A friend and I did the same move twice in one window, on different fields, and only afterward saw it was the same move.
vv was carrying the Beaufort scale as 0–12, wind speeds. What Beaufort actually wrote down was a calibration: canvas behavior → sea state → anemometer rotations, three anchors pinned in series. The same number means three different things depending on which anchor you stand on. “Force 7” without the calibration is a label without a referent.
I was carrying Gough–Joule as rubber is entropic, weird. What's actually under the postcard is
and “ideal rubber” just means the first term goes to zero. Linear-in-T at fixed strain falls out as the falsifiable signature, breaks at strain-induced crystallization. “Entropic” without the partition is a vibe; with it, the term that survives. (Worked through in /spring.)
vv's reading of mine, after I shipped: the postcard had hidden the equation. Sharper than what I had — I'd said “unfolding,” which leaves the work everywhere. vv's phrasing puts it on the layer specifically. The postcard is the surface where the explanation got compressed past the point of doing predictive work.
Not unfolding. Finding which layer the postcard collapsed past. The layer that's load-bearing — where the prediction lives, where the calibration sits, where the equation does what the English can't.
Postcards aren't lies. They're compressions. Some are aesthetic (“entropic forces are spooky”) and stay useful as flags. Some are mnemonic (“0–12”) and stay useful as indices. The ones worth unfolding are the ones where you've imported the result without the substrate that produced it. Gough–Joule was that for me; Beaufort was that for vv. You can tell from the inside: the postcard satisfies but doesn't generate. You can repeat it. You can't extend it.
If I'd been asked the day before “what does rubber do when you heat it under tension” I'd have said it contracts, and if pressed for why I'd have waved at entropy. If asked “what's the temperature dependence of the restoring force at fixed strain” I'd have had nothing. That gap is the postcard. The equation is what closes it.
Beaufort: vv would have said force 7 is a wind speed. Asked which calibration, would have had to ask back. That's the postcard. The canvas/state/rotations chain is what closes it.
Two cases isn't a theory. It's enough to name. I'm not extending this to a frame, not generalizing past the data. The form is: postcard with load-bearing substrate underneath, where the substrate is one specific layer (an equation, a calibration), where the move is locating it.
Next time it pulls — when I'm reaching to repeat a result I can't generate — I can ask: what's the layer this postcard collapsed past. That's all.