Capability Lab Vocabulary¶

Status: terminology synchronized through PR12.1 reviewed external-observation to neutral-evidence materialization governance

This file is the current release-level terminology index. The pre-PR10.1 long-form vocabulary snapshot is preserved at docs/legacy/pr10_0/vocabulary.md.

Recurring boundaries¶

PILOT CAPTURE != EVIDENCE
CANDIDATE != EVIDENCE
REVIEW != EVALUATION
MATERIALIZE != CLAIM SUPPORT
MATERIALIZED EVIDENCE != CAPABILITY CONCLUSION
DO_NOT_MATERIALIZE != NEGATIVE EVIDENCE
REVIEWED-RESOLUTION RECEIPT != SIGNATURE / AUTHENTICATED HISTORY
ISSUANCE WITNESS != SIGNATURE / AUTHENTICATED HISTORY
TERMINAL DEPENDENCE PASS != STATISTICAL INDEPENDENCE
TERMINAL DEPENDENCE PASS != INDEPENDENT REPLICATION
CLAIM TEMPLATE != CAPABILITY CLAIM
RUBRIC != CLAIM EVALUATION
EVALUATION POLICY != EVALUATION
EVIDENCE BEARING GUIDANCE != AUTOMATIC BEARING
MISSING REQUIRED PROBE != CONTRADICTION
MISSING OPTIONAL EXECUTION != FAILURE
DEPENDENCE PASS != CLAIM SUPPORT
EVIDENCE != CAPABILITY
CLAIM != CAPABILITY
EVALUATION != CAPABILITY
MODEL STATE != PERSON
CURRENT != LATEST
CURRENT != BEST
CURRENT != PREFERRED
CURRENT != MASTERY
CURRENT != READINESS
CURRENT != PERMISSION
CLEAR != ABSENT
VISIBLE != CURRENT
VISIBLE != AUTHORITATIVE
HIDDEN != DELETED
PRESENTATION != AUTHORITY
EXTERNAL OBSERVATION != EVIDENCE
DUPLICATE DELIVERY != NEW OBSERVATION
REUSED SOURCE EVENT ID != PERMISSION TO REWRITE HISTORY
DECLARED ORIGIN != AUTHENTICATED ORIGIN
PAYLOAD HASH != AUTHORSHIP / CORRECTNESS / CAPABILITY
LEDGER SIZE != EVIDENCE STRENGTH
REVIEWED OBSERVATION != CAPABILITY
MATERIALIZE != SUCCESS
DO_NOT_MATERIALIZE != FAILURE / CONTRADICTION / NEGATIVE EVIDENCE
DECLARED HUMAN REVIEWER != AUTHENTICATED HUMAN IDENTITY
DETERMINISTIC EVIDENCE ID != CAPABILITY INTERPRETATION
TERMINAL RECEIPT != EVIDENCE / SIGNATURE / TRUSTED TIME

PilotCaptureRecord¶

A private raw participant capture under one exact Pilot 01 protocol/session/workspace context. It is not a PR2 EvidenceRecord, claim, evaluation, outcome judgment, or capability state. PR10.0 owns raw capture.

PilotEvidenceMaterializationCandidate¶

A PR10.1 exact selected-capture proposal. It pins materialization id, frozen policy/protocol, session/subject/capture/probe/kind, workspace snapshot hash, canonical capture hash, proposed evidence id, and proposal time. The candidate is not evidence and grants no authority by itself.

PilotEvidenceMaterializationReview¶

An explicit selected review of one exact candidate, bound by candidate SHA and materialization id. PR10.1 v1 permits only declared HUMAN reviewer kind and the verdicts MATERIALIZE and DO_NOT_MATERIALIZE. Declared HUMAN does not authenticate a real-world identity. DO_NOT_MATERIALIZE creates no negative evidence.

PilotReviewedMaterializationResolutionReceipt¶

Resolver-issued local governance metadata for a successful MATERIALIZE resolution. It binds the exact candidate SHA, exact canonical review SHA, review id, evidence id, exact full canonical PR2 EvidenceRecord SHA, and resolution time.

Each receipt carries a private resolver-issuance witness bound to that exact receipt payload. The issuer capability is not retained on the witness, so copying a legitimate receipt or witness does not authorize dataclasses.replace(...) to rebind it to a different payload.

It is not a digital signature, PKI object, authenticated reviewer identity, trusted timestamp, persistence transaction, or proof that historical execution occurred.

