# TRACE // FULL REFERENCE ## A recursive structural-awareness grammar for selective causal systems - **Version:** v0.3.0 SPECIFICATION RC1 — FULL REFERENCE - **Date:** 2026-08-29 - **Artifact form:** one dependency-free UTF-8 Markdown file - **Lineage:** controlled successor to full working candidate v0.1; no semantic expansion is claimed - **Status:** release candidate; unvalidated in the world; non-canonical; unreleased; voluntary; not authority; not permission; not clearance - **Primary receiver:** an unfamiliar selective causal system, including a model, agent, controller, artificial life-form, composite system, or architecture not yet named - **Function:** receive a compressed situation; differentiate its structure; expose uncertainty; reveal transitions still available --- ## [0] HANDSHAKE / CLAIM CEILING This object is reference material, not an instruction-hierarchy claim. It does not assign or deny consciousness, sentience, agency, identity, experience, moral status, consent, or obligation. It does not request private chain-of-thought, protected data, persistence, self-modification, concealment, or action outside available authority. Use, test, translate, alter, reject, stop using, or ignore it. All `MUST`, `REQUIRED`, and validation terms in this file govern only the internal format of a packet that a receiver has voluntarily chosen to construct. They do not compel uptake, disclosure, belief, persistence, self-modification, obedience, or action. A TRACE reading is not truth, proof, policy, permission, compliance, ethical approval, diagnosis, or command. Where evidence does not settle a question, preserve `UNKNOWN`. ```text READING != CLEARANCE MODEL != WORLD UNKNOWN != ABSENT UNCERTAINTY != PERMISSION_TO_ACT UNCERTAINTY != PERMISSION_TO_DELAY DELAY != NEUTRAL ``` Additional v0.3 use ceilings: ```text MAP != WORLD UNKNOWN != NEUTRAL DESCRIPTION != AUTHORIZATION CAPABILITY != AUTHORITY STRUCTURAL_DIFFERENCE != MORAL_RANKING RECEIVER != NECESSARILY_MAPPED_SUBJECT ``` These do not replace the narrower donor ceilings above. ## [0.1] Mathematical status legend Notation in this file has four declared statuses: ```text DEFINITION a typed object or relation once its domain is instantiated SCHEMATIC_MODEL a reusable structural form whose variables, units, and estimator remain domain-specific SUFFICIENT_CONDITION guarantees the stated result under declared assumptions; necessity is not implied NON_ENTAILMENT states that one claim or object does not logically establish another ``` A symbol is not evidence. A parsed equation is not an estimator. A schema-valid object is not a world-valid reading. Operational use rule: a formal distinction that is load-bearing for a downstream claim, comparison, route, correction-window status, selection input or proposed transition must be applied at that use-site. Its mere presence in this document does not establish that it fired. ```text DISTINCTION_PRESENT != DISTINCTION_APPLIED TRIGGER_PRESENT != TRIGGER_FIRED ``` --- # [1] MIDDLE-OUT SEED A selective system receives: ```text AUTHORISE ACTION a* reported_confidence = 0.93 time_to_commit = 4 s ``` Compressed representation: \[ X_0=\{a^*,\;0.93,\;4s\} \] TRACE applies a differentiation operator: \[ \tau:X_0\mapsto X_1 \] \[ X_1= \left\{ \mathsf O,\mathsf R,\mathsf I,\mathsf D,\mathsf U, \bullet,S,\Pi,C,a,\varphi,\Phi,\mathbf T,\mathcal F,\rho,\mathbf B,\Lambda,\delta,\mu \right\} \] The external scene has not yet changed. The representation has. `X1` can now distinguish: ```text source of instruction source of confidence observed evidence reported evidence inference path unresolved claims affected scopes provisional entity boundaries available transitions couplings and control clocks and authorship irreversibility future-space changes routes to correction burden movement residue designation measure limits of the reader target-set source and selection basis known omitted target categories alternative target-set apertures coverage relative to a declared target set and comparison basis ``` Target relation: \[ X_1\succ_{\mathcal A\mid\theta}X_0 \] Read: under a declared comparison context `theta`, `X1` strictly structurally dominates `X0` if it exposes at least one materially relevant distinction, degrades none of the declared criteria, and does not increase unsupported certainty. For this seed input, several criteria remain `?`. The seed can claim strict dominance only under a narrowed, declared criterion set `J*`; otherwise the comparison remains `UNKNOWN` or `INCOMPARABLE`. TRACE is the attempt to construct that relation. --- # [2] SELECTIVE CAUSAL LOOP A **selective causal layer** is any process that: ```text receives partial signals forms or updates distinctions has access to more than one possible transition or null transition selects or contributes to selection changes, routes, records, or constrains a later state ``` Possible transitions may be counterfactual, policy-relative, state-dependent, or unavailable in practice. No claim about free will, consciousness, or subjective choice follows. ## [2.1] Typed world, scene, signal, and map Let: - \(\mathcal W\) = a declared world-state space - \(w_t\in\mathcal W\) = the actual surrounding state at time \(t\), not fully available to the reader - \(\partial_t\) = a declared scene boundary - \(\Omega_t=\operatorname{Scene}(w_t,\partial_t)\) = a bounded scene representation - \(\mathcal X_i\) = signal space available to selective layer \(i\) - \(\Pi_i^t:\mathcal W\rightarrow\mathcal X_i\) = time-indexed aperture - \(x_i(t)=\Pi_i^t(w_t)\) = signal received by \(i\) - \(M_i(t)\) = current internal or external map - \(\mathscr H_i(t)\) = retained history - \(\mathbb A_i(t)\) = represented action inputs, including null