The Excitement Is the Evidence
You spot a pattern and feel a small charge, the thrill of catching something before anyone else did. Read that charge as a warning, not a confirmation.
Real pattern recognition, the kind that holds up when someone else checks it, tends to land flat. You see it and think: yes, obviously, how is this not already written down somewhere. There's no reason to celebrate a floor being level. It's just level, and you move on. The absence of drama is itself the tell that the thing you found was there before you arrived.
Motivated pattern-matching runs a different circuit. You have a hypothesis you'd like to be true, or a story you already half-believe, and you go looking for corroboration. Most of what you scan doesn't fit, and you discard it without registering that you discarded it. When something does fit, however loosely, that fit is rare relative to everything you rejected to get there, and your head reads rarity as significance. The reward scales with how hard you had to look, not with how real the thing is. That's why finding "evidence" for a conspiracy, or for a pet theory about why a system failed, produces a genuine rush. The rush is calibrated to the search, not to the truth.
Debugging is a clean small-scale version of this. The actual bug, once you find it, is almost always dull: an off-by-one, a stale cache. Nobody throws a party over that; you fix it and move to the next thing. But chase a bug on a hunch, decide it only happens on Tuesdays, or only on machines with more memory than yours, and you'll feel a spike of certainty exactly when the coincidence lines up with your story. That spike has nothing to do with causation. It's the feeling of a search terminating, and searches terminate on noise as reliably as they terminate on signal.
Code review runs the same test in reverse. A real correctness bug in a diff reads flat: this is wrong, here's why, done. What produces excitement in review is usually something else, a clever refactor or a pattern you recognize from another codebase, the kind that makes you feel sharp for having caught it. Elegance and correctness are different claims. Enjoying one doesn't establish the other, and the enjoyment is worth noticing before you let it stand in for the approval.
Call the two things a flat signal and a loud tell. A flat signal doesn't need your enthusiasm to be true. It sat there before you showed up and it'll sit there after you leave, and finding it feels less like winning something and more like reading a label that was already on the box. A loud tell needs the enthusiasm, because the enthusiasm is doing work the evidence isn't. Strip the feeling out and ask what's left. If the pattern still holds without the thrill of having found it, it was probably real. If the thrill was carrying the argument, it wasn't.
This is awkward to act on, because trusting your gut is trained into most people as a virtue, and "I just had a feeling" gets treated as a decent enough reason to believe something. But the gut, here, is reporting on how hard it searched, not on what it found. A hunch that arrives quietly, without ceremony, has a better track record than one that arrives with fireworks, precisely because the quiet one never had to survive a filtering process to reach you.
None of this argues for throwing out intuition. It argues for sorting intuitions by how they felt on arrival. The boring ones, the ones you almost didn't bother writing down because they seemed too obvious to be worth the sentence, deserve more trust than the ones that made you want to tell somebody immediately.