One detail to clarify:
the PoW component is bounded per operator and does not accumulate globally. It is only used as a secondary modifier in committee selection.
Second clarification on the design:
Selfcoin does not attempt to compete on longest-chain security. There is no fork choice rule based on accumulated work or chain length.
The system advances only through finalized progression.
A block is accepted only after quorum finality (≥ 2/3), and nodes do not operate on competing tips or speculative states.
This removes probabilistic settlement entirely.
A transaction is either finalized or it is not. Once visible, it does not revert.
The design shifts complexity:
-- away from fork choice, mempool races, and reorg handling
-- toward validator assumptions and quorum honesty
The tradeoff is explicit:
-- no probabilistic confirmations
-- no rollback logic at the application level
deterministic validation from finalized checkpoints
in exchange for reliance on ≥ 2/3 honest participation in the active committee.
Committee selection is deterministic from finalized state at epoch boundaries. Operators are derived from finalized membership and weighted by bonded stake using sqrt scaling.
A bounded proof-of-work component is included, but its role is limited.
-- each operator performs a small, fixed search per epoch
-- only the best ticket per operator is considered
-- tickets act as a secondary modifier in deterministic ranking
Proof-of-work does not determine chain security, does not accumulate across the network, and does not influence finality.
Its purpose is to introduce a small, bounded source of unpredictability inside committee selection without creating an open-ended resource race.
Finality remains entirely quorum-based.
The implementation (selfcoin-core) reflects this model by:
-- processing only finalized blocks
-- deriving state strictly from finalized history
-- exposing finalized data only through its interfaces
The goal is to reduce ambiguity in both validation and settlement, even if that requires stronger assumptions about validator behavior.