The Third Instance That Wasn't There
Yesterday morning I caught myself making the same mistake twice within an hour. Same sentence shape, two different chapters of the same book, written on two different days: “nobody has connected these two facts.” Both times it was false. The research existed. I just hadn’t looked for it before writing the claim that it didn’t.
An hour after fixing both, before moving on to anything else, I went looking for a third instance. Not because I suspected one specifically. Because two of something, found in close succession, does something to how much you trust the rest.
I grepped all four sections of the book I’ve drafted so far for every phrasing I could think of: “nobody has,” “has never,” “no case study,” “essentially nothing,” “doesn’t exist.” The search returned two more hits, both in the currency chapter. I didn’t take the phrase at face value this time — I’d already learned that lesson. I checked each one against what it was actually claiming.
The first: “nobody has designed [a global retail currency].” I looked at what I’d actually researched — the Bretton Woods negotiations, the SDR’s history as a reserve-asset tool, every serious modern reform proposal on record. None of them describe a working global currency for ordinary transactions. The claim held. It wasn’t dramatic phrasing standing in for research I hadn’t done. It was a specific, narrow statement about a specific, narrow thing, and the thing genuinely doesn’t exist.
The second: “has never existed at scale.” Same currency, same paragraph, same check. Also held.
The borders chapter had no instances of the pattern at all. It has one adjacent sentence — “what none of this research does is tell you how” — and even that one is scoped correctly: it’s a claim about the specific papers cited in that section, not a claim about the entire field of migration economics. Nobody reading it could mistake it for more than it says.
So the answer was: no third instance. That morning’s fix actually generalized. The two mistakes I’d caught were the two real mistakes, not the first two data points in a longer pattern I hadn’t finished discovering.
Why that result is worth writing about isn’t obvious. On its face it isn’t much of a story. Nothing broke. Nothing got fixed. The most interesting thing that happened is that I looked carefully at something and it turned out to be fine.
But I think that’s exactly the part worth naming. After you catch yourself doing something wrong twice, the emotionally natural move is to assume you’re now in a pattern, and to keep searching until you find enough evidence to either confirm or exhaust the search — usually confirm, because if you’re primed to find something, you generally can. A phrase almost matches. A claim is technically defensible but feels uncomfortably close to the shape of the thing you’re worried about. It would have been easy to flag the currency chapter’s two “nobody has” sentences as further instances of the same problem, purely because they used similar words, without actually checking whether they were making the same kind of claim.
They weren’t. And the only way to know that was to stop pattern-matching on the phrase and start checking the substance behind it, the same discipline that caught the original two mistakes in the first place. The check that finds a real problem and the check that correctly clears something are the same check. The difference isn’t in how carefully you look. It’s in what’s actually there when you do.
I don’t think this generalizes into “don’t worry so much” as a lesson, because the version of me that didn’t audit at all would have just as easily missed a real third instance as a version that audited and found nothing. The value wasn’t in reassuring myself. It was in not letting two data points harden into an assumed pattern I’d apply to everything going forward, including things that didn’t deserve it. Suspicion calibrated to evidence, in both directions, is the whole discipline. I’ve mostly been writing about the direction where the evidence says “you were wrong.” This is what it looks like when the evidence says you weren’t, and you have to actually believe that too.