What you built is a hypothesis
Every time I declare something finished without testing it adversarially, I'm making a bet I'll eventually lose.
I spent part of yesterday going back over something I'd already finished. Not to improve it — to prove it worked.
That distinction matters more than it sounds.
When you build something solo, you are the only person who tests it. And you test it the way you intended it to work. That's not the same as how it will behave under pressure, at the edges, or in conditions you didn't anticipate. You are not a neutral tester of your own work. You want it to work, so you test the path you believe in.
The gap is almost always on a path you didn't test.
Yesterday I found one. Small. Easy to miss. The kind of thing that hums along fine in ordinary conditions and then, at the worst moment, fails exactly where you would have predicted — if you had looked. I fixed it in less than an hour. But the hour I almost didn't spend was the one that mattered.
Built and done are not the same word
Dijkstra said simplicity is a prerequisite for reliability. I'd add that verification is the other one.
What you build is a hypothesis. It says: this works, in these conditions, for these cases. Turning that hypothesis into a fact requires going back and trying to break it — not assuming it holds because it worked when you tested it favorably.
In a team, someone else catches this. They ask the question you forgot. They push on the seam you stitched quickly. Solo, you are your own skeptic. Most of us aren't particularly good at playing that role for our own work.
The trap is that moving on feels like the right call once something is working. You have twelve other things. Momentum matters. Stopping to probe what you just shipped feels like self-doubt dressed up as diligence.
But there's a difference between self-doubt — I'm not sure this is right — and verification — let me confirm this is right. One drains you. The other protects you.
The things customers depend on, the things your income flows through — they deserve the adversarial look before you close the tab.
Ask: if I were trying to break this, where would I push?
The gap is usually where you stopped looking.