14.11 What surprising outcomes would mean

Composition creates possibilities that single-workflow experiments cannot expose. Several outcomes would be scientifically informative.

Synergy.

A causal distinction becomes discoverable only after OPTIC constructs a probe that RELIC learns when to deploy.

Antagonism.

A confident causal proposal narrows OPTIC’s search too early and performs worse than an uncertainty-preserving composition.

Order dependence.

\(C\to O\to R\) and \(R\to C\to O\to R\) produce different admitted theories under equal budgets.

Creative bottleneck migration.

After one missing structure is supplied, the limiting obstruction moves from causal language to skill language or from skill language to decision language.

Meta-accommodation.

No registered ordering succeeds, but adding a mediator or new feedback edge produces a transportable improvement.

Negative results are equally valuable. If the full system never exceeds the best pairwise composition, the third workflow may be redundant in that testbed. If removing certificate checks improves reward but increases false accommodation, the experiment quantifies the price of epistemic discipline. If a flat learner matches performance and transport, the structural machinery has not yet earned its complexity.