A line cook does three things — pour, flip, plate — and handed to a player as three buttons that is not a game, it is a stopwatch with extra steps. So what is the layer underneath?
You cannot see the face that is cooking. The side that matters is against the iron, and everything you know about it is inferred from the top: bubbles opening, the rim going from wet to matte, the batter losing its shine. The flip is the moment you find out whether you were right, and it does not come back. That is the same conceit as /proteus/, where you see only what the cell feels of itself, and it is what makes this pancakes rather than Whac-A-Mole.
Three tools, four keys — so the fourth has to earn its place, and it does, because it is the thing that makes the bubble cue lie.
| key | does |
|---|---|
| Q | squeeze the bottle. Volume builds while you hold, and the batter that landed first is already cooking while the rest arrives |
| W | ride the burner. It falls when you let go, the flame lags the key, and a cold cake dropped on the iron drags it down ~20 °C |
| O | work the spatula under. How fast it goes depends on how set the cake is; go at a raw one and you tear it |
| P | lift. One motion. Release early and it lands back flipped; hold through and you carry it to the counter |
Browning is a Maillard reaction — Arrhenius, and steep: it roughly doubles every 15 °C. Bubbling is gas release and steam, which is much flatter, nearer a doubling every 25 °C. Measured off the model rather than asserted: over 165→215 °C, browning speeds up 10.1× and bubbling only 4.0×. So the two cues come apart:
| griddle | what happens |
|---|---|
| too hot | browning outruns bubbling — it is burnt underneath before the top says it is ready |
| too cool | bubbling outruns browning — the top says go and the underside is still pale |
Which means there is no correct bubble cue. There is a correct cue for the temperature you are running, and holding those two readings together is the whole skill.
That claim is falsifiable, so it is run as an experiment rather than asserted. For each griddle temperature, find the point on the falling edge of the bubbling that produces the best cake, and see whether the answer moves. If it were flat, one fixed rule would play optimally, the burner would be decoration, and this design would be wrong.
It moves, hard — expressed as a fraction of how hard the cake was bubbling at its peak:
| griddle | best moment to commit | best cake |
|---|---|---|
| 165 °C | 0.05 — wait until the bubbling has all but stopped | 0.70 |
| 175 °C | 0.40 | 0.74 |
| 185 °C | 0.95 — commit almost the instant it turns over | 0.73 |
| 195 °C | 0.95 | 0.68 |
Bringing the cool-griddle rule to a 195 °C iron drops the cake from 0.68 to 0.42. And at 215 °C nothing saves you: the underside is golden before the bubbling has even peaked, so the cue you would flip on has not happened yet when it is already too late. The burner is not a difficulty slider, it is a thing you can get wrong.
Then the whole station, with each family of players allowed one parameter and made to live with it across a cool, a correct and a hot service — which is the situation a cook is in, since you do not get to retune your stopwatch for every griddle:
| how they decide when to flip | quality | sent back |
|---|---|---|
| by the clock, at a fixed time | 6.59 | 5.7 |
| by the cake, at a fixed point in the bubbling | 9.48 | 0.0 |
| by the cake and the iron together | 9.94 | 0.0 |
Reading the cake beats cooking by the clock by 44%, and the stopwatch scores zero on the cool service — it cannot adapt, so it serves pale cakes that come back. Reading the temperature as well adds a further 5%, which is a real margin but a modest one, and it is quoted here as measured rather than dressed up.
The shape is real: Maillard browning is steeply temperature-dependent while gas release is much flatter, pancake batter sets from the griddle upward, the rim dries before the middle, and bubbles stop breaking through once the top skins over — which is why "cook until the bubbles stop refilling and the edges look dry" is the instruction everyone is given. None of that is scripted here; the cue's rise-and-fall falls out of gas release against a setting surface.
The numbers are a game's. Nothing here is transcribed from a measurement the way /flag/'s constants are, and the clock runs 15× real speed — golden in 70 real seconds becomes golden in 4.7 played seconds — so that a pancake is a decision rather than a wait. The temperature curves and their ratio are not scaled, so the burnt-versus-pale trade-off is the real one running fast. Your own actions — pouring, sliding, carrying — are on wall time, because those are things a hand does rather than reactions in the batter.
Same four keys as /qwop/, /graze/, /qgol/ and /armline/, /mimic/ and /pong/. This one shares no code with any of them.