Orogen Docs

Appearance

Yuma scoring ​

Governed validator admission, top-K submission permits, and median-clipped stake-weighted per-epoch operator scoring.

How it works ​

Per epoch (360 blocks, about 36 minutes at the current 6-second block time):

  1. Governance adds or removes validator accounts in pallet-yuma-consensus, assigning each account a stake weight and entity id.
  2. Governance rotates the active Yuma permit set for the epoch to the top permitted validators by stake weight, with deterministic account-id tie breaking.
  3. Only validators permitted for that epoch can submit weight vectors over operators.
  4. Admission enforces a bounded validator set and an entity concentration cap.
  5. compute_epoch_incentives clips each submitted score to the per-operator median, aggregates clipped scores by validator stake weight, then normalizes operator incentives to basis points.

Validator allocation ​

The target economics reserve 15% of per-job emission split for validators, opML watchers, and zkML provers. Validator allocation is intended to be proportional to:

  • Sampling volume × match-rate-to-consensus.
  • Low coverage → lose emission.
  • Drift from consensus → lose emission and bond stake.

Validator permit ​

The implemented runtime enforces a bounded governed validator set and a separate epoch-scoped permit set. rotate_permits(epoch) selects the top permitted validators by governed stake weight before weight submissions begin for that epoch. In the current runtime configuration the governed set is capped at 64 validators and all 64 can be permitted.

Concentration cap ​

The current Yuma membership extrinsics enforce an entity stake-share cap after bootstrap, using the entity id attached to each validator admission record.

See also ​