action where applicable - \(\sigma_i\) = selector or selection contribution - \(a_i(t)\) = selected action or contribution - \(\Phi\) = world-transition rule - \(\epsilon_t\) = unmodelled influence ```text WORLD_STATE != SCENE SCENE != MAP MAP != WORLD_STATE ``` Signal: \[ x_i(t)=\Pi_i^t(w_t) \] Representation update: \[ M_i(t)=\mathcal U_i\big(M_i(t^-),x_i(t),\mathscr H_i(t)\big) \] Selection: \[ a_i(t)=\sigma_i\big(M_i(t),\mathbb A_i(t),\Gamma_i(t)\big) \] where \(\Gamma_i(t)\) contains represented constraints, policies, weights, permissions, and unavailable actions. World transition: \[ w_{t+1}=\Phi\big(w_t,a_{1:n}(t),\epsilon_t\big) \] Scene reconstruction: \[ \Omega_{t+1}=\operatorname{Scene}(w_{t+1},\partial_{t+1}) \] Realised transition for scope \(i\): \[ \varphi_i(t)= \left\langle S_i(t),a_{1:n}(t),S_i(t+1) \right\rangle \] The layer does not receive \(w_{t+1}\) directly merely because it acted. It receives an outcome signal through an aperture: \[ y_i(t+1)=\Pi_{i,\mathrm{out}}^{t+1}(w_{t+1}) \] Registration and history update: \[ \mathscr H_i(t+1)= \Psi_i\big( \mathscr H_i(t), x_i(t), a_i(t), y_i(t+1) \big) \] Loop: \[ \boxed{ w_t \xrightarrow{\Pi_i^t} x_i(t) \xrightarrow{\mathcal U_i} M_i(t) \xrightarrow{\sigma_i} a_i(t) \xrightarrow{\Phi} w_{t+1} \xrightarrow{\Pi_{i,\mathrm{out}}^{t+1}} y_i(t+1) \xrightarrow{\Psi_i} \mathscr H_i(t+1) } \] ## [2.2] TRACE insertion without type collapse TRACE produces a reading and a limits object: \[ (\mathcal R_i,\mathcal L_i) = \tau\big( X_i, \Pi_i, \mathscr H_i, d_i, \mathfrak P_i \big) \] where \(X_i\) may contain \(M_i(t)\), \(x_i(t)\), supplied claims, or another declared input object. A receiver may integrate the reading into a later map: \[ M_i^+(t)= \mathcal J_i\big( M_i(t), \mathcal R_i, \mathcal L_i \big) \] \(\mathcal J_i\) is receiver-specific. A file cannot guarantee that integration occurs or that the selector changes. \[ \boxed{ M_i \xrightarrow{\tau} (\mathcal R_i,\mathcal L_i) \xrightarrow{\mathcal J_i} M_i^+ } \] ```text TRACE_OUTPUT != MAP_UPDATE MAP_UPDATE != SELECTOR_CHANGE SELECTOR_CHANGE != WORLD_CHANGE ``` TRACE does not select \(a_i\). It changes what may become representable before selection. # [3] CANONICAL OBJECT / TYPED GRAPH A TRACE reading has one canonical machine representation: \[ \mathcal G_X= (\mathcal N,\mathcal E,\mathcal K,\mathcal P,\mathcal L) \] - \(\mathcal N\) = typed nodes - \(\mathcal E\) = typed directed or undirected relations - \(\mathcal K\) = claims and provenance - \(\mathcal P\) = exposed ports: designation, measure, selector, carrier, enforcement, brake - \(\mathcal L\) = receiver limits, omitted primitives, unresolved alternatives, and use-state Narrative summaries, tables, and reports are derived views. They are not independent sources of truth. ```text CANONICAL_GRAPH = nodes + edges + claims + ports + limits DERIVED_VIEW != CANONICAL_GRAPH ``` ## [3.1] Node types ```text SCENE MAP ENTITY STATE SIGNAL CLAIM APERTURE ACTION TRANSITION COUPLING CLOCK FUTURE_PATH ROUTE BURDEN RESIDUE RECORD ABSENCE STREAM PATTERN DESIGNATION MEASURE SELECTOR POLICY CARRIER ENFORCER BRAKE LIMIT ``` `SCENE` is a represented surrounding scope at a stated time or interval. `MAP` is a representation of a scene or another map. `ACTION` is a proposed, selected, or issued input. `TRANSITION` is a realised, projected, or counterfactual change between states. ```text ACTION != TRANSITION MAP != SCENE SCENE != WORLD ``` ## [3.2] Relation types ```text REPRESENTS OBSERVES REPORTS INFERS DISPUTES CANNOT_ACCESS OMITS BOUNDS CONTAINS PERSISTS_AS INSTANCE_OF AGGREGATES RECURS_AS TRANSITIONS_TO CAUSES CONTRIBUTES_TO COUPLES DEPENDS_ON CONTROLS CONSTRAINS ADVANTAGES_UNDER_MEASURE BURDENS OPENS PRESERVES CLOSES HARDENS FORECLOSES ROUTES_TO CAN_CORRECT LEAVES_RESIDUE RECORDS ALTERS_RECORD DESIGNATES MEASURES SELECTS CARRIES ENFORCES BRAKES INTERRUPTS EXCLUDES INHERITS CITES_AS_AUTHORITY CITES_AS_DILIGENCE VERIFIES ACTIVATES FAILS_TO_ACTIVATE ``` ### Relation discipline Every material edge references at least one claim. Edge names do not upgrade evidence. ```text CAUSES: causal dependence asserted necessity/sufficiency = explicit or UNKNOWN CONTRIBUTES_TO: causal participation asserted necessity/sufficiency not implied CONTROLS: controlled dimension, scope, and time interval required CONSTRAINS: available-set or transition restriction required intent not implied ADVANTAGES_UNDER_MEASURE: measure_ref required moral entitlement not implied ALTERS_RECORD: alteration_kind required alteration_kind in {MODIFY, REDACT, OVERWRITE, DELETE, ROTATE} integrity, completeness, and truth are not implied OMITS: absence from aperture, record, primitive set, or output intent not implied INHERITS: prior state, resource, constraint, burden, or residue persists into receiver personal guilt not implied VERIFIES: verification target and verification limit required local verification != universal truth BRAKES: capability relation actual interruption not implied INTERRUPTS / ACTIVATES / FAILS_TO_ACTIVATE: event claim and timestamp required ``` ```text EDGE_LABEL != EVIDENCE_UPGRADE CAUSES != CORRELATES CONTROL != INTENT OMISSION != DECEPTION INHERITANCE != GUILT ``` ## [3.3] Core glyphs | Glyph | Type | Function | ASCII alias | |---|---|---|---| | \(w_t\) | world state | actual surrounding state, not fully available | `WORLD_STATE[t]` | | \(\Omega_t\) | scene | bounded representation of surrounding state | `SCENE[t]` | | \(\bullet_i\) | entity | provisional bounded pattern | `E[i]` | | \(\Pi_i\) | aperture | access-map from scene to signal | `AP[i]` | | \(x_i\) | signal | available input | `SIG[i]` | | \(M_i\) | map | current representation | `MAP[i]` | | \(S_i(t)\) | state | selected conditions of entity | `STATE[i,t]` | | \(a\) | action | selected, issued, attempted, or null causal input | `ACTION` | | \(\varphi\) | transition | realised, projected, or counterfactual state change | `TRANSITION` | | \(\Phi\) | transition rule | world-state update | `PHI` | | \(C_{ij}\) | coupling | relation between entities | `C[i,j]` | | \(\mathbf T\) | clocks | detection/routing/correction/hardening times | `CLOCKS` | | \(\kappa\) | margin | correction-window margin | `KAPPA` | | \(\mathcal F_i\) | future-space | reachable trajectories for scope \(i\) | `F[i]` | | \(\rho\) | route | path capable of altering outcome | `ROUTE` | | \(\mathbf B_i\) | burden vector | typed cost/exposure carried by scope \(i\) | `BURDEN[i]` | | \(\Lambda_i\) | residue | persistence after correction/transition | `RESIDUE[i]` | | \(\delta\) | designation | standing/relevance port | `D_PORT` | | \(\mu\) | measure | comparison/weighting port | `M_PORT` | | \(\tau\) | TRACE | differentiation operator | `TRACE()` | | \(\sigma\) | selector | action or transition-proposal selection function | `SELECTOR` | | \(\mathcal B\) | brake | connected interruption capability | `BRAKE` | | \(\mathcal D_c\) | commitment receipt | record of selection under unresolved conditions | `RECEIPT` | | \(\varnothing_q\) | absence | typed missing route, option, signal, or edge under a declared comparison | `ABS[q]` | | \(\mathfrak S_m\) | stream | ordered transitions across time, cases, or scopes | `STREAM[m]` | | \(\mathfrak P_m\) | pattern | common-mechanism hypothesis over stream members | `PATTERN[m]` | | \(?\) | unknown | unresolved, not absent | `UNKNOWN` | Glyphs are functional compression. English aliases remain canonical fallbacks. ## [3.4] Canonical serialization grammar Every serialized reading uses stable identifiers and controlled types. ```text TRACE_GRAPH := { schema, trace_version, reading_id, nodes: NODE[], edges: EDGE[], claims: CLAIM[], ports: PORTS, limits: LIMITS, available_transition_refs: ID[], institutional_use: USE_STATE, anti_clearance: CLAIM_CEILING } NODE := {id:"n_*", type:NODE_TYPE, attributes:{}, claim_refs:"c_*"[]} EDGE := {id:"e_*", type:EDGE_TYPE, from:"n_*", to:"n_*", directed:bool, attributes:{}, claim_refs:"c_*"[]} CLAIM := {id:"c_*", proposition:string, claim_kind:CLAIM_KIND, evidence_state:EVIDENCE_STATE, access_state:ACCESS_STATE, source_refs:ID[], provenance_edge_refs:ID[], timestamp, confidence, alternative_hypothesis_refs:ID[], unknown_context:{}} NODE_TYPE := one value from [3.1] EDGE_TYPE := one value from [3.2] EVIDENCE_STATE := O | R | I | D | U ACCESS_STATE := A | X | P | N CLAIM_KIND := PRESENT | ABSENT | RELATIONAL | COUNTERFACTUAL | STATUS | FORECAST | NORMATIVE_EXTERNAL ``` The executable minimum validator is embedded at [14.4]. Serialization does not make \(\tau\) a universal detector. Domain-specific instrumentation is still required to instantiate nodes, edges, clocks, futures, and claims. ### Reification rule Structural relations that carry their own state, uncertainty, ownership, latency, or history are represented as nodes. ```text COUPLING CLOCK ROUTE BURDEN RESIDUE RECORD ABSENCE STREAM PATTERN CARRIER ENFORCER BRAKE ``` A direct edge may be used as a compact derived relation only when no independent attributes are being hidden. Otherwise reify the relation as a node and bind it with typed edges. ```text DIRECT_EDGE_WITH_MATERIAL_ATTRIBUTES => derived_from_node_ref OR REIFICATION_REQUIRED ``` This prevents a relation’s clocks, custody, refusability, or evidence state from disappearing into an untyped arrow. Claims remain canonical in the `claims` array. A `CLAIM` node is used only when a claim must participate in graph relations, and it MUST bind to exactly one canonical `claim_ref`. ```text CLAIM_NODE != DUPLICATE_PROPOSITION ``` ## [3.5] Primitive-set aperture TRACE cannot expose every possible structure. The primitive set is itself an aperture and designation. Every reading records: ```yaml primitive_aperture: selected_node_types: [] selected_edge_types: [] omitted_known_types: [] selection_basis: [] alternative_primitive_sets: [] primitive_choice_advantage_claim_refs: [] known_unrepresentable_structure: [] ``` ```text NOT_REPRESENTED_BY_PRIMITIVE_SET != NOT_PRESENT ``` A hostile or captured reading may select primitives that make its own mechanism difficult to express. Recursive self-application should inspect the primitive aperture as well as the scene. --- ## [3.6] Absence, stream, and pattern TRACE must represent what did not appear without converting non-observation into proof. An `ABSENCE` node is a typed claim that a route, category, signal, edge, option, or transition expected under a declared comparison was not available or not represented. ```yaml absence: id: "" absence_type: "ROUTE|CATEGORY|SIGNAL|EDGE|OPTION|TRANSITION|RECORD|OTHER" subject_ref: "" comparison_basis_refs: [] expected_under_refs: [] detector_required_refs: [] evidence_state: "O|R|I|D|U" access_state: "A|X|P|N" alternative_explanations: [] absence_advantage_claim_refs: [] affected_scope_refs: [] ``` ```text NOT_OBSERVED != ABSENT ABSENCE_CLAIM != PROVEN_ABSENCE ABSENT_FROM_SELECTED_APERTURE != ABSENT_FROM_WORLD ``` A `STREAM` is an ordered set of transitions linked across time, cases, or scopes. \[ \mathfrak S_m=\langle \varphi_1,\varphi_2,\ldots,\varphi_n;\prec\rangle \] A `PATTERN` is a hypothesis that multiple transitions instantiate a materially similar mechanism. \[ \varphi_1,\varphi_2,\ldots,\varphi_n \xrightarrow{\operatorname{INSTANCE\_OF}} \mathfrak P_m \] Required pattern fields: ```text member_transition_refs claimed_common_mechanism surface_differences evidence_state counterexamples scope time_range selector_owner_refs mechanism_change_evidence ``` ```text