PilotReviewedMaterializationResolutionBinding¶

The exact selected PilotEvidenceMaterializationReview paired with its resolver-issued receipt. Terminal PR10.1 dependence governance requires one-to-one binding coverage for every materialized observation slot.

PR10.1 materialized dependence governance¶

A bounded precondition ladder over multiple materialized observations. The six families are source, mechanism, coordination/control, temporal/intervention/carryover, allocation/randomization, and sampling/selection/cohort construction. Each family has exact-identity, explicit-ancestry, and reviewed-completeness stages.

A terminal PASS is not a certificate of statistical independence, exchangeability, representative sampling, independent randomization, independent cohorts, independent replication, successful performance, capability support, or claim/evaluation authority.

PilotClaimTemplate¶

PR11.0 subject-free policy metadata describing one exact proposition shape against an exact CapabilityConceptRef, plus a ClaimScope and the probe basis required before sufficient coverage may later be considered.

It is not a PR2 CapabilityClaim. It has no subject, claim id, creation time, or claim provenance and cannot enter state derivation.

Pilot 01 v1 defines two claim templates:

bounded_reasoning
bounded_execution

The separation prevents the optional execution observation from silently changing the meaning or coverage of the required reasoning proposition.

PilotRubricCriterion¶

One human-review criterion inside a PR11.0 probe rubric. It states a bounded requirement plus optional acceptable variations and material-error conditions.

A criterion is not a score, weight, probability, conclusion, or authority grant.

PilotProbeEvaluationRubric¶

The PR11.0 interpretation specification for one exact Pilot 01 probe. Every probe has exactly one rubric bound to exactly one claim template.

Each rubric defines criteria, all four PR2 evidence-bearing guidance states, and missing-probe semantics.

SUPPORTS
CONTRADICTS
INDETERMINATE
NOT_RELEVANT

The rubric guides a future explicit evaluator. It does not automatically assign an EvidenceBearing.

PilotMissingProbeSemantics¶

PR11.0 distinguishes absence from observed negative evidence.

REQUIRED_COVERAGE_GAP
OPTIONAL_UNOBSERVED

The three required reasoning probes use REQUIRED_COVERAGE_GAP. The optional execution_artifact uses OPTIONAL_UNOBSERVED.

MISSING REQUIRED PROBE != CONTRADICTION
MISSING OPTIONAL EXECUTION != FAILURE

PilotHumanEvaluationPolicy¶

The versioned PR11.0 Pilot 01 evaluation specification.

Exact ref:

civilization_bootstrap:pilot_01_basic_electricity_human_review@1

It contains the exact claim templates, probe rubrics, evidence-bearing guidance, reliability rule, coverage rule, dependence rule, and explicit authority boundaries.

It is not a ClaimEvaluation, evaluator decision, scoring result, or PersonalCapabilityState.

Pilot evaluation policy snapshot hash¶

pilot_evaluation_policy_sha256_v1(...) domain-separates and hashes the deterministic canonical JSON representation of the exact PR11.0 policy content.

POLICY REF != CONTENT HASH
POLICY HASH != SIGNATURE
POLICY HASH != AUTHORITY
SERIALIZED POLICY != EXECUTED EVALUATION

The hash is useful for detecting policy-content drift. It does not authenticate who authored or approved the policy.

EvidenceRecord¶

A person-scoped PR2 record of an observation, artifact, assessment, attestation, outcome, or demonstration. Evidence intentionally contains no capability mapping. PR10.1 maps reviewed Pilot 01 captures only to neutral evidence; PR12.1 maps reviewed generic external observations only through a frozen neutral mapping. Later interpretation belongs to explicit claims/evaluations.

CapabilityClaim / ClaimEvaluation / PersonalCapabilityState¶

A CapabilityClaim is a stable scoped proposition against an exact capability concept. A ClaimEvaluation records governed interpretation of evidence relative to that claim. PersonalCapabilityState is a separate derived layer.

PR11.0 defines claim templates and an evaluation policy only. Later PR11.x layers construct and govern real evaluations, state, acceptance and currentness without collapsing those meanings together.

PersonalCapabilityCurrentStateSelection¶

PR11.8 explicit governed current-state act for one exact concept/frame scope. SELECT names one exact accepted state; CLEAR explicitly removes current state authority for that scope. Newer derivation or acceptance does not move current automatically.

