Evidence and limits
Finding: Models that performed well on isolated tasks struggled when those tasks were interleaved. Long-term-memory augmentation sometimes matched or exceeded larger context windows.
Impact: Strengthen interleaving and recovery tests. Reject context-window size as continuity evidence.
Boundary: Synthetic conversations do not reproduce repository authority or real user stakes. Next: can fresh-session evaluation separate retrieval success from correct task ownership and stale-context rejection?
Flags: synthetic conversations; model-version sensitivity
Record State: Catalog review recorded; not a performance endorsement
What was reviewed: Structure and relevance screened; not a full evaluation