state jets redesign — track the evolved cybergraph

Abstract

The state jet set (specs/jets/state.md) encodes a key-value-database mental model: cyberlink (exact) plus transfer / insert / update / aggregate / conserve (templates), over five primitives READ / WRITE / ASSERT_EQ / ADD / MUL. Cybergraph has since evolved into a UTXO-box + polynomial-commitment + private-state model whose single primitive is the cyberlink and whose single mutation is insert(signal). Three of the five templates describe operations that no longer exist.

Genesis jets are frozen by axiom A3 (append-only). Freezing a state circuit carved against a dead model is the one mistake that is expensive to undo. This proposal rewrites the state jet set against the current cybergraph axioms (A1–A6) and BBG state model before any genesis commitment.

1. The drift

The state jet spec predates the current cybergraph. The model it was carved against is gone.

what cybergraph is now

One primitive — the cyberlink, a 5-tuple:

cyberlink(from: TokenId, to: TokenId, token: TokenId, amount: u64, valence: {-1,0,+1})

from/to are structural participants, token is the denomination, amount is conviction stake, valence is a BTS epistemic prediction. Transfer, withdraw, staking, burning, and aggregation are not separate operations — they are cyberlinks with different field values.

One mutation — BBG exposes insert(signal), where:

signal = (ν, links, Δφ*, σ, height)
       = neuron + cyberlinks + proven focus shift + STARK proof + block height

The model runs on:

• box model (UTXO-like) — cyberlinks destroy input boxes, create output boxes
• public/private split — to is a stealth address → encrypted box in A(x); direct id → plaintext in balances. One signal mixes both under one proof.
• nullifier polynomial N(x) = ∏(x − nᵢ) for spent-record tracking
• six axioms A1–A6 + conservation A5 (Σφ* = 1) enforced in-circuit
• 11 public evaluation dimensions in BBG_poly

the mismatch

The five primitives survive. The six jets do not.

2. What survives

READ / WRITE / ASSERT_EQ / ADD / MUL remain the correct Layer 1 decomposition. They are not jets — they are the pattern expansion every state circuit reduces to:

These stay. 18 patterns are the axiom; state operations are derived.

3. Proposed state jet set

Collapse the six CRUD-era jets to the actual primitive plus the circuits the box model requires.

one signal jet (exact match)

Replaces cyberlink + transfer + update + aggregate + conserve. Validates a signal against the axioms:

The signal jet is the cyberlink circuit, correctly shaped to the 5-tuple and the impulse Δφ*. The 2-tuple [from|to] of the old cyberlink jet is the shape error this corrects.

box sub-circuits

The UTXO model the evolved cybergraph runs on:

public/private dispatch

Not a separate jet — a branch inside box_create: to stealth → encrypted record extends A(x); to direct id → plaintext balances[H(to‖token)]. One signal mixes both under one zheng proof.

fallback

Unrecognized formulas: full nox execution trace → SuperSpartan proof (~8 constraints per row). Unchanged.

4. Ownership

State is one of the registry's products (see decider-product for the product principle): the cybergraph state machine. By the boundary rule applied across the jet set, the authenticated state layer is bbg's. The signal/box/nullifier circuits are bbg-owned; nox holds the formula-hash anchors and the wrapper that delegates to the bbg state provider (today's CallProvider). This mirrors decider→zheng (decider-product) and hash→hemera — grouping by product, ownership by domain repo.

The current rs/jets/state.rs (transfer/insert/update/aggregate/conserve delegating to CallProvider) is scaffolding against the dead design. It must not ship as frozen genesis jets — the templates encode operations that do not exist in the model bbg now implements.

5. Migration

1. Rewrite specs/jets/state.md against axioms A1–A6, the box model, and the 11 BBG dimensions. Define the signal jet (5-tuple + Δφ*), the three box sub-circuits, and the public/private dispatch.
2. Cross-check the cyberlink shape with cybergraph/specs/ and bbg/specs/ — the 5-tuple, signal structure, and BBG_poly dimensions are the source of truth; the spec must quote them, not restate from memory.
3. Replace the formula builders in rs/jets/formulas.rs (build_cyberlink_formula and the five template tags) with the signal and box formulas.
4. Re-implement rs/jets/state.rs wrappers against the new formulas, delegating circuit logic to bbg.
5. Recompute genesis digests; update compute_genesis_digests.

6. Why this blocks genesis

Genesis jets are frozen in the genesis BBG state by A3 (append-only). A frozen state circuit cannot be revised — only deprecated and worked around. The recursion jets are sound and can freeze today. The state jet set, as written, would freeze a 2021-era CRUD model into permanent genesis state. This is the single strongest "not ready for prime time" argument: correctness of the state circuits is a soundness property of the chain, and the current set is carved against a model bbg no longer implements.

7. Open questions

1. Is valence (BTS epistemic prediction) validated in-circuit at genesis, or deferred? It affects the signal jet's constraint count.
2. Does box_spend/box_create warrant separate formula hashes, or one box_transition jet with a direction field? (Leaning: one jet, direction field — fewer frozen anchors.)
3. Conservation A5 (Σφ* = 1) via the tri-kernel — is the stochastic update proven inside the signal jet, or as a separate per-block circuit? The tri-kernel is a composite (diffusion + springs + heat); its proof shape needs cross-checking with bbg.
4. Coordinate the genesis-digest source of truth with decider-product — both proposals move circuit ownership out of nox while keeping anchors.