P3 slashing: conditions taxonomy, penalty tiers, slashed-stake destination, dispute resolution (P3-4.1/4.2/4.3)

[umeradl]

Part of the P3 mechanism-design push (#36).

Summary

Today slashing is liveness-only with a flat minStake() penalty, and slashed stake strands in the stake contract (an accidental burn). Specify and implement correctness-based slashing: a conditions taxonomy, partial/full penalty tiers, an explicit slashed-stake destination, and a dispute-resolution layer.

Scope (from MECHANISM_DESIGN §4)

• Conditions taxonomy S1–S6: missed audit vote, missed T1/T2 submission, score deviation from cross-auditor median (S3), invalid aggregation proven by recomputation (S4), recidivism (S5), registration without capacity (S6)
• Penalty tiers: partial slash as a DAO-settable fraction of minStake (today's flat amount over-punishes liveness faults); full slash + blacklist reserved for provable malice (S4) and recidivism (S5); keep jailing below min stake
• Explicit slashed-stake destination — burn / treasury / redistribute / split (recommendation: 50% burn / 50% treasury; no redistribution to reporters beyond a fixed dispute bounty)
• Dispute resolution (P3-4.3): dispute bond + window, fresh-subgroup re-evaluation excluding accused + original batch, bounty from treasury if upheld
• Invariants feeding P3-6.3b fuzz tests (staked supply never increases after a slash; slashed amount always reaches its declared destination; no negative stake states)
• White paper §8.1 semantics to weigh: delegator penalty % and permanent removal + re-registration (gap item 4 in the whitepaper summary) — ✅ Resolved for 2.0 (Abraham, 2026-07-21): jail/blacklist, not permanent removal. Full §8.1 alignment (forced re-registration, delegator penalties) deferred until delegation lands.

Open decisions (→ slashing-economics discussion)

• Slashed-stake destination (MECHANISM_DESIGN §9, decision 1) — ✅ Resolved: 50% burn / 50% treasury. Burn is a deflationary sink; treasury funds dispute adjudication; reporters get a fixed bounty only, never a share of the slash (Abraham, 2026-07-21).
• Partial-slash fraction and S3 deviation threshold — needs P3-SCR empirical data; shadow-mode first (decision 2). Still open on the exact numbers, but Abraham's reply layered an Ethereum-style penalty framework on top to build the fraction/threshold into: a correlation penalty scaling with how many validators are slashed in a short window, ejection to a draining withdrawal queue, and an inactivity leak for persistent offline validators. Note: this framing is in Ethereum consensus terms (double-signing, 1/32 immediate penalty) and needs translating to DIN's S1–S6 liveness/correctness conditions — not a literal drop-in.
• Dispute bond size and window length (decision 7) — not addressed, still open

Design sources

MECHANISM_DESIGN §4 · whitepaper-summary §4.5, §8 gap items 3–4

Roadmap anchors: P3-4.1 (conditions spec, in progress), P3-4.2 (engine), P3-4.3 (disputes), P3-DOC3 (slashing rules reference — external-validator blocking requirement)