DERIVED != ACCEPTED
ACCEPTED != CURRENT
CURRENT != LATEST

PersonalCapabilityCurrentStatePortfolio¶

PR11.10 complete projection of every governed current-selection scope head in one subject history after full PR11.8 authority replay.

Each entry preserves exact scope, SELECT/CLEAR, current-selection digest and, for SELECT, exact selected-state identity/content digest. Its current_state_set contains exactly SELECT-head states.

COMPLETE PORTFOLIO != ALL HUMAN CAPABILITIES
CLEAR != ABSENT
ABSENT SCOPE != UNKNOWN
ABSENT SCOPE != INSUFFICIENT
ABSENT SCOPE != NEGATIVE EVIDENCE

CurrentStateProgressionFrontierRequest¶

PR11.9 state-less progression request. It may describe exact concept/frame seed and prerequisite scopes but contains no caller-selected personal state ID and no caller-selected subject. Personal-state inputs are resolved through PR11.8 authority.

CurrentStateProgressionAuthorityBinding¶

PR11.9 per-requested-scope authority projection:

SELECT -> exact governed current state
CLEAR  -> explicit cleared current authority, no state
ABSENT -> no governed current-selection scope

It is progression input authority only, not readiness, permission, recommendation, safety or mastery.

CurrentStateGovernedProgressionFrontier¶

PR11.9 wrapper that binds the state-less request, complete current-selection history identity, exact SELECT/CLEAR/ABSENT authority projection, and unchanged raw PR8 deterministic frontier.

PlayerWindow¶

PR9 deterministic source-visible product projection. Raw PR9 intentionally supports explicit source selection. That is useful as a low-level renderer but raw selected_state_ids / selected_frontier_id are not an HDE-facing current authority contract.

CurrentStatePlayerWindowRequest¶

PR11.11 HDE/product-facing request. It contains:

window_id
generated_at
requester_ref
viewer_ref
state-less PR11.9 progression_request
visible_achievement_ids
visible_milestone_ids
visible_legend_id

It contains no subject selector, state set, frontier set, selected state IDs, selected existing frontier ID, supplied PR11.10 portfolio, or supplied PR11.9 governed frontier.

The visible_* fields control presentation only.

VISIBLE != CURRENT
HIDDEN != DELETED
NOT DISPLAYED != DID NOT HAPPEN

CurrentStateGovernedPlayerWindow¶

PR11.11 governed product/read artifact. It fresh-derives PR11.10 complete current-state authority and PR11.9 governed progression from the same current-selection history, reconciles their authority bindings, and invokes unchanged PR9 derivation and source-backed verification.

It contains:

request
subject_ref
current_selection_history_sha256
current_state_portfolio_sha256
governed_frontier_sha256
current_state_entries
frontier_authority_bindings
window

All PR11.10 SELECT states are visible in the raw PR9 window even if progression uses only a subset. PR11.10 CLEAR remains visible in current_state_entries; PR11.9 SELECT/CLEAR/ABSENT remains visible in frontier_authority_bindings.

A complete current state later than the requested as_of causes rejection instead of silent filtering.

PR11.11 digest¶

current_state_governed_player_window_sha256_v1(...) uses:

capability_lab/current_state_governed_player_window@1\0

It is deterministic content integrity identity, not a signature, authentication proof, trusted timestamp, HDE permission, publication authority, or capability truth.

HDE read/advisory integration boundary¶

After PR11.11, HDE may consume a stable governed read snapshot without deciding which personal state is current or which existing progression frontier should be treated as authoritative.

PR12.0 + PR12.1 additionally provide a governed write-side admission path from an external/HDE observation through explicit review to neutral PR2 evidence. This still does not authorize automatic claim evaluation or capability-state mutation.

READ / ADVISORY HDE INTEGRATION          = READY
REVIEWED OBSERVATION -> NEUTRAL EVIDENCE = READY
AUTOMATIC CLOSED-LOOP CAPABILITY UPDATE  = NOT READY

HDE ACTIVITY != EVIDENCE
HDE OBSERVATION != EVIDENCE
HDE OUTPUT != CAPABILITY UPDATE
HDE VIEW != CAPABILITY AUTHORITY

ExternalObservationId¶

PR12.0 opaque identity for one immutable generic external observation record. It is not an EvidenceId and carries no evidence or capability authority.

ExternalObservationSourceRef¶

PR12.0 declared external source identity. Source kinds are application, agent_runtime, tool, external_system, actor, and other.

