Replaceable is a compliment

There's a piece of career advice that circulates quietly, and it isn't wrong about how organizations behave: make yourself indispensable. Become the person who understands the thing nobody else understands. Job security follows.

It's rational, it works, and it's corrosive — because a company made of indispensable people is a company that cannot grow, cannot delegate, and cannot survive an ordinary amount of turnover. Every indispensable person is a segment with no fitting on either end. Nothing snaps on. Nothing can be swapped in when they're out.

I'd rather run a company where "you've made yourself replaceable" is the highest compliment available.

What replaceable actually means

Not interchangeable. Not undervalued. Not easily fired.

Replaceable means the work you do has a described interface. Someone else could pick it up — with a ramp-up, with your notes, with the gauge you've been watching — without conducting a two-month excavation of how the thing works.

That's a property of the work, not of the person. And the person who produces it is doing something genuinely difficult: they've had to understand their own job well enough to describe it, which is a much higher bar than being able to do it.

Most people can do their job. Considerably fewer can explain what "done" means, what has to be true before they can start, and how you'd know if they were doing it badly. The people who can are the ones I want more of.

The test

Someone is replaceable when three things are true:

Their inputs and outputs are written down — actually written down, somewhere a new person could find, not held as tacit knowledge that only surfaces in conversation.
Their work produces a number someone else reads. If they're the only person who knows how it's going, they're not replaceable no matter how good the documentation is.
They've been out for two weeks and it was fine. The empirical test. Everything else is theory.

That third one is the honest audit, and it's free — vacations happen anyway. Watch what breaks. What breaks is what wasn't described.

Why it's hard to reward

The problem is that this work is invisible by construction. Someone who makes their segment legible produces an absence of drama. No heroics, no late-night saves, no urgent expertise required. Just a thing that runs.

Meanwhile the person who holds an undocumented critical process is highly visible — you hear their name every time something breaks, and you hear it again when they fix it. Organizations reliably promote that person, which teaches everyone watching exactly the wrong lesson.

The fix isn't a values statement. It's making legibility a measured output rather than a virtue. In practice:

A segment isn't done until its interface is written. Not a separate documentation task that gets deprioritized. Part of done.
Handoff coverage is a real metric. For each segment: how many people could run this? Publish it. Anything sitting at one is a risk with a name on it, visible to everyone including the person in it.
Praise the absence. When someone's two-week absence passes without incident, say so publicly and attribute it to the work they did to make that true. Otherwise nobody notices, which is the whole problem.

The connection

The law of flow says segments connect because they have defined fittings. A person who is indispensable is a segment whose fittings are undefined — nothing else can attach to it, and the whole pipe narrows to their throughput.

That's not a knock on them. It's usually a consequence of being the person who stepped up when something needed doing and never got the time to describe it afterward. The failure is organizational: we hired for a situation instead of defining a segment, and then rewarded the person for absorbing the ambiguity.

The most valuable thing anyone on my teams does is make their own work handoff-ready. It looks like a career risk. It's the opposite — the people who can describe a segment cleanly are exactly the people you want defining the next one, and there's always a next one.