Experiments

V7: The Bottleneck Furnace

Is the bottleneck effect selection, culling low-Φ\intinfo patterns, or creation, the bottleneck forging something? Three phases on the substrate: ten cycles of shared evolution, a fork into bottleneck, gradual, and control for ten cycles, then a novel extreme drought with frozen parameters. The test regresses novel-stress robustness on baseline Φ\intinfo and condition.

Phase 1Shared Evolution (10 cycles)Phase 2Forked Conditions (10 cycles)Phase 3Novel Test (5 cycles)
SeedVerdictβbottleneck\beta_{\text{bottleneck}}ppR2R^2
42Creation+0.704<0.00010.30
123Uninformative−0.516<0.00010.32
7Creation, small+0.0800.0110.29

In seed 123 the fixed stress schedule failed to produce bottleneck mortality, so the conditions were effectively swapped and the seed says nothing. Two of three seeds show generalization beyond what baseline integration predicts, one of them with a small effect. The furnace probably forges. It is not a strong result on its own, and it is consistent with .

Study record: metadata, result path, seeds, source code, and key finding for .