There is a pattern to leaving jobs. It starts the same way every time: you realize the system doesn't work, and you don't believe anyone is going to fix it. You’re going to leave.

What happens next separates people.

Some people leave badly. They have months of notice, and they check out immediately. They do the minimum. They let things slide. They withhold knowledge. They delete their notes. They leave the codebase in the state they found it, or worse. They tell their colleagues what they really think, all of it, with maximum damage and minimum mercy. They vanish on their last day and feel satisfied that they got the final word.

These people feel like they’ve won something. They haven’t.

The other way is harder. You know the system is broken, you know you're going to leave, and you do your job anyway. You document what you know. You leave the code in better shape than you found it. You finish the work in progress if you can, or hand it off if you can't. You take the colleague who's been difficult and leave them set up to succeed without you. You have the conversations that matter, and you don’t have the ones that don’t. You’re generous on your way out.

This costs something. It costs time you could spend job-hunting. It costs emotional energy you could spend being angry. It costs the small satisfactions of leaving land mines and letting other people step on them. It costs the convenience of just walking away.

But it buys you something too: it buys you optionality. It buys you the ability to return if circumstances change. It buys you recommendations from people who actually know your work. It buys you the credibility to say “I tried my best to make that work,” and to have people believe you.

In research, this is the discipline of the reproducible experiment. You document your method not because it's easy, but because someone else needs to be able to verify what you did. You leave your harness in a state someone can build on. You don't hide the failed approaches; you document them, because the next person might learn from exactly the thing that didn't work. The people who do this become the people whose work outlasts them. The people who don't become the ones whose work dies the day they leave.

The sunk cost fallacy says: “I've already invested this, so I might as well leave the damage.” Integrity says: “I'm leaving anyway, so I might as well leave well.”

The real problem with leaving badly isn't moral. It's practical. The world is smaller than you think. The person you decided was useless turns out to be brilliant at a different organization. The team you sabotaged gets better without you and then hires your next boss. The code you left broken costs the company thousands, and that story finds you. The recommendations you didn't get from people you could have left well with—those cost you actual money on your next contract.

But there's also something deeper. How you leave is how you say “here is what I value.” If you value only your own vindication, you leave badly and spend years looking over your shoulder. If you value the system continuing to work, even after you're gone, you leave well and you don't.

The pattern shows up everywhere. In open source, the maintainers who step back gracefully hand off to people who care and whose code thrives. The ones who sabotage their own projects on the way out are remembered, but not kindly. In ventures, the founders who leave their teams set up to succeed raise the credibility of everyone they'll ever work with again. The ones who extract maximum value and leave destruction behind spend the rest of their careers explaining why people don't trust them.

None of this means you should stay in a broken system and pretend it's not broken. Leave when you need to leave. But leave with the understanding that you're about to become a reference for everyone you just worked with. Leave knowing that the code you commit is the last thing you'll be associated with that place. Leave knowing that the person you help on your last week will remember it for years.

The hard part isn't the leaving. It's maintaining integrity under exit pressure. It's having the discipline to do the right thing when nobody would ever know if you didn't. It's choosing to leave well even when leaving badly would feel so much better.

That's the difference. That's the pattern. That's what separates the people who burn their ships on the way out from the people who leave them seaworthy for whoever comes next.