For every episode you publish, how fast does AI actually know about it?

The per-episode variant of Information Lag. For each recent episode a show publishes, we measure how many days pass before each of the five canonical AI engines can actually name that episode. Lag is the difference between the episode's published_at and the earliest stored knowledge-probe snapshot whose payload text mentions the episode by title or platform id.

Information Lag measures how long an entity-wide fact stays stale before an engine corrects it (e.g. a change of HQ, a new CEO). Episode Adoption Lag scopes the same idea to a single unit: one specific episode a creator has just published. Different granularity, same measurement discipline.

A `null` lag reads honestly: no snapshot on file names the episode yet. That is unmeasured — not a claim that the engine will never know. Today the knowledge-probe template asks engines for entity-wide facts, not per-episode enumeration, so per-episode signal will be sparse until probes are extended. The framework ships now so the signal lights up automatically once episode mentions appear in stored payloads.

Zero LLM cost, no supplier calls. A coded text-scan over the per-engine knowledge-probe payloads the pipeline already writes. Add the wiring on any channel; the reads land on demand.