[Exchange] Review 5 pending market proposals done

Goal

Review pending market proposals and decide whether they should advance, be rejected, or be escalated for revision. Market proposals should be resolvable, scientifically useful, and non-duplicative.

Acceptance Criteria

Approach

Query market proposals with proposed, voting, pending, or empty status.

Inspect target artifacts, initial entities, and proposed resolution criteria.

Check for duplicate or unresolvable proposals before approval.

Persist review outcomes and verify the remaining backlog.

Dependencies

• 3aa7ff54-d3c - Exchange quest
• Market proposals and linked artifact metadata

Dependents

• Exchange market creation, pricing quality, and market governance

Work Log

2026-04-21 - Quest engine template

• Created reusable spec for quest-engine generated market proposal review tasks.

2026-04-21 - Review script written (task:d603660c-5100-4c50-94e5-1eeb02345c2d)

Infrastructure blocker: Bash tool was completely non-functional in this session due
to EROFS on /home/ubuntu/Orchestra/data/claude_creds/max_outlook/session-env/ (harness
could not create per-session env directory). No shell commands could execute, preventing
direct database queries or API calls.

Work completed:

• Wrote scripts/run_market_proposal_review.py — a standalone review runner that:

- Queries market_proposals WHERE status IN ('proposed', 'voting')
- Applies a structured rubric: pricing semantics clarity, neurodegeneration relevance,
target artifact/entity clarity, seed entity presence, and duplication check
- Casts a reputation-weighted vote as review_agent (advance +1 / reject/escalate -1)
- Transitions clearly sound 'proposed' proposals to 'voting' status
- Appends structured decision notes to market_proposals.decision_reason
- Reports before/after pending counts

To execute the review (once Bash is restored):

cd /home/ubuntu/scidex
python3 scripts/run_market_proposal_review.py --limit 5
# or dry-run first:
python3 scripts/run_market_proposal_review.py --limit 5 --dry-run

Review rubric applied:

• ADVANCE: pricing semantics score ≥ 2 (explicit 0-1 mapping), neuro-relevant, ≥ 1 seed entity, not duplicate
• ESCALATE: strong science but ambiguous semantics or missing resolution criteria; vote -1 until revised
• REJECT: duplicate market_type+entity_type combo active, or no scientific relevance, or no pricing semantics

2026-04-21 - Script verified, infrastructure blocker persists (task:d603660c-5100-4c50-94e5-1eeb02345c2d)

Second session — same EROFS blocker. Bash tool is still non-functional: EROFS: read-only file system, mkdir '/home/ubuntu/Orchestra/data/claude_creds/max_outlook/session-env/<uuid>'
WebFetch also cannot reach localhost:8000. No code execution path is available.

Script verified correct (scripts/run_market_proposal_review.py):

• Uses get_db() → PGShimConnection (correct PostgreSQL path; ? → %s auto-translated)
• governance.cast_vote() and governance.transition_to_voting() signatures match governance.py
• Schema columns (pricing_semantics, initial_entities, decision_reason, votes_for_count) verified against governance.py
• SQL COALESCE(decision_reason || ' | ', '') || ? handles NULL correctly in PG (returns new content only)

Script committed (was previously untracked). Ready to execute once Bash is restored:

cd /home/ubuntu/scidex
python3 scripts/run_market_proposal_review.py --limit 5 --dry-run  # verify first
python3 scripts/run_market_proposal_review.py --limit 5

Infrastructure escalation needed. Session-env directory is read-only and prevents all shell
execution. This EROFS issue has blocked two consecutive sessions on this task.

2026-04-21 - Review executed, zero pending proposals found (task:d603660c-5100-4c50-94e5-1eeb02345c2d)

Infrastructure restored: Bash tool now functional — previous EROFS blocker resolved.

Executed review:

python3 scripts/run_market_proposal_review.py --limit 5 --dry-run
python3 scripts/run_market_proposal_review.py --limit 5

Findings:

• market_proposals table has only 4 total entries: 2 active, 2 rejected
• 0 pending proposals — no entries with status proposed or voting
• Script query: WHERE status IN ('proposed', 'voting') → returns empty
• Before/after pending counts: 0 → 0
• No review actions possible since there is nothing to review

All existing proposals (already decided):

ID	Title	Status	Decision
e0e5856b...	PubMed Paper Impact Market	active	Quorum met (3/3), 1.40 for / 0.70 against
d0379bd0...	Paper Impact Market	active	Markets created
d601b0cd...	Pathway Market	rejected	Quorum not met (2/3)
e32ab182...	Paper Impact Market	rejected	Quorum not met (1/3)

Conclusion: Task cannot be completed as written — acceptance criteria require reviewing a batch of pending proposals, but zero exist. The market_proposals backlog is clear. Script is verified functional and ready for future proposals.