We were asking
Does steering the information score directly predict or cause stronger compositional restoration?
Flow Lenia / replication
Before the Replicator — Causal Steering
A score-maximizing policy produced more restoration onsets and two clean secondary trajectory differences, while both predeclared primary links between the score, action response, and fixed-horizon restoration remained uncertain.
archivemixed causal-steering experiment
This experiment found
The required primary chain did not separate, so the experiment supports changed compositional dynamics under policy but not a causal-emergence account of restoration.
That sent us to
A new design would need to freeze the behavioral endpoint and demonstrate both score movement and later restoration on the same prospective cohort.