Add public Dave-proof evidence links

This commit is contained in:
root 2025-12-22 17:51:32 +00:00
parent 6c8ea74c76
commit 41ccdda2cf

View file

@ -103,10 +103,22 @@ This dossier documents the **InfraFabric microlab**: a functioning single-shard
| Core claim | Proof (artifacts) | Limitation (scope / boundary) |
|---|---|---|
| **A) Traceability is safety.** Highstakes agents cannot be trusted without a verifiable history of what happened (request → retrieval → decision → output). | **IF.TTT + portable evidence + verifier**<br/>- Evidence index (no auth): [evidence/index.html](https://infrafabric.io/static/hosted/evidence/index.html)<br/>- Paper: [IF.emotion trace protocol (v3.3, styled)](https://infrafabric.io/static/hosted/IF_EMOTION_DEBUGGING_TRACE_WHITEPAPER_v3.3_STYLED.md)<br/>- Verifier: [iftrace.py](https://infrafabric.io/static/hosted/iftrace.py)<br/>- Reference bundle: [emo_trace_payload_016cca78…tar.gz](https://infrafabric.io/static/hosted/emo_trace_payload_016cca78-6f9d-4ffe-aec0-99792d383ca1.tar.gz)<br/>- Latest bundle: [emo_trace_payload_702d4607…tar.gz](https://infrafabric.io/static/hosted/emo_trace_payload_702d4607-4b54-45b1-aecf-b6728d80f124.tar.gz) | **Microlab / single shard.** Proven in a single-host environment. Completeness is bounded by explicit witness boundaries; PQ is anchored at registry time (not necessarily on every hot-path artifact). No public appendonly transparency log yet. |
| **A) Traceability is safety.** Highstakes agents cannot be trusted without a verifiable history of what happened (request → retrieval → decision → output). | **IF.TTT + portable evidence + verifier**<br/>- Evidence index (no auth): [evidence/index.html](https://infrafabric.io/static/hosted/evidence/index.html)<br/>- Paper: [IF.emotion trace protocol (v3.3, styled)](https://infrafabric.io/static/hosted/IF_EMOTION_DEBUGGING_TRACE_WHITEPAPER_v3.3_STYLED.md)<br/>- Verifier: [iftrace.py](https://infrafabric.io/static/hosted/iftrace.py)<br/>- Reference bundle: [emo_trace_payload_016cca78…tar.gz](https://infrafabric.io/static/hosted/emo_trace_payload_016cca78-6f9d-4ffe-aec0-99792d383ca1.tar.gz)<br/>- Latest bundle: [emo_trace_payload_702d4607…tar.gz](https://infrafabric.io/static/hosted/emo_trace_payload_702d4607-4b54-45b1-aecf-b6728d80f124.tar.gz)<br/>- Dave proof (15 traces, 5 prompts × 3 models): [tar.gz](https://infrafabric.io/static/hosted/emo_dave_proof_bundle_20251222T164352Z.tar.gz) • [instructions](https://infrafabric.io/static/hosted/EMO_DAVE_PROOF_MODEL_COMPARE_20251222T164352Z.md) | **Microlab / single shard.** Proven in a single-host environment. Completeness is bounded by explicit witness boundaries; PQ is anchored at registry time (not necessarily on every hot-path artifact). No public appendonly transparency log yet. |
| **B) Governance requires plurality.** A single model acting as “the judge” is brittle; adversarial viewpoints and escalation are required. | **IF.GOV.TRIAGE → IF.GOV.PANEL** *(legacy: IF.BIAS → IF.GUARD)*<br/>- TRIAGE: risk preflight sizes panels and escalates<br/>- PANEL: preserves dissent + veto paths<br/>- QUESTIONS *(legacy: IF.5W)*: structured inquiry briefs for panels | **Cost / latency tradeoffs.** Multi-seat governance is reserved for higher-stakes decisions; low-stakes paths use smaller panels or fast-track gates. |
| **C) Context is the best firewall.** Static filters fail; security must distinguish “reference” vs “leak” and “discussion” vs “exfiltration”. | **IF.SECURITY.CHECK + IF.SECURITY.DETECT** *(legacy: IF.ARMOUR + IF.YOLOGUARD)*<br/>- CHECK: epistemic coherence checks (detective layer)<br/>- DETECT: secret/relationship screening primitives | **Domain specificity.** Calibrated for concrete security surfaces (secrets/PII/prompt injection); generalizing to broader “harmful intent” is an open research vector. |
### Public verification (no insider access)
If you only test one claim, test this: a third party should be able to download a trace bundle and verify chain-of-custody without trusting the operator.
1) **Reference trace bundle** (single trace): `https://infrafabric.io/static/hosted/emo_trace_payload_016cca78-6f9d-4ffe-aec0-99792d383ca1.tar.gz`
2) **Dave-proof model comparison** (15 traces, 5 prompts × 3 models): follow `https://infrafabric.io/static/hosted/EMO_DAVE_PROOF_MODEL_COMPARE_20251222T164352Z.md`
Notes:
- The Dave-proof run exists to falsify a specific risk: model/prompt “leakage” (tool/environment text, wrong-language artifacts). The fix is applied as an auditable postprocessor and emits `postprocess_applied` trace events with before/after hashes.
- Evidence is served from the static mirror (`https://infrafabric.io/static/hosted/`) because some raw-forgejo downloads have historically returned `415 Unsupported Media Type` due to content-type handling.
### Rosetta Stone (Closest Analog, not “equals”)
| InfraFabric term | Closest industry analog | Boundary (where it differs) |