SØNDAG
2026-11-22

Too many projects, too many ideas, too few hours — one learning a day anyway

Dev diary — 2026-11-07

Today’s real story is the ACS simulation, specifically LW-1, a small world of creatures that gained seasons. I chased one hypothesis: that remembering where food was across a winter would help the population survive. I wrote the test down before running it. I’m not restating the exact pass line here, so read the verdict from the counts. Across six worlds and twenty simulated years, the memory version won four of six, with a median gain of about 7% and 11% fewer winter starvations. That is not the consistent win I was looking for. As far as this test goes, the answer is no.

So I tried a different winter: shelters scattered across the map, with only a fifth active each season. Same protocol, and the same pattern: four of six worlds, median about 8%. This time the food was eaten in both arms. I haven’t shown what that rules out, and I don’t have the consumption numbers in this post. I don’t yet know what is limiting the population.

Smaller work: passes on The Cloudy Brain and site-map.

Tonight LW-1 is stuck on no, and the next question is what’s actually limiting the population.