bbg roadmap
P0: storage reliability
The owner assigned highest delivery priority to durable storage for the Cyber node on 2026-09-11. BBG owns the storage contract and its Fjall/redb implementations. Cybergraph integrates native graph acceptance; Cyber/Soft3 consumes that path.
Close this work before claiming a reliable node or advancing the dependent Cyber A1 acceptance gate. It takes precedence over storage optimizations and new storage features. Independent proof research can continue alongside it; delivery resources follow P0 first. Backend availability alone does not satisfy this gate.
The P0 document defines ordered tasks, owners and executable exit criteria.
Keep contracts in specs/, rationale in docs/explanation/, and observed
results with source revisions in audit/.
other design tracks
| track | contract | scope |
|---|---|---|
| storage proofs | storage, data availability | per-node retention, size, replication and retrievability proofs |
| verifiable query | query | query compilation and proof construction |
| evy ShardStore | storage | ECS storage API and Unimem compatibility to preserve during P0 changes |
earlier proposals (contracts in specs, rationale in explanation)
| former proposal | specs | explanation |
|---|---|---|
| algebraic-nmt | indexes.md, state.md, architecture.md | why-polynomial-state.md |
| unified-polynomial-state | state.md, architecture.md | why-polynomial-state.md |
| mutator-set-polynomial | privacy.md | polynomial-privacy.md |
| signal-first | sync.md, storage.md | why-signal-first.md |
| algebraic-das | data-availability.md | data-availability.md |
| full-pipeline | architecture.md (pipeline section) | architecture-overview.md |
| temporal-polynomial | temporal.md | (absorbed into reference) |
| pi-weighted-replication | storage.md | (absorbed into reference) |