Visible becomes usable
A claim that may be displayed is silently treated as analytical evidence.
OSOIX EXPERIMENTAL POLICY PROTOCOL · VERSION 1
RBP-1 says what a machine claim is. ECB-1 decides whether that claim may cross into a stronger decision context.
A missing requirement opens the circuit. Uncertainty is never converted into permission.
A claim that may be displayed is silently treated as analytical evidence.
A model output leaves a research context without fresh evidence or formal uncertainty.
A human-facing recommendation acquires safety meaning without observation or authority.
High-quality public data is mistaken for permission to command or control.
The requested context is permitted within the declared RBP and ECB boundaries.
Use is bounded and requires every mandatory disclosure and restriction.
Promotion is blocked by a hard boundary, missing evidence or absent authority.
Each context is evaluated independently. A lower-level allowance can never be carried upward.
Show the claim only while its RBP class, temporal basis and limitations remain visible.
Use valid claims for non-operational analysis; UNKNOWN cannot become positive evidence.
Require bounded freshness, evidence support and contradiction controls.
Require observation basis, formal uncertainty, reconciliation and suitable authority.
Remain open until a separately authorised command architecture exists.
ECB-1 opens the circuit when a required predicate cannot be evaluated. Missing source age, covariance, observation, reconciliation or authority is recorded as a denial reason and next-evidence requirement.
{
"protocol": "osoix.ecb.v1",
"context": "advisory",
"circuitState": "OPEN",
"capsuleDigest": "...",
"claims": [{
"claimId": "claim:future-state",
"decision": "DENY",
"reasons": [
"Projected orbital state lacks formal uncertainty."
],
"missingEvidence": [
"Formal covariance or equivalent uncertainty state."
]
}],
"boundaries": {
"safetyApprovalGranted": false,
"commandAuthorityGranted": false,
"commandPath": false
}
}The circuit evaluates declared predicates, not the probability that a claim is true.
Availability and derivability are not converted into reliability percentages.
Even a satisfied predicate still requires external operational validation.
Command rights require a separate authorised architecture and cannot emerge from evidence quality.
The private implementation evaluates every claim in an Orbital Evidence Capsule, shows exactly where the promotion ladder opens, and identifies the evidence required to close it.