Scope: permissionless writer admission and turn-assignment for a sovereign Logos channel (Bedrock/Mantle). Expected regime ~10 writers. Targets sovereign zones, which are built on channels ([1.1.0][Overview] Bedrock Architecture ]], [1.5.0] Mantle ]). Native zones have their own ordering path (Execution Tickets + partitioned ledger; Preliminary Research: Bedrock Mantle Specification (Native Zones) , Execution Tickets (1) ) and are out of scope.
Legend: * marks a claim that needs an experiment before it justifies a decision. Build-status tags: EXISTS (in code today), UNBUILT (intended, net-new), FORBIDDEN (excluded).
Sequencers are known parties; each signs its inscription with its public key. There is no identity to hide, so there is no privacy machinery here: no ZK membership proof, no commitment tree, no nullifier set, no secret leader. The v1 contribution is small and concrete:
accredited_keys list. Make inscription onto it stake-gated: lock a fixed registration amount to join the round-robin; register K times to hold K positions. This is the Ethereum model, where every validator stakes the same 32 ETH and participation is weighted by registering more than once.posting_timeout as the liveness floor (EXISTS). Because stake-per-registration is fixed and turns rotate, there is no election to run; the schedule is the registered set in rotation. First-write-wins (FWW) remains a separate selectable schedule mode; round-robin is the one specified here.CHANNEL_WITHDRAW committee path (D-C1).The channel stays dumb; it orders opaque inscriptions, never reads zone-transaction content or runs the STF, and holds only a typed, op-gated stake registry it reads over itself.
Open for review: registration and maturity parameters; the turn unit (channel-iteration vs L1-block-range); stake denomination (does not block v1, D-C2). Concentration is engaged in section 8; no seat cap is proposed.
The channel is an ordering service for opaque inscriptions. It checks each inscription’s envelope (turn, signature, no equivocation, well-formed fields) and appends; it never reads the zone transactions inside, never runs the state transition function. Sequencers are a known, public set; each signs with its public key, so everyone sees exactly who wrote what.
Two halves, independent:
| Half | Decides | v1 status |
|---|---|---|
| Membership | who may be a writer; how they register and leave | the actual new work; stake-gated registration + lifecycle (UNBUILT) |
| Turn assignment | among registered writers, who writes next | reuses round-robin + posting_timeout (EXISTS); no election step |
The same registered set runs under round-robin or FWW; the same turn rule works under any admission policy.
The only new persistent state is a stake registry: a typed, op-gated list the channel maintains over itself (who is registered, how many positions, lifecycle status). It is read only by the channel about its own writer set, never over zone-transaction content, so it does not make the channel less dumb about zone state.