Salience, Ranking, and Metabolism: Three Conflated Signals in Long-Running Agent Memory Systems

Who destroyed the LLM's memory? Not the model. Not RAG. Years of unexamined industry habit. A scoring formula written in 2023 for a simulation demo—relevance plus recency plus importance—moved into production hearts without a checkup, letting the three jobs of "importance" (sedimentation, ranking, retirement—which we later separate as salience, ranking, and metabolism) strangle one another inside a single signal. Our position rests on a service chain: memory serves the LLM; the LLM serves the human—and memory must be memory: self-describing and self-metabolizing, its salience and retirement decided by data and lifecycle rather than one-shot human-delegated verdicts (Invariant I0). Skip the middle link and even the most elaborate memory system drifts toward RAG. From 14+ months of production telemetry on a single conversational system (including its pre-refactor predecessor; ~38K raw memory items, ~148K association links as of the 2026-08 audit, the live store now holding ~108K), we argue that "importance" is at least three orthogonal signals that are best kept physically separated: salience (intrinsic to the data), ranking (a query-time service), and metabolism (lifecycle retirement). We document three production failures of conflation—a 93× semantic skew inside one counter, a structure-blind quality metric reporting 0.999 while the store's structure-aware coherence read 0.101, and 88% of raw memories being "islands" (73% at the time of writing, still declining without cleanup)—yet containing the most load-bearing roots. The retrieval-side pressure is measurable: an access-distribution Gini coefficient of 0.960, with the top 1% of nodes absorbing 51.9% of all accesses. We contribute a signal taxonomy with four invariants (I0–I3), six design laws, and two same-day audit tools (the Silence Test and the RAG Test). No benchmark supremacy is claimed; all evidence comes from our own failures, of which we have plenty.