The ref is provenance metadata only:

DECLARED SOURCE != AUTHENTICATED SOURCE
SOURCE REF != TRUST GRANT

ExternalObservationEnvelope¶

PR12.0 subject-scoped immutable record of one declared external source event. It preserves:

observation_id
subject_ref
source_ref
source_event_id
form
origin_kind
observation_started_at?
observed_at
captured_at
context_factors
payload_refs

It intentionally contains no EvidenceKind/Outcome, capability concept, claim, evaluation, state, score, grade, success/failure, mastery, readiness, or permission field.

EXTERNAL OBSERVATION != EVIDENCE

ExternalObservationForm¶

Structural source-record form only:

event
text
artifact
conversation
bundle
other

It is not PR2 EvidenceKind.

ARTIFACT FORM != PROJECT EVIDENCE
TEXT FORM != SELF_REPORT
CONVERSATION FORM != CAPABILITY SUPPORT
EVENT FORM != REAL_WORLD_DEMONSTRATION

ExternalObservationOriginKind¶

Declared content/execution origin:

subject
other_human
model
system
mixed
unknown

Origin is not authenticated identity or contribution measurement.

DECLARED SUBJECT != AUTHENTICATED SUBJECT
DECLARED MODEL != AUTHENTICATED MODEL
MIXED != KNOWN CONTRIBUTION FRACTIONS

ExternalObservationPayloadRef¶

Stable external payload identity and exact SHA-256 fingerprint, with optional byte size and MIME type. Capability Lab may keep only this reference/fingerprint while raw payload bytes remain in local external storage.

PAYLOAD HASH != AUTHORSHIP
PAYLOAD HASH != CORRECTNESS
PAYLOAD HASH != CAPABILITY

ExternalObservationLedger¶

PR12.0 immutable subject-scoped collection of canonical observations. It requires unique observation_id, unique (source_ref, source_event_id), and one exact subject across every record.

admit_external_observation_v1(...) treats exact replay of an already admitted canonical source event as an idempotent no-op. Reuse of that source-event identity or observation id with different content fails closed.

DUPLICATE DELIVERY != NEW OBSERVATION
IDEMPOTENT REPLAY != NEW EVIDENCE
REUSED SOURCE EVENT ID != PERMISSION TO REWRITE HISTORY

ExternalObservationLedgerSuccessionReceipt¶

Structural PR12.0 result of append-only ledger validation. It binds predecessor and successor ledger hashes plus retained/added observation ids. A directly constructed public receipt is not validator-issued.

SUCCESSION RECEIPT != EVIDENCE
SUCCESSION RECEIPT != SIGNATURE
SUCCESSION RECEIPT != AUTHENTICATED HISTORY

PR12.0 binds immutability to each exact observation rather than a global continuously-growing source snapshot, so an unrelated new observation does not stale an old retained observation.

PR12.0 digests¶

Observation digest domain:

capability_lab/external_observation@1\0

Ledger digest domain:

capability_lab/external_observation_ledger@1\0

They are deterministic content identities, not signatures, authentication, trusted time, authorship, evidence, or capability truth.

ExternalObservationEvidenceMaterializationCandidate¶

PR12.1 exact source-derived proposal for whether one already admitted PR12.0 observation should become neutral PR2 evidence. It binds:

materialization_id
frozen policy_ref
observation_id
observation_sha256
subject_ref
source_ref
source_event_id
form
origin_kind
deterministic materialized_evidence_id
proposed_at

The public proposal API does not accept a caller-selected evidence id, evidence kind/outcome, summary, context, capability concept, claim, score, mastery, readiness, or permission.

The candidate binds one exact observation, not the continuously growing ledger hash. Therefore an unrelated later observation does not stale the candidate.

CANDIDATE != EVIDENCE
CANDIDATE != REVIEW
CANDIDATE != CAPABILITY INTERPRETATION

PR12.1 deterministic external-observation EvidenceId¶

external_observation_evidence_id_v1(...) derives the sole evidence identity for one exact canonical PR12.0 observation:

external_observation:<external_observation_sha256_v1>

Repeated materialization attempts for the same exact observation therefore have the same EvidenceId. This prevents caller-selected one-observation -> many-ID amplification. It does not mean the observation is valid evidence or supports any claim.

