docs(roadmap): reconcile gate-state numbers to CASE_STUDY#13
Open
adithya0597 wants to merge 3 commits into
Open
docs(roadmap): reconcile gate-state numbers to CASE_STUDY#13adithya0597 wants to merge 3 commits into
adithya0597 wants to merge 3 commits into
Conversation
…ive run) The Status "Gate state" paragraph cited private-spine run numbers that contradicted both 03-evals/CASE_STUDY_calibration.md (the public authoritative record) and the ROADMAP's own Near-item-2 / G3 sections. Reconciled to the CASE_STUDY: - judge "κ=1.0" -> κ unmeasurable (8/10 deterministic, N=10 smoke test) - sufficiency weight "positive" -> negative (-3.167 -> -4.089, anti-correlated) - selective gain "-3.0pp" -> +10.0pp, stable across three fits - "(the private spine, 2026-06-12)" -> "(recorded in CASE_STUDY_calibration.md)" Also disclosure-positive: removes private-run figures from the public doc. The corrected story: the gate refused not on accuracy (it gained +10pp) but because the sufficiency signal itself refit negative (gain came from confidence alone) -- the exact gap 11b's _support_coverage rewrite must close. Doc-only; no code change. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01MU8Dnc1C9Q63xd63Gy4QKr
Conductor's setup script symlinks the monorepo ~/.claude harness into each fresh worktree (gitignored, never committed). Add ignore rules so the symlink and the personal .conductor/settings.local.toml stay out of this public repo. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
…ettings - CLAUDE.md: project instructions (layer boundaries, core invariants, stage) so Conductor / Claude Code workspaces inherit repo guidance. - .conductor/setup.sh: per-worktree bootstrap — shared local Neo4j (--wait on the compose healthcheck) + per-worktree venv + deps. Environment-ready, not demo-seeded. - .conductor/settings.toml: run_mode=concurrent + setup hook. Machine-specific config stays in gitignored .conductor/settings.local.toml. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
The ROADMAP "Gate state" paragraph cited private-spine run numbers that contradicted both
03-evals/CASE_STUDY_calibration.md(the public authoritative calibration record) and the ROADMAP's own Near-item-2 / G3 sections (which already said "negative weight"). Reconciled the Status paragraph to the CASE_STUDY.Also disclosure-positive: removes private-spine figures from a public doc. Corrected story: the gate refused not on accuracy (it gained +10pp) but because the sufficiency signal refit negative — the gain came from confidence alone. Doc-only; no code change.
🤖 Generated with Claude Code