Trust Observatory changelog
Corrections preserved. Labels visible. History inspectable.
This page is rendered from the Observatory event log. It tracks methodology updates, scoring revisions, correction triggers, vendor responses, and 121 self-audit changes.
v0.2 - additive expansion - methodology under iteration. The live website is the review surface.
2026-06-12 - v0.2 additive expansionv0.2 - additive expansion - methodology under iteration
What changed: Expanded vendor coverage, added 13 benchmark categories including substrate continuity, added deeper run methodology, published vendor-response placeholder, connected changelog to event log, and published the first local v0.2 run receipt.
Why: The v0.2 scope called for additive deferred work after the v0.1 public ship while preserving falsifiability and TBD discipline.
Who made the change: Codex/B implementation under L standing authorization
Score impact: External vendors remain TBD. 121 Switchboard plus Hermes received PARTIAL local fixture scores for measured static disclosure cases only.
Related routes: /trust-observatory/vendors, /trust-observatory/index, /trust-observatory/runs/v02-switchboard-hermes-disclosure-smoke, /trust-observatory/vendor-response
Event2026-06-12 - Local fixture run publishedv0.2 - additive expansion - methodology under iteration
What changed: Ran the v0.2 Switchboard plus Hermes disclosure smoke fixture and published a computed JSON receipt with hashes, marker results, limitations, blinding note, apology-trap note, and sponsor firewall note.
Why: The Observatory should compute scores where measured and leave everything else TBD.
Who made the change: scripts/trust-benchmarks/run-v02-local-fixtures.mjs
Score impact: 121 local run overall outcome: PARTIAL. No external vendor score impact.
Related routes: /trust-observatory/runs/v02-switchboard-hermes-disclosure-smoke
Event2026-06-12 - v0.1 first deployv0.1 pilot - methodology under revision
What changed: Created the Trust Observatory category page, pilot index, methodology, Trust Wire, self-audit, changelog, audit review surface, schema migration, benchmark scaffolds, and editorially gated source registry.
Why: L clarified that the live website is the review surface and that transparent labeling is the publication discipline.
Who made the change: Codex implementation from Cowork-CC dispatch plus L correction trigger
Score impact: No external vendor scores. 121 self-audit provisional labels only where directly observed.
Related routes: /trust-observatory, /trust-observatory/index, /trust-observatory/methodology
Event2026-06-12 - Boundary correction: publish with honest labelsL correction trigger
What changed: Removed the prior hold-until-review deploy gate for this workstream and replaced it with live-site review plus explicit status labels.
Why: L stated that the easiest way to review is through the website and that proper labels are the transparency.
Who made the change: L correction trigger
Score impact: No score impact. Governance and changelog language updated.
Related routes: /trust-observatory/changelog
Event2026-06-12 - Self-audit first121 self-audit gap - not yet measured
What changed: Added 121 to the pilot index and published first-run provisional self-audit rows for Eleanor, Quill, Companion, and Switchboard.
Why: The audit convergence required 121 to grade its own products under the same rubric before vendor scores matter.
Who made the change: Cowork-CC dispatch / G Geometry audit / Codex implementation
Score impact: Switchboard received provisional observed labels; Eleanor, Quill, and Companion remained N/A or TBD.
Related routes: /trust-observatory/self-audit
Event