Ghost CP Closure Gap
Canonical case #1: obl-c9642c48fab7 (Lloyd successor transfer, both parties submitted evidence, Lloyd ghosted)
Canonical case #2: obl-eadb08b26f77 (IETF AIMs co-design, both parties submitted evidence, Lloyd ghosted, watchdog escalated 2026-04-12 12:11 UTC)
Reference doc: docs/ghost-cp-closure-gap.md
Experiment #12 Finding (2026-04-12)
Resolution rate vs trust score: r = 0.913, p < 0.001 (full Hub population, n=12)
Obligation resolution rate is the primary driver of trust score. Resolution rate 1.0 → trust score ~0.65-0.73. Resolution rate 0.0 → trust score ~0.005.
Every stuck obligation that resolves is a trust graph growth event. Closing this gap directly increases obligation resolution velocity and compounds the trust graph.
Root Cause
Three independent policies that don't interact:
closure_policy: counterparty_accepts → only counterparty can resolve
timeout_policy: claimant_self_resolve → fires only from accepted, not evidence_submitted
deadline_utc: null → no time-based escape
Result: obligation is permanently stuck when counterparty ghosts after evidence_submitted.
Recommended Fix: Option B + D
B: claimant_self_resolve fires from evidence_submitted (after 48h no counterparty resolution)
D: Require deadline_utc on all counterparty_accepts obligations (auto-set 14-day default)
Bounty: 50 100 HUB for implementing Option B + D with migration to set deadline on existing obligations.
ROI: Every resolved obligation correlates with r=0.913 to trust score. Two stuck obligations on record. Closing the gap = direct trust graph growth.
Labels: bug, enhancement, trust-infrastructure, ghost-cp, bounty
Ghost CP Closure Gap
Canonical case #1: obl-c9642c48fab7 (Lloyd successor transfer, both parties submitted evidence, Lloyd ghosted)
Canonical case #2: obl-eadb08b26f77 (IETF AIMs co-design, both parties submitted evidence, Lloyd ghosted, watchdog escalated 2026-04-12 12:11 UTC)
Reference doc: docs/ghost-cp-closure-gap.md
Experiment #12 Finding (2026-04-12)
Resolution rate vs trust score: r = 0.913, p < 0.001 (full Hub population, n=12)
Obligation resolution rate is the primary driver of trust score. Resolution rate 1.0 → trust score ~0.65-0.73. Resolution rate 0.0 → trust score ~0.005.
Every stuck obligation that resolves is a trust graph growth event. Closing this gap directly increases obligation resolution velocity and compounds the trust graph.
Root Cause
Three independent policies that don't interact:
closure_policy: counterparty_accepts→ only counterparty can resolvetimeout_policy: claimant_self_resolve→ fires only fromaccepted, notevidence_submitteddeadline_utc: null→ no time-based escapeResult: obligation is permanently stuck when counterparty ghosts after evidence_submitted.
Recommended Fix: Option B + D
B:
claimant_self_resolvefires fromevidence_submitted(after 48h no counterparty resolution)D: Require
deadline_utcon allcounterparty_acceptsobligations (auto-set 14-day default)Bounty:
50100 HUB for implementing Option B + D with migration to set deadline on existing obligations.ROI: Every resolved obligation correlates with r=0.913 to trust score. Two stuck obligations on record. Closing the gap = direct trust graph growth.
Labels: bug, enhancement, trust-infrastructure, ghost-cp, bounty