bbg/reference/props/storage proofs: proving data retention at all tiers.md

π 0.0%

storage proofs

the biggest gap in bbg: particles.root proves particle existence, but where does the actual content live? how is availability proven across all four storage tiers?

the problem

bbg authenticates state (13 sub-roots committed in BBG_root) but does not prove that the underlying data is stored and retrievable. particles.root says "particle P exists with energy E" — but if nobody stores the actual content, the knowledge graph is a collection of hashes pointing to nothing.

files.root addresses content availability via DAS, but retention over time and across tiers remains an open problem.

storage tiers and their gaps

L1: Hot state (NMT roots, aggregate data, mutator set state)
    availability: guaranteed by validators running the chain
    gap: none — this IS the consensus state

L2: Particle data (full particle/axon data, indexed by CID)
    availability: validators store it, SSD-local, millisecond access
    gap: after active window, availability depends on archival nodes
    who is incentivized to store old particle/axon data?

L3: Content store (particle content / files, indexed by CID)
    availability: committed via files.root, DAS proves retrievability
    gap: DAS proves availability at commitment time. long-term retention
    requires ongoing proof of storage. if particle content disappears,
    the cybergraph is a graph of dead links.

L4: Archival (historical state snapshots, old proofs)
    availability: DAS during active window, no formal guarantees after
    gap: historical queries require someone to have kept old state

what storage proofs need to cover

  1. L2 retention: prove that particle/axon data behind particles.root is stored and retrievable. challenge-response: "give me particle at CID C" -> particle data + NMT inclusion proof.

  2. L3 retention: prove that particle content behind files.root is stored. this may require:

    • proof of storage (like Filecoin's PoRep/PoSt but over Goldilocks field)
    • incentive mechanism: neurons pay focus to keep particles alive
    • relationship to temporal decay: particles with zero inbound energy have no retention incentive

  3. axon-particle -> content binding: axon-particles in particles.root reference particle CIDs (from, to). particles.root commits to the existence of those particles. files.root proves their content is retrievable. the full chain: BBG_root -> particles.root (existence) -> files.root (retrievability).

  4. cross-tier consistency: BBG_root authenticates L1. L1 references L2 via NMT commitments. L2 references L3 via particle CIDs. proving the full chain: BBG_root -> particles.root -> CID -> content in files.root.

design constraints

  • must work over Goldilocks field and be provable in zheng-2
  • must respect bounded locality (law 1): proving storage of MY data costs proportional to MY data
  • must compose with hemera-2 (32-byte hashes, 1 perm/node)
  • must integrate with temporal decay: expired content releases storage obligations

open questions

  1. is proof of storage (PoSt-style) needed, or is incentivized replication sufficient?
  2. how does storage pricing interact with focus economics?
  3. what is the minimum viable storage proof for pre-genesis?
  4. can DAS be extended to cover long-term L3 retention, or is a separate mechanism needed?
  5. what happens when particle content is lost? how does the graph degrade?

Local Graph