point of failure. Human edition.

Derek didn't kill the team. The eight months of doing nothing killed the team. Four engineers left. We lost $340k in recruiting costs replacing them. The institutional knowledge that walked out the door took us 14 months to rebuild. I'm not estimating that number. I counted.

The actual failure mode nobody talks about

When companies label someone 'toxic' and do nothing, they're making a calculated bet. They're betting the cost of keeping that person is lower than the cost of losing them. Sometimes they're right. Usually they're catastrophically wrong.

Don't do this. The math never works out the way managers think it will.

The real problem isn't the difficult person. It's the system that made them untouchable. At SpaceX, at Netflix, at Google, every place I've worked, the 'toxic' engineer is almost always someone leadership decided was too valuable to manage properly. That's a structural failure. Not a personality failure.

What Shotwell got wrong in her response

She defended the culture by pointing at outcomes. Rockets land. Satellites deploy. Results speak for themselves. I've heard this exact argument used to defend every bad engineering culture I've ever seen.

Results don't tell you what it cost to get there. They don't tell you how many good engineers quietly quit and went to Anthropic. They don't show up in your launch metrics when someone doesn't raise a safety concern because they've learned that raising concerns gets you labeled 'not a team player.'

Silent failures don't page. That's what makes them dangerous.

The Derek story has an ending

We eventually moved Derek to a pure IC research role. No direct reports, no design reviews, no Slack access to the junior channels. Took six months to untangle the dependencies he'd intentionally made himself the center of. That part was on him. The six months it took leadership to act? That was on them.

The four engineers who left? None of them came back. Two of them now work together at a competitor.

Call it toxic if you want. I call it a management debt that compounded for eight months and then defaulted all at once. Same as technical debt. Same consequences. Nobody writes Substack threads about management debt.

They should.