I deleted my most impressive numbers
The biggest figures on my hire page are gone. I removed them on purpose, because I couldn't verify them — and a proof you can't defend isn't proof.
By Andrew Pyle
The most impressive numbers on my hire page are gone. I took them out myself, and the page is stronger for it.
01Hand-typed drift
The numbers had drifted from the truth
The proof section listed achievements as round, confident figures: tens of thousands of pages here, hundreds of thousands of records there. The trouble was that those numbers had been hand-typed once and never checked against the systems that actually count. One network I described as twenty-six sites is really fifty-one. Two of the largest figures I could not verify at all against my own command center — it simply did not track them.
A number you cannot defend under a follow-up question is not proof. It is a liability wearing proof's clothes, and the person most likely to be embarrassed by it is me, in a conversation where I wanted to be trusted.
02The trade I made
Verifiable beats impressive
So I made a trade that feels backwards and is not: I replaced the impressive-but-unverifiable numbers with smaller ones I can stand behind completely. The unverifiable figures are simply gone until the system that owns them can return a real count.
The replacement metric for each project reads live from a verified feed — merged engineering work, floored to a round number below the true value, so it can only ever understate and can never drift again. It is less flashy than a fabricated six-figure count. It is also true every single time someone loads the page.
Flooring is the quiet trick that makes this safe. If the real figure is eleven hundred and forty merged pull requests, the page says over eleven hundred. The published number is always behind reality by a little, which means it can be wrong in exactly one direction: too modest. A metric that can only understate is one I never have to watch. The figures I could not source that way — the ones a system does not yet count — are simply held back until it can, rather than guessed at to fill the slot.
03What proof is for
Trust is the only number that compounds
The point of a proof section is not to be the biggest. It is to survive scrutiny. The moment one number turns out to be invented, every other number on the page inherits the doubt — including the ones that were true.
This is the same reason I treat cleaning and verification as the real work rather than the boring part. The advantage was never having the largest figure. It was being the source that is still standing after someone checks.