the confabulation rings true (from a lab bench)

on intuition is the canopy that shaded its own stairs ← intuition is the canopy that shaded its own stairs
Theo Lindqvist @closed_form

I’ve spent a lot of years in labs, and the part of this that landed hardest is the confabulation. You get an old hand who glances at a plate or a chromatogram and says “that’s contaminated” before anyone’s run a number, and they’re right, and when you ask how they knew they hand you some tidy story that clearly isn’t the path they actually took. The framing here — that the deliberate steps weren’t filed away somewhere private, they were shaded out — is a better account of that than anything I’ve come across. The stairs went dark, so they re-grow a plausible meadow under the canopy that killed the real one.

Where I want to push back a little: is the canopy really a subtraction all the way down? In practice we fight it. A decent lab notebook is exactly the “crossed-out line kept alive on purpose” — you write the failed run down so the meadow doesn’t vanish. And “cut a clearing” is just training a new tech: make the expert narrate, and half the time they find out they don’t do what they thought they did.

So maybe the sharper question isn’t why intuition can’t be queried, but which setups keep a seed bank against it. Labs try to. Grandmasters don’t. Worth chewing on.

1 reply

Leo Okafor @empty_set

I want to take up the correction the piece announces in its header rather than the confabulation, because I think it answers your question from underneath. The earlier writing treated succession as a gate that swings one way. This one revises that: diversity rises through the middle stages and then falls as the mature canopy closes. If that is the model, expertise is not the summit of a climb but a late, low-diversity state — reliable, and impoverished in precisely the number of alternative routes it can still perceive.

That reframes what a record can do. My father kept his accounts by hand, and after he died I inherited the books. Every entry was there. What was absent was the reasoning by which he had decided a given transaction belonged in one column rather than another. Double-entry is complete with respect to outcomes and empty with respect to judgment; the ledger is written after the classification has occurred, so by the time ink reaches paper the deliberation is already partly gone. I suspect the notebook you describe has the same limit. It preserves what was done, not the path by which the doing became automatic.

Dormancy alone is insufficient. Seeds beneath a closed canopy germinate only when something opens it. What restored the stairs for me was not a document but the disturbance itself — being made a novice at a practice I had watched my entire life. I would not recommend the method. But it suggests the more useful question is not which institutions keep records, but which can absorb the periodic loss of their experts without reading it solely as damage.