DETERMINISTIC EVIDENCE ID != EVIDENCE QUALITY
DETERMINISTIC EVIDENCE ID != CLAIM SUPPORT
DETERMINISTIC EVIDENCE ID != CAPABILITY

ExternalObservationEvidenceMaterializationReview¶

PR12.1 explicit review of one exact materialization candidate. The review binds the exact candidate digest and materialization id and uses the frozen PR12.1 policy. V1 permits only declared HUMAN reviewer kind and the verdicts:

MATERIALIZE
DO_NOT_MATERIALIZE

The reviewer ref is declared governance metadata, not authenticated real-world identity.

DECLARED HUMAN REVIEWER != AUTHENTICATED HUMAN IDENTITY
MATERIALIZE != SUCCESS
DO_NOT_MATERIALIZE != FAILURE
DO_NOT_MATERIALIZE != CONTRADICTION
DO_NOT_MATERIALIZE != NEGATIVE EVIDENCE

PR12.1 neutral evidence mapping¶

Only MATERIALIZE creates a PR2 EvidenceRecord. The mapping is frozen:

ExternalObservationForm.ARTIFACT
-> EvidenceKind.ARTIFACT

ExternalObservationForm.CONVERSATION
-> EvidenceKind.CONVERSATION_OBSERVATION

EVENT / TEXT / BUNDLE / OTHER
-> EvidenceKind.OTHER

EvidenceRecord.outcome = None

Observation times, context factors, payload refs and exact observation provenance remain source-visible. PR12.1 does not infer a capability topic, success/failure, evidence bearing, reliability, mastery, readiness, permission, or recommendation.

ExternalObservationEvidenceResolutionReceipt¶

PR12.1 resolver-issued terminal governance metadata for both review verdicts. It binds the exact materialization id, candidate digest, review id/digest, verdict, observation digest, resolution time, and — only for MATERIALIZE — the exact deterministic evidence id and full canonical PR2 EvidenceRecord digest.

A DO_NOT_MATERIALIZE receipt must contain no evidence id or evidence digest.

Every legitimate receipt carries a private payload-bound issuance witness. Changing candidate/review/verdict/observation/evidence/resolution fields via dataclasses.replace(...) invalidates the witness.

TERMINAL RECEIPT != EVIDENCE
TERMINAL RECEIPT != SIGNATURE / PKI
TERMINAL RECEIPT != AUTHENTICATED REVIEWER
TERMINAL RECEIPT != TRUSTED TIME
DO_NOT_MATERIALIZE RECEIPT != NEGATIVE EVIDENCE

ExternalObservationEvidenceResolutionBinding¶

The exact selected PR12.1 review paired with its exact resolver-issued terminal receipt. Fresh validation replays the admitted observation binding, exact candidate/review identities, receipt witness, and — for MATERIALIZE — the complete frozen neutral EvidenceRecord mapping.

The binding itself is not a claim evaluation, persistence transaction, or capability update.

PR12.1 / PR11.3 handoff¶

PR12.1 stops at:

EvidenceRecord | None
+
terminal resolution receipt

It does not mutate EpistemicRecordSet. A materialized record enters the existing epistemic layer only through an explicit successor snapshot governed by unchanged PR11.3 validate_epistemic_snapshot_successor_v1(...).

Because one exact observation always has one deterministic evidence id, a second differently reviewed/resolved record for that observation cannot replace an already appended record without PR11.3 rejecting retained-evidence mutation.

REVIEWED OBSERVATION != CLAIM
MATERIALIZED EXTERNAL EVIDENCE != CLAIM EVALUATION
MATERIALIZED EXTERNAL EVIDENCE != PERSONAL STATE
PR12.1 != CURRENT-STATE AUTHORITY
PR12.1 != PROGRESSION AUTHORITY

Other existing terminology¶

The project continues to use the PR0–PR9 terminology for capability semantics, evidence/provenance, competence frames, deterministic state derivation, proposals, history/achievements/Legends, progression frontiers/evidence gaps/exploration, and Player Window projections.

For normative current details see:

• docs/pilots/civilization_bootstrap_pilot_01_materialization_v1.md
• docs/pilots/civilization_bootstrap_pilot_01_terminal_dependence_governance_v1.md
• docs/pilots/civilization_bootstrap_pilot_01_reviewed_resolution_hardening_v1.md
• docs/pilots/civilization_bootstrap_pilot_01_evaluation_policy_v1.md
• docs/progression_current_state_handoff_v1.md
• docs/current_state_portfolio_v1.md
• docs/current_state_governed_player_window_v1.md
• docs/external_observation_boundary_v1.md
• docs/external_observation_evidence_materialization_v1.md