Deleting the source error doesn't fix the chat: context pollution benchmark, 72 cases
Lopsided_Scarcity979 · reddit · 2026-08-20
The author treats context pollution as a context-engineering problem and builds a pilot benchmark: if an error has already been repeated, derived from, or summarized in later turns, editing or deleting the source message doesn't repair the conversation.
Design: 9 synthetic task families at propagation depths 1–3, with the same final question under 5 conditions (clean, polluted, source-prune, subgraph-prune, recompute-in-dependency-order). A flagship example has a parts count corrected from 30 to 24, falsely restored to 30, then propagated into answers of 120 and 131. Four model endpoints at temperature 0 produced 540 conditions, yielding 72 paired cases where the model was right when clean but wrong after pollution.
Results: source deletion alone repaired 68/72; deletion + recomputing descendants 71/72; removing the contaminated subgraph 72/72. In the flagship case, Gemma 4 26B and GPT-OSS 20B still answered 131 after the source was deleted — not hidden memory, since the value 30 remained explicitly in the serialized context.
Takeaway: removing erroneous evidence and repairing text derived from it are distinct operations; context-management systems need explicit invalidation — mark downstream turns stale, prune the subgraph, or regenerate descendants in dependency order. Limitations acknowledged: synthetic tasks, single sampling, unnormalized reasoning settings; planned extensions include length-matched controls and a local-model track.
More from coding & agent
- AI agents struggle to close the loop on hardware boot failures — lucasmeijer · 2026-08-20
- Teknium: Use cloud browsers as agent backends for captcha breaking and proxies — Teknium · 2026-08-20
- SkillGate: Fixing Credit Starvation in Long-Horizon Agent Skill Selection — SJTU · 2026-08-20
- Voice Agent Global Scaling: Unified vs. Per-Market Agents? — Warm-Moose6028 · 2026-08-20
- Hermes Integrates NVIDIA SkillEvaluator for Pre-Install Security Scans — max_paperclips · 2026-08-20
- Help: Building an On-Premise AI Agent for Internal Documents — Remarkable_Mine_1622 · 2026-08-20