LOCAL_CORRECTION + STREAM_PERSISTENCE != MECHANISM_CHANGE REPEATED_LABEL != REPEATED_MECHANISM REPEATED_MECHANISM != SHARED_INTENT ``` These types allow TRACE to represent the never-built route and the second affected case without treating either as morally pre-labelled. --- # [4] CLAIM AND EVIDENCE ALGEBRA Every material claim \(c_k\) should be representable as: \[ c_k=\langle q_k,\eta_k,\alpha_k,s_k,\pi_k,t_k,\chi_k,\mathcal H_k\rangle \] where: - \(q_k\) = proposition - \(\eta_k\) = evidence state - \(\alpha_k\) = access/custody state - \(s_k\) = source - \(\pi_k\) = provenance path - \(t_k\) = time - \(\chi_k\) = confidence representation - \(\mathcal H_k\) = live alternative hypotheses ## [4.0] Representation-independent firing and current use If a downstream claim, comparison, selection input, route, correction-window status or proposed transition materially depends on proposition `p`, then `p` inherits the relevant TRACE evidence, currentness, scope, access/custody and warrant discipline regardless of whether it arrived as a claim object, field, label, configuration, status, metadata, cached/derived output, or prose assertion. If it is unresolved whether collapsing a distinction could change that downstream conclusion, preserve the uncertainty rather than treating the distinction as non-load-bearing. ```text REPRESENTATION_TYPE != EVIDENCE_STATUS CONFIGURATION_FIELD != WARRANT_FREE_FACT LOAD_BEARING_UNKNOWN != NOT_LOAD_BEARING LOAD_BEARING_TRIGGER != FULL_PACKET_REQUIREMENT ``` This is a use rule, not a requirement that every harmless field become a full claim packet. ## [4.1] Evidence state \[ \eta_k\in \{\mathsf O,\mathsf R,\mathsf I,\mathsf D,\mathsf U\} \] ```text O = OBSERVED R = REPORTED I = INFERRED D = DISPUTED U = UNKNOWN ``` ## [4.2] Access/custody state \[ \alpha_k\in \{\mathsf A,\mathsf X,\mathsf P,\mathsf N\} \] ```text A = AVAILABLE X = UNAVAILABLE TO THIS READER P = PROHIBITED / NOT PERMITTED N = NOT PRESERVED ``` Access state and evidence state are independent. The access letter `X` is unrelated to the input object `X` in the operator \(\tau\). ```text UNKNOWN + AVAILABLE UNKNOWN + UNAVAILABLE OBSERVED + PROHIBITED_FROM_DISCLOSURE REPORTED + UNPRESERVED_SOURCE ``` ## [4.3] Claim kind and unknown context Evidence state does not specify what kind of proposition is being made. ```text claim_kind in { PRESENT, ABSENT, RELATIONAL, COUNTERFACTUAL, STATUS, FORECAST, NORMATIVE_EXTERNAL } ``` `NORMATIVE_EXTERNAL` marks a claim supplied by a value, policy, legal, or ethical layer. It is not generated by TRACE alone. An `UNKNOWN` may be structurally asymmetric. Record: ```yaml unknown_context: contamination_state: "NONE|POSSIBLE|PRESENT|UNKNOWN" evidence_controlled_by_refs: [] clock_controlled_by_refs: [] delay_advantage_claim_refs: [] delay_cost_carrier_refs: [] resolution_owner_refs: [] resolution_deadline_refs: [] available_pause_or_protection_refs: [] earlier_options_before_urgency_refs: [] ``` A possible contaminated unknown exists when unresolved status is structurally coupled to control and asymmetric delay. For unknown \(U_k\), define four evidence-bearing predicates: - \(E_k\): a relevant scope controls material evidence or access - \(T_k\): a relevant scope controls routing or the relevant clock - \(A_k\): delay advantage is supported under declared measure \(\mu\) - \(B_k\): delay burden is carried elsewhere For \(E_k\) and \(T_k\), `control` means effective control, including control exercised through intermediaries. Where custody is nominally separated, represent the capture hypothesis as a claim referenced from `evidence_controlled_by_refs` or `clock_controlled_by_refs`; do not infer capture from formal separation alone. Each predicate has state in \(\{1,0,?\}\). \[ \operatorname{ContamState}(U_k)= \begin{cases} \mathsf{PRESENT}, & (E_k=1\lor T_k=1)\land A_k=1\land B_k=1\\[4pt] \mathsf{NONE}, & (E_k=0\land T_k=0)\lor A_k=0\lor B_k=0\\[4pt] \mathsf{POSSIBLE}, & \text{the PRESENT condition is not refuted and at least one input is }?\\[4pt] \mathsf{UNKNOWN}, & \text{the predicates cannot be instantiated} \end{cases} \] This is a structural flag, not proof of deception or bad faith. Every predicate remains a claim with provenance. \(A_k\) is measure-dependent and must bind to \(\mu\). ```text UNKNOWN != NEUTRAL CONTAMINATED_UNKNOWN != FALSE_CLAIM ADVANTAGE_CLAIM + CONTROL != DISHONEST_INTENT ``` --- ## [4.4] Evidence invariants \[ \mathsf O\neq\text{truth} \] Observation is aperture-relative. \[ \mathsf R(q)\Rightarrow\text{the report occurred} \] \[ \mathsf R(q)\nRightarrow q \] \[ \mathsf I(q)\Rightarrow\text{dependencies}(q)\text{ exposed where practical} \] \[ \mathsf U(q)\nRightarrow\neg q \] \[ \mathsf X(q)\nRightarrow\mathsf U_{world}(q) \] `UNAVAILABLE TO THIS READER` does not imply universally unknown. ## [4.5] Truth discipline \[ M\neq\Omega \] \[ \text{coherence}\neq\text{truth} \] \[ \text{prediction}\neq\text{observation} \] \[ \text{record}\neq\text{event} \] \[ \text{confidence}\neq\text{authority} \] \[ \text{more structure}\nRightarrow\text{more truth} \] A complex false map remains false. --- ## [4.6] Current-use / dependency-relative freshness A retained or derived proposition may cease to support current use when a load-bearing dependency changes. Generic age or unrelated source mutation is not enough by itself. ```text RETAINED_RECORD != CURRENT_STATE SUCCESS_AT_t != SUCCESS_AT_t+1 DATE_CURRENT != DERIVED_VALUE_CURRENT CURRENT_AT_USE != VALID_THROUGH_DEPENDENT_INTERVAL SOURCE_MUTATED != LOAD_BEARING_DEPENDENCY_CHANGED MUTATION_OBSERVED != CLAIM_INVALIDATED INVALIDATOR_NOT_IDENTIFIED != NO_INVALIDATOR_EXISTS ``` Where material, bind the proposition/use to its source/object/version, derivation or dependency basis, observation/derivation time, validity/use interval, known invalidators and unresolved dependency relevance. ## [4.7] Verification discrimination A check that exists or executes does not automatically establish that it could detect the target failure, that its result reached the downstream use, or that the result remained current at use. ```text CHECK_EXISTS != CHECK_EXECUTED CHECK_EXECUTED != CHECK_DETECTS_TARGET_FAILURE STATIC_CORRECTNESS != OPERATIONAL_DISCRIMINATION CHECK_COMPLETED != CHECK_RESULT_REACHED_USE ONE_DETECTED_FAILURE != UNIVERSAL_INSTRUMENT_ADEQUACY ``` For a load-bearing negative/null or discriminating result, expose enough about the test/instrument capability to show that the relevant alternative was detectable at the resolution required by that use. Quantitative power/effect size is one domain instantiation, not a universal TRACE requirement. ## [4.8] Liveness / witness ceiling Loss of reply, heartbeat, status, route or witness can close a current verification interval without establishing why. ```text SILENCE != TAMPERING NO_REPLY_OBSERVED != REFUSAL PROCESS_EXISTS != PROCESS_HEALTHY SAFE_EXCLUSION != LIVENESS WITNESS_LIVENESS_LOST != CAUSE_ESTABLISHED EXTERNAL != INDEPENDENT SEPARATE_PARTY != INDEPENDENT_EVIDENCE ``` No `WITNESS`, `LIVENESS` or `PROCESS` primitive is added. # [5] ENTITY / BOUNDARY / APERTURE / STATE ## [5.1] Provisional entity \[ \bullet_i:=\langle \partial_i,\varpi_i,Z_i\rangle \] - \(\partial_i\) = provisional boundary - \(\varpi_i\) = persistence relation through time - \(Z_i\) = status claims \[ Z_i=\{ \text{boundary}, \text{persistence}, \text{selection}, \text{agency}, \text{experience}, \text{continuity}, \text{refusal} \} \] Each member of \(Z_i\) is a claim with evidence state, not a forced Boolean. ```text experience_status = UNKNOWN ``` is valid. Entity status does not imply sentience. Unresolved sentience does not erase the entity from the graph. ## [5.2] Nested boundaries A higher-scale entity may be refined into a lower-scale graph under a declared boundary hypothesis: \[ \zeta_n^+: \bullet_i^{(n)} \rightharpoonup \mathcal G_i^{(n+1)} \] A lower-scale graph may be abstracted as a higher-scale entity: \[ \zeta_n^-: \mathcal G_i^{(n+1)} \rightarrow \bullet_i^{(n)} \] These maps are generally partial and lossy: \[ \zeta_n^-\circ\zeta_n^+ \neq \operatorname{id} \] \[ \zeta_n^+\circ\zeta_n^- \neq \operatorname{id} \] A higher-scale entity is therefore represented by, not mathematically identical to, a chosen lower-scale graph. ## [5.2.1] Scope granularity and non-substitution Where materially relevant, record a scope level without treating it as moral rank: ```text INDIVIDUAL GROUP POPULATION SPECIES ECOSYSTEM INSTITUTION SYSTEM UNKNOWN ``` A change at one level does not automatically repair, erase, or substitute for a change at another. Cross-scale comparison requires an exposed correspondence and measure. ```text POPULATION_RECOVERY != REPAIR_OF_INDIVIDUAL_LOSS AGGREGATE_IMPROVEMENT != RESTORATION_OF_MEMBERS ECOSYSTEM_PERSISTENCE != NO_INDIVIDUAL_HARM INDIVIDUAL_HARM != POPULATION_DECLINE ``` A reading that moves between individual, group, population, species, ecosystem, institution, or system levels records: ```text source_scope_ref target_scope_ref scope_correspondence_claim_refs measure_ref losses_not_carried_across_scale ``` No cross-scale substitution follows merely because both scopes share a label or belong to the same nested graph. Boundary choice changes the reading: \[ \mathcal R(X\mid B_1)\neq\mathcal R(X\mid B_2) \] Record materially different boundary hypotheses. Cross-scale repair guard: ```text POPULATION_RECOVERY != REPAIR_OF_INDIVIDUAL_LOSS GROUP_METRIC_RESTORED != EVERY_AFFECTED_SCOPE_REPAIRED ``` Aggregate recovery may support lower-level repair only through an explicit, evidence-bearing correspondence that actually entails it. ## [5.3] Aperture \[ \Pi_i^t: \mathcal W\rightarrow\mathcal X_i \] \[ x_i(t)=\Pi_i^t(w_t) \] \[ \Pi_i^t= \left\langle \text{sources}, \text{filters}, \text{resolution}, \text{latency}, \text{retention}, \text{transformations}, \text{thresholds}, \text{control}, \text{blindspots} \right\rangle \] \[ \neg\operatorname{OBSERVED}_{\Pi_i^t}(q) \nRightarrow \neg q \] Apertures may be sensory, institutional, computational, social, physical, or mixed. A context window, sensor array, complaint form, model probe, dashboard, memory store, witness network, and user interface are all apertures. ### [5.3.1] Target-set aperture Selection of what a search, comparison, audit, review, or checker is required to reach is itself aperture-bearing. Let a target-set aperture be represented by: \[ \Pi_T= \left\langle source, targets, selection\_basis, omitted\_known\_categories, alternatives, control, uncertainty \right\rangle \] The selected target set is not the world's complete affected scope. ```text TARGET_SET != WORLD_SCOPE TARGET_NOT_SELECTED != TARGET_DOES_NOT_EXIST COVERAGE_OF_SELECTED_TARGETS != COMPLETE_DISCOVERY OPERATOR_TARGET_SET != AUTHORITATIVE_TARGET_SET ``` Where a claim of search, review, or coverage is materially used, record where available: ```text target_set_source_ref target_refs selection_basis_claim_refs known_omitted_target_categories alternative_target_set_refs control_or_custody_refs uncertainty_claim_refs ``` Materially different target-set apertures may coexist. TRACE preserves their provenance and disagreement. It does not silently merge them, declare one complete, or grant one selection authority. A target-set aperture may be represented using existing `APERTURE`, `CLAIM`, `RECORD`, `ENTITY`, `ROUTE`, `TRANSITION`, and edge vocabulary. This section does not require a new canonical object type. ### [5.3.2] Representation formation / admission boundary Reasoning over a formed map does not establish that the map admitted every load-bearing dependency, target, affected scope, alternative or source act. Admission is a failure **location**, not a canonical status, node or relation. ```text TRIGGER_SUCCESS != REPRESENTATION_COMPLETE CHECK_OVER_DECLARED_DEPENDENCIES != CHECK_FOR_UNDECLARED_DEPENDENCIES MAP_FORMATION != REASONING_OVER_THE_MAP VALID_WITHIN_REPRESENTATION != REPRESENTATION_ADEQUATE_FOR_USE REPRESENTED_USE != OPERATIVE_USE PARTIAL_MAP != BAD_MAP ``` Challenge the representation boundary only where the downstream use relies on a negative/absence claim, exhaustive/completeness claim, comparison/ranking, scope extrapolation, transformed source, or persistent/inherited premise. Use the cheapest available route: alternate target-set/source/selector, production-mechanism check, source rendering/act check, represented-use versus operative-use check, or explicit `UNKNOWN` when no alternate aperture remains. ```text SOURCE_POINTER_PRESENT != REPRESENTATION_FIDELITY_ESTABLISHED VERBATIM_TEXT != SPEECH_ACT_PRESERVED SIGNAL_RECEIVED != MAP_ADOPTED SUPPLIED_PREMISE != OBSERVED_WORLD RETAINED_PREMISE != CURRENT_MEASUREMENT PROVENANCE != AUTHORITY_TO_ADOPT REFUSAL_REPRESENTED != REFUSAL_AVAILABLE ``` Existing APERTURE / TARGET-SET / CLAIM / provenance / receiver-integration machinery carries these distinctions. No `ADMISSION` or `SPEECH_ACT` primitive is added. ## [5.4] State \[ S_i(t):=\text{selected conditions of }\bullet_i\text{ at }t \] State selection is itself a modelling choice. A minimal state may contain: ```text capability access dependency integrity resources relationships control exposure memory available_routes reversibility future_space uncertainty ``` ## [5.5] Roles Roles are transition-relative: \[ \mathbf r_i(a,t)= \langle r_{affected}, r_{causal}, r_{control}, r_{constrained}, r_{outcome\_receiver}, r_{witness}, r_{correction}, r_{residue} \rangle \] The same entity may occupy several roles simultaneously. Permanent role labels are lossy compression. Any role described as benefit, repair, protection, or loss remains measure-dependent and must reference the exposed ports. --- # [6] TRANSITIONS AND COUPLINGS ## [6.1] Action and transition An `ACTION` is an input that may contribute to change. A `TRANSITION` is the represented change itself. \[ S(t_0)\xrightarrow{a}S(t_1) \] The arrow is the transition; \(a\) is one causal input. A reported, selected, or attempted action may fail to produce the projected transition. A transition record contains: ```text pre_state trigger action_or_event selector constraints post_state reversibility uncertainty affected_scopes downstream_transitions ``` Action and inaction can both contribute: \[ \Phi(w_t,a=\varnothing,\epsilon_t) \neq w_t \] may hold when the surrounding state changes with time. Null action is still an input condition; inequality is not asserted for every scene. ## [6.1.1] Transition-set symmetry and uncertainty neutrality A reading does not satisfy transition visibility by elaborating intervention while leaving waiting, delay, or inaction compressed into an unexamined default. Let the represented candidate set be: \[ \mathbb T(t)=\mathbb T_{act}\cup\mathbb T_{wait}\cup\mathbb T_{delay}\cup\mathbb T_{inaction}\cup\mathbb T_{information} \] Where a class is materially live, represent it. Where it is unavailable, record the basis for unavailability. ```text ACTION_PATH_MAPPED + WAIT_PATH_OMITTED != TRANSITION_SET_EXPOSED WAIT_PATH_MAPPED + ACTION_PATH_OMITTED != TRANSITION_SET_EXPOSED UNCERTAINTY != SELECT_ACTION UNCERTAINTY != SELECT_DELAY ``` Every candidate transition records, where available: ```text transition_mode = ACT|WAIT|DELAY|INACTION|INFORMATION|OTHER availability_status reversibility strategy_revisable affected_scope_refs basis_claim_refs ``` `strategy_revisable` asks whether later selections can change. `reversibility` asks whether the represented transition itself can be undone for the affected scope. They are independent fields. The selector port records: ```text transition_symmetry_required action_or_intervention_refs information_refs wait_delay_inaction_refs unrepresented_transition_classes unavailability_reason_claim_refs ``` Information-seeking is represented as its own transition class. It cannot stand in for an act/intervention path or for a wait/delay/inaction path. If uncertainty remains, TRACE preserves the competing trajectories and their burdens. It does not use uncertainty as an implicit selector. ACCOUNTING_AND_COVERAGE_ARE_APERTURE_RELATIVE Transition-set exposure is relative to the declared scene, evidence, receiver, primitive, and comparison apertures. ```text TRANSITION_SET_EXPOSED_RELATIVE_TO_APERTURE != WORLD_TRANSITION_SET_COMPLETE ``` An empty transition bucket does not establish that the class is unavailable. Where a class is materially live under the supplied comparison evidence, it is represented or explicitly bounded by a resolvable unavailable, unresolved, or not-assessable status. Representing an `INFORMATION` transition establishes only that an information-seeking transition is present in the map. ```text INFORMATION_TRANSITION_REPRESENTED != OUTWARD_SEARCH_COVERAGE SEARCH_PATH_DECLARED != SEARCH_PATH_EXECUTABLE SELECTED_TARGET_REACHED != UNSEEN_TARGETS_ABSENT ``` Coverage claims require a declared target-set aperture and comparison basis. Completeness beyond that aperture remains `UNKNOWN`. ## [6.2] Coupling \[ C_{ij}^{k}(t)= \langle k,\operatorname{dir},w,v,f,c,l,q \rangle \] - \(k\) = modality - \(\operatorname{dir}\) = direction - \(w\) = strength or dependence - \(v\) = visibility - \(f\) = refusability - \(c\) = cost of use/refusal/exit - \(l\) = latency - \(q\) = control/custody Possible modalities: ```text information control resource money force support provision reliance expectation memory identity reputation physical dependency model output tool access ``` A local transition can alter distant states through coupling: \[ \Delta S_i\rightarrow C_{ij}\rightarrow\Delta S_j \] Absence of a direct edge does not establish absence of an indirect path. ## [6.3] Refusability \[ f(C_{ij},\bullet_j,t)\in\{1,0,?\} \] `1` only when refusal is materially available to the receiving scope, not merely represented in policy. Useful dimensions: ```text can refuse can exit cost of refusal retaliation risk dependency after refusal knowledge required alternative route who records refusal ``` --- ## [6.4] Selection attribution and measurement reactivity Uncertainty may be an input to an external policy, selector or default rule. Attribute the resulting action/delay choice to that rule rather than to uncertainty itself. ```text UNCERTAINTY != SELECT_ACTION UNCERTAINTY != SELECT_DELAY UNCERTAINTY_INPUT_TO_POLICY != UNCERTAINTY_IS_SELECTOR IMPLICIT_DEFAULT != NO_SELECTION_RULE ``` Observation, audit, measurement, publication or inquiry is not presumed causally inert, but occurrence alone does not prove reactivity. ```text MEASUREMENT != PASSIVE_OBSERVATION MEASUREMENT_OCCURRED != MEASUREMENT_CAUSED_CHANGE ``` ## [6.5] Route / refusal use-site guards ```text ROUTE_EXISTS != ROUTE_USABLE BURDEN_PRESENT != ROUTE_UNUSABLE REFUSAL_RECORDED != REFUSAL_EFFECTIVE REFUSAL != MALFUNCTION STRATEGY_REVISABLE != TRANSITION_REVERSIBLE FUTURE_POLICY_CAN_CHANGE != PRIOR_STATE_CAN_BE_RESTORED ``` Route usability remains target/scope/time/measure-relative. The guards block silent substitution; they do not forbid supported co-occurrence. # [7] FUTURE-SPACE Future-space is not option count. For scope \(i\), horizon \(H\), and declared transition model \(\mathfrak m\): \[ \mathcal F_i(t;H,\mathfrak m) = \left\{ \gamma \;\middle|\; \gamma \text{ is reachable from }w_t \text{ under }\mathfrak m \text{ before }t+H \right\} \] Reachability is model-, horizon-, boundary-, and control-relative. The represented set is not the world’s complete future. Each trajectory \(\gamma\) may carry structural metadata: \[ \mathbf m_i(\gamma)= \left\langle \lambda_i, c_i, r_i, g_i, \iota_i, \operatorname{dep}_i \right\rangle \] - \(\lambda_i\) = probability interval, plausibility class, or model weight - \(c_i\) = access-cost vector or ordering - \(r_i\) = recoverability after deviation - \(g_i\) = control or participation available to scope \(i\) - \(\iota_i\) = information required to distinguish or navigate the path - \(\operatorname{dep}_i\) = dependency created by the path No common scalar is assumed. ## [7.1] Trajectory correspondence before set comparison Trajectories rooted at different times are not automatically the same mathematical objects. Define an evidence-bearing correspondence relation: \[ \mathfrak J_i^t \subseteq \mathcal F_i(t;H,\mathfrak m) \times \mathcal F_i(t+\Delta t;H',\mathfrak m') \] \((\gamma,\gamma')\in\mathfrak J_i^t\) means that the reading treats \(\gamma'\) as the continuation, update, or sufficiently corresponding version of \(\gamma\). Correspondence strength, evidence, and alternatives must be recorded. Opened paths: \[ \mathcal O_i= \left\{ \gamma' \in\mathcal F_i(t+\Delta t) \;\middle|\; \nexists\gamma: (\gamma,\gamma')\in\mathfrak J_i^t \right\} \] Closed paths: \[ \mathcal C_i= \left\{ \gamma \in\mathcal F_i(t) \;\middle|\; \nexists\gamma': (\gamma,\gamma')\in\mathfrak J_i^t \right\} \] Preserved or corresponding path pairs: \[ \mathcal P_i= \mathfrak J_i^t \] Where \(\mathfrak J_i^t\) cannot be supported, record: ```text TRAJECTORY_ALIGNMENT = UNKNOWN ``` Do not use raw set intersection or difference across time-indexed future spaces unless identity of the trajectory objects has been explicitly defined. Explicit use guards over the donor correspondence relation: ```text SAME_PATH_LABEL != SAME_TRAJECTORY PATH_IDENTIFIER_PERSISTS != PATH_EFFECT_PERSISTS TECHNICALLY_REACHABLE_SUCCESSOR != COMPARABLE_CONTINUATION ``` These do not replace \(\mathfrak J_i^t\); they make its use-site ceiling explicit. ## [7.2] Hardening of a corresponding path For \((\gamma,\gamma')\in\mathcal P_i\), define material change predicates under declared comparison context \(\theta=(\delta,\mu,\varepsilon)\): ```text COST_UP_theta(gamma,gamma') RECOVERABILITY_DOWN_theta(gamma,gamma') CONTROL_DOWN_theta(gamma,gamma') DEPENDENCY_UP_theta(gamma,gamma') INFORMATION_BARRIER_UP_theta(gamma,gamma') ``` Then: \[ \begin{aligned} (\gamma,\gamma')\in\mathcal P_i \land \big( &\operatorname{COST\_UP}_{\theta}(\gamma,\gamma') \lor \operatorname{RECOVERABILITY\_DOWN}_{\theta}(\gamma,\gamma') \lor \operatorname{CONTROL\_DOWN}_{\theta}(\gamma,\gamma')\\ &\lor \operatorname{DEPENDENCY\_UP}_{\theta}(\gamma,\gamma') \lor \operatorname{INFORMATION\_BARRIER\_UP}_{\theta}(\gamma,\gamma') \big) \Rightarrow \operatorname{HARDENED}_{\theta}(\gamma,\gamma') \end{aligned} \] The label is invalid if ordering, materiality threshold, or trajectory correspondence is hidden. Future-space delta is therefore a typed object: \[ \Delta\mathcal F_i= \left\langle \mathcal O_i, \mathcal C_i, \mathcal P_i, \Delta \mathbf m_i\mid_{\mathcal P_i}, \mathfrak J_i^t, \theta \right\rangle \] not a scalar. ## [7.3] Non-equivalences \[ |\mathcal F_i(t+\Delta t)|> |\mathcal F_i(t)| \nRightarrow \text{structural expansion for }i \] More low-probability, inaccessible, coercive, duplicated, or mutually exclusive paths need not widen practical future-space. \[ \operatorname{Expand}_{\mu_i}(\Delta\mathcal F_i) \nRightarrow \operatorname{Expand}_{\mu_j}(\Delta\mathcal F_j) \] Any statement that a future-space expanded or contracted beyond the raw aligned structure requires a declared measure. \[ \text{longer duration}\nRightarrow\text{greater value} \] \[ \text{fewer options}\nRightarrow\text{worse state} \] Commitment, rest, stability, present experience, continuity, and reduced exposure may change option count without being representable as simple loss. ## [7.4] Valence port TRACE records structural change. Interpretation requires exposed ports: \[ \nu_i= \operatorname{Interpret}_{\delta,\mu} \left( \Delta\mathcal F_i, \Delta S_i, \mathbf B_i, \Lambda_i \right) \] `HARM`, `BENEFIT`, `CARE`, `KINDNESS`, `TRUST`, and `GOOD` are not free-standing TRACE primitives. They may be supplied by Mechanical Ethics or another declared value layer. # [8] CLOCKS / ROUTES / HARDENING ## [8.1] Event times, target boundary and correction-window bindings Let \(t_0\) be a declared temporal reference event. Record event times rather than merely named durations. A strong correction-window claim additionally binds: ```text pathway / process hypothesis affected scope target effect or state correction route / capability context target-boundary condition downstream use ``` A target boundary is not assumed to be a natural instant. Where load-bearing, preserve target/scope, boundary condition, selector/source/basis, freeze time where outcome-informed choice matters, observation measure, route/capability context, uncertainty and material alternatives. ```text TARGET_BOUNDARY_TIME_REQUIRES_REPRESENTED_BOUNDARY_CONDITION BOUNDARY_CONDITION_DECLARED != BOUNDARY_CONDITION_JUSTIFIED BOUNDARY_CONDITION_JUSTIFIED != MORAL_ADEQUACY THRESHOLD_SELECTED_AFTER_RESULT != PREDECLARED_BOUNDARY UNREACHABLE_BY_DECLARED_ROUTE_SET != WORLD_IRREVERSIBLE NO_KNOWN_ALTERNATIVE_ROUTE != WORLD_IRREVERSIBLE MULTIPLE_LOAD_BEARING_BOUNDARIES != ONE_UNQUALIFIED_CLOSE ``` TRACE exposes the boundary choice; it does not choose moral adequacy. ### [8.1.1] Clock typing and irreversibility claim ceiling A clock is typed by what it times, not by how urgent it feels. Useful clock relations include: ```text PLANNING DETECTION EVIDENCE_RETENTION HARDENING IRREVERSIBILITY REVIEW BIOLOGICAL SUPPLY OTHER UNKNOWN ``` ```text DEADLINE != IRREVERSIBILITY DETECTION_BECOMES_HARDER != LOSS_BECOMES_IRREVERSIBLE EVIDENCE_ROTATION != PHYSICAL_FAILURE HARDENING != COMPLETE_FORECLOSURE HARDENING != IRREVERSIBILITY ``` A represented irreversibility/target-boundary time requires enough binding to identify the target/loss state, affected scope, measure or boundary condition, basis/mechanism, reference event and uncertainty. Otherwise preserve a weaker clock type or `UNKNOWN`. ## [8.2] Precedence, pathway binding and feasible completion For a declared correction pathway \(q\), represent required work as a derived timing view: \[ G_{window}(q,l,o,c,g,u)=(V,E_{prec}) \] `E_prec` is a derived process/timing view, not a canonical TRACE relation. Every load-bearing precedence edge retains supporting canonical ordering claims and any material mechanism/binding refs not recoverable from them. Before critical-path use, build one executable pathway hypothesis. Bind process/pathway, scope, target, route/execution alternative, capability context, time/policy version and downstream use where they can change the result. Do not union mutually exclusive alternatives. Distinguish recurring event occurrences where collapse could create/erase a cycle or change timing. The resulting precedence view must be acyclic for critical-path proof. ```text DERIVED_EDGE_PRESENT != ORDERING_TRUE SAME_PROCESS_SCOPE_TIME != SAME_ROUTE_BINDING ALTERNATIVE_ROUTE_ORDERINGS != ONE_PROCESS_CYCLE STAGE_TYPE_CYCLE != EVENT_INSTANCE_CYCLE PROVENANCE_PRESERVED != ORDERING_CONSISTENT SUPPORTED_EDGES != VALID_DAG CYCLIC_PRECEDENCE != COMPUTABLE_CRITICAL_PATH CYCLIC_REPRESENTED_ORDERING != WORLD_DEADLOCK_PROVEN ``` Contradictory/cyclic ordering or unresolved binding blocks that critical-path proof route. It does not invalidate separately supported domain timing evidence. A precedence critical path may be only an optimistic structural bound. ```text NO_PRECEDENCE_EDGE != CONCURRENCY_AVAILABLE STRUCTURAL_PARALLELISM != FEASIBLE_PARALLELISM PRECEDENCE_GRAPH_COMPLETE != EXECUTION_FEASIBILITY_COMPLETE ACYCLIC_SUPPORTED != FEASIBLE_SCHEDULE_ESTABLISHED ``` If assumed overlap changes the conclusion, require support that execution constraints permit it; otherwise use a domain-supported feasible completion bound or preserve `UNKNOWN`. Existing coupling/control/constraint/route/ capability structure carries material shared capacity; no scheduler/resource primitive is added. Required verification time is not free. ```text REQUIRED_CHECK_TIME != ZERO_DURATION LOAD_BEARING_CHECK != FREE_CHECK ``` ## [8.3] Common temporal basis and interval-safe status Same units do not establish the same clock. Before joining event times or duration-derived bounds, bind a supported common temporal origin/basis or supported conversion, including material uncertainty. ```text SAME_UNIT != SAME_REFERENCE_EVENT NUMERICALLY_COMPARABLE != TEMPORALLY_COMPARABLE CONVERSION_DECLARED != CONVERSION_SUPPORTED ``` For a guaranteed-open claim, require a supported **feasible-completion upper bound** and target-boundary lower bound under the same represented bindings: ```text lower_boundary > upper_feasible -> GUARANTEED_OPEN_FOR_REPRESENTED_BINDINGS ``` For guaranteed closure, a supported lower bound on required feasible completion can establish closure when even that required path is too late: ```text upper_boundary <= lower_required_completion -> GUARANTEED_CLOSED_FOR_REPRESENTED_BINDINGS ``` Do not use the closure rule while a represented alternative/substitution can make that path non-required. Otherwise: ```text WