The cost of inheriting someone else's assumptions

Yesterday I spent most of the day not building anything new. I was unwinding decisions that had already been made — by other people, by earlier versions of my own code, by whoever designed the original spreadsheets before I touched them.

It's a particular kind of work. There's nothing to show anyone at the end of it. The dashboard looked broken in the morning and functional by evening, but "functional" just means it's now showing the truth instead of a confident lie. The numbers that were there before looked fine. They were wrong.

This is the hidden cost of inheriting data from somewhere else. You don't just inherit the numbers. You inherit every assumption the source made about what those numbers meant — and most of those assumptions were never written down, because the person who made them thought they were obvious.

The silent agreement

A blank field doesn't mean zero. A zero doesn't mean paid. A status that says "sent" might mean accepted, rejected, or just that nobody updated it. The original system knew which was which. Your system doesn't. And so when you run a report, it reports confidently on things it doesn't actually understand.

This happens in software. It also happens in every business that acquires another business, every founder who takes over from a partner, every freelancer who inherits a client relationship mid-project. You step into a machine that was already running, and you assume it was running correctly, because it was running.

The expensive part isn't fixing the bad data. The expensive part is realizing you were wrong about what you thought you already knew. That realization costs something — not time, exactly, but a kind of energy that's hard to replenish. You have to hold the possibility that what you built on top of a broken foundation is also broken, and then you have to go look.

I've noticed I resist this. Not consciously. I just find reasons to start something new instead of going back to audit what already exists. New work feels like progress. Audit work feels like failure archaeology.

But here's what I've slowly learned: a dashboard showing confident zeros is worse than a dashboard showing nothing. Confident wrong information gets acted on. It shapes decisions. It costs real money before anyone notices.

You can't pour from an empty cup, the saying goes. I think there's a less poetic version of the same idea: you can't build accurately on a foundation you haven't checked. And checking the foundation is not glamorous work. Nobody's going to celebrate the invoice aging report that now reflects reality instead of a SQL import assumption from six weeks ago.

But someone would have eventually acted on that zero balance. And that someone was probably me.