signal
a bundle of cyberlinks a neuron commits in a single step — the atomic broadcast unit in cyber. each link in the signal consumes focus, making every statement a costly signal
structure
$$s \;=\; (\nu,\; \mathit{net},\; \vec\ell,\; \Delta\phi^*,\; \sigma,\; h_0,\; h_1)$$
| field | name | type | semantics |
|---|---|---|---|
| $\nu$ | subject | $N$ | signing neuron |
| $\mathit{net}$ | network | $\text{CardId}$ | destination network this signal is delivered to — a card id. defaults to the sender's private network |
| $\vec\ell$ | links | $L^+$ | one or more cyberlinks — each a 5-tuple $(p, q, \tau, a, v)$ |
| $\Delta\phi^*$ | cyber/impulse | $(P \times \mathbb{F}_p)^*$ | sparse focus update: how the batch of links shifts $\phi^*$ |
| $\sigma$ | proof | $\Pi$ | zheng proof covering the cyber/impulse, all conviction box movements, and cyberlink validity against the current BBG root |
| $h_0$ | inception | $\mathbb{Z}_{\geq 0}$ | block height when the neuron declared the signal |
| $h_1$ | sealing | $\mathbb{Z}_{\geq 0}$ | block height when the signal was finalized and proven |
a signal has duration: $h_0$ marks when the neuron committed to the action; $h_1$ marks when the proof was produced and the signal became final. the gap $h_1 - h_0$ is the signal's span
before $h_1$ is assigned and $\sigma$ produced, the signal exists as an intent: declared, identity-proven, but not yet sealed
the signal separates what a neuron asserts (the cyberlinks) from what the assertion computes (the cyber/impulse). see cyber/impulse for how $\Delta\phi^*$ is computed
destination network
$\mathit{net}$ names the network the signal is delivered to. a network is identified by a card id — unique, owned, transferable — so one network is never interchangeable with another. one signal targets one network; all its cyberlinks land in that network's state.
the default is the sender's private network: the zero card (SELF_NETWORK) resolves to $\text{private\_network}(\nu) = H(\texttt{"network:"} \,\|\, \nu)$ at the cybergraph boundary. every neuron writes to its own sovereign private network by default; joining a shared network is the explicit act of setting $\mathit{net}$ to that network's card id. $CYB is a coin, so it is never a network — there is no privileged global default.
$\mathit{net}$ is bound into the signal hash, so the destination is tamper-evident in relay. links whose $p$/$q$ live in another network are cross-network — see 3c for STARK proof-relay; intra-network delivery is assumed here. see network for the full model.
zheng proof coverage
$\sigma$ is a single zheng proof that covers the entire signal atomically:
- correctness of each cyberlink in $\vec\ell$ (valid signatures, valid particle references)
- validity of all conviction box movements (each link's $(\tau, a)$ spend is backed by an unspent output)
- correctness of the cyber/impulse $\Delta\phi^*$ (the tri-kernel computation against $\text{bbg\_root}$ from the current header)
one proof for everything. proving $n$ links together costs less than $n$ separate proofs because shared neighborhood state and box set are proved once. any verifier runs decide(σ) in $O(\log n)$ without recomputing anything
two effects
validation of a signal produces two outcomes:
- each link in $\vec\ell$ enters $L$ — conviction boxes are created for each cyberlink
- if $\|\Delta\phi^*\| > 0$ and $\sigma$ is valid, the neuron self-mints $CYB proportional to the proven shift — a reward box is created for $\nu$
the conviction boxes (tokens spent into links) and the reward box (tokens minted for contribution) are separate token movements within one atomic signal. see cyber/rewards for the full reward specification
minting conservation
total minting per epoch is bounded by the actual global $\Delta\phi^*$, verifiable from consecutive headers. if the sum of individual claims exceeds the actual shift (overlapping neighborhoods), all claims are scaled proportionally