task_210726_6 — Scoring Validation & Auditor Evaluation Mechanism + Reward Engine Bootstrap (Similoluwa)

[umeradl]

cc @Abidoyesimze
Task assigned for Jul 21–27, 2026 (5 working days; deadline Monday, July 27). Full spec: Developer/tasks/task_210726_6.md

The task file has the full specification — exact function signatures, scope boundaries, and the deliverables checklist. This post is just the summary and assignment notice.

Summary

Issues #38 (slashing), #39 (scoring), #40 (auditing), and #41 (rewards) are all open and unassigned. This task is scoped to the parts of #39/#40/#41/#38 that have zero dependency on task_210726_5 (Robbert's concurrent tokenomics/fees/staking task this same week), plus the parts of #38 safe to do regardless of merge order.

In scope:

• P3 scoring: BlockFLow validation (P3-SCR), on-chain median, scenario-driven scoring redesign #39 (scoring) — map the existing fold-in/permutation-averaging claims in MECHANISM_DESIGN.md against actual code (they may not exist yet), pytest coverage for poison detection and scoring gates, confirm/fix the on-chain median at finalizeEvaluation, ship the S3 deviation threshold in shadow mode (settable, not gating)
• P3 auditing: auditor evaluation mechanism — commit-then-reveal, encrypted test data, deviation accountability #40 (auditing) — commit-then-reveal auditor score submission, per-validator encrypted test-data key mapping, test-set resampling policy (design doc if no hook point exists)
• P3 rewards: per-GI reward engine and pull-payment claim flow for clients, auditors, aggregators #41 (rewards, core scope) — per-GI reward pool (giRewardPool), DIN-denominated deposits, DAO-settable role split (bps-based, defaulted 60/20/15/5 pending Umer's P3-5.2 simulation), on-chain settlement at endGI, pull-payment claimRewards(), treasury forwarding stubbed with a TODO
• P3 slashing: conditions taxonomy, penalty tiers, slashed-stake destination, dispute resolution (P3-4.1/4.2/4.3) #38 (slashing, stretch only) — S1–S6 taxonomy spec, invariant/fuzz tests against current slash(), dispute-resolution scaffold — attempt only if Parts 1–3 finish early

Explicitly out of scope: protocol-emission top-up, DinTreasury/DinFeeRouter forwarding (needs Robbert's task merged first), final reward-split percentages, DAO-settable partial-slash fraction, jail/blacklist tier wiring.

Reference material

• Developer/design/MECHANISM_DESIGN.md §5 (Rewards), §6 (Scoring & Auditing)
• Developer/design/whitepaper-summary.md §4.4, §8
• Developer/design/scoring-auditing-references.md — new prior-work mapping for this task: BlockFlow's median-scoring/commit-reveal mechanism (arXiv:2007.03856), 2CP's marginal-gain contributivity (arXiv:2011.07516), Shapley-value data valuation (arXiv:1902.10275, arXiv:1908.08619), IPFS content-addressing (arXiv:1407.3561), and a runnable permutation-averaging Shapley reference implementation in LabeliaLabs/distributed-learning-contributivity
• P3 scoring: BlockFLow validation (P3-SCR), on-chain median, scenario-driven scoring redesign #39, P3 auditing: auditor evaluation mechanism — commit-then-reveal, encrypted test data, deviation accountability #40, P3 rewards: per-GI reward engine and pull-payment claim flow for clients, auditors, aggregators #41, P3 slashing: conditions taxonomy, penalty tiers, slashed-stake destination, dispute resolution (P3-4.1/4.2/4.3) #38 — the tracking issues this task implements against

If anything in the task spec doesn't feel right, is ambiguous, or needs clarification — comment here and cc @umeradl.