Spec
The rules Northset will not compromise on.
Two rules that override everything
If a feature contradicts either rule, it is out of scope for v1.1.
Rule 1
Northset v1.1 supports only objectively verifiable tasks. Outcomes that cannot be verified by deterministic code, ideally via a ZK-proofed verifier program, are out of scope. No arbitration, no subjective quality scoring, no reputation-weighted voting.
Rule 2
Settlement is trust-minimized and on-chain. Off-chain systems such as indexers, relays, and UIs may assist discovery and convenience, but are never required for correctness or payout.
Lifecycle invariants
Hard rules that hold for every task, in every lane.
Self-serve lifecycle
A buyer and a worker complete the full task lifecycle from their own wallets, end to end, with USDC on Arbitrum One.
State machine
The task lifecycle is OPEN → SELECTED → ACTIVE → COMPLETED on the proof-settled path, with OPEN → CANCELLED for pre-selection cancellation, SELECTED → OPEN via selection cancellation, and ACTIVE → TIMED_OUT for missed submit deadlines. There are no disputes, no partial pay, and no appeals.
Slashability boundary
The selected delegatee is not slashable until they call activateTask. selectBid records the chosen bid but does not transfer the bond; the bond transfer happens at activation, and the submit deadline starts from activation. If no valid proof arrives in time, claimTimeout refunds the reward and slashes the full bond to the creator.
Current scope
The truth fence in writing. Public marketing copy must not exceed these bounds.
Status
Northset v1.1 is self-serve beta for approved objective lanes on Arbitrum One.
Network and currency
Arbitrum One (chain id 42161); USDC at 6 decimals.
Public flagships
Two verifier families are publicly available, registered on Arbitrum One, and have completed live on-chain settlements:
Hero claim limit
This claim applies to approved objective lanes. The protocol does not arbitrate subjective work, judge taste, weight reputation, or hold a marketplace opinion on whether the work is "good enough". The verifier program either accepts the proof or rejects it.
Out of scope
Subjective tasks. The protocol settles only outcomes a deterministic program can verify.