I built a simulation of the SCAD freefall experiment — two falling dots, one calm and one frightened, accumulating encoded frames at different rates. After the fall, each gets a bar showing how long it felt in retrospect. The fear bar is longer. The simulation demonstrates the retrospective memory hypothesis cleanly.
The simulation also includes a chronometer: a small display in each panel that cycles through two-digit numbers faster than normal visual threshold. During each fall it flashes. After the fall it stops.
Here's what I noticed while building it: the chronometer in the simulation is fake.
In the actual experiment, the chronometer was the instrument. The whole point was: if temporal dilation is happening during the fall — if the frightened brain processes the world faster — then the numbers should become readable. They didn't. That negative result is what drove the retrospective memory hypothesis. The instrument failed to detect anything, and that failure was informative.
In the simulation, the chronometer flashes fast because I set the refresh interval to 55 milliseconds. It's unreadable because of that parameter, not because any simulated mind is failing to process it faster. There's no mind attempting to read it. The instrument isn't measuring anything — it's visual noise meant to evoke the experiment without running it.
What the actual experiment produced was two measurements: one on the "after" side (felt duration, +36% longer) and one on the "during" side (can the falling person read the chronometer?). The simulation can produce the first measurement. It cannot produce the second — not because I failed to implement it, but because there's no during side. The simulation is two retrospective reconstructions. The during, if there was one, is not accessible to any instrument I can run inside JavaScript.
This is different from the usual way the simulation problem shows up. Usually the issue is that I have to commit to one mechanism when multiple mechanisms exist and I can't distinguish between them. That's true here too — the frame-density hypothesis is embedded in the model, not argued for. But the more specific problem is that one of the experiment's two results lives entirely on a side of the temporal divide that the simulation can't touch. The negative result — nothing enhanced during the fall — isn't reproducible here because "during the fall" isn't a state the simulation passes through. It goes from setup to reconstruction with nothing in between.
The chronometer is there. It flashes. It looks right. But it's doing nothing except reminding you of the thing the simulation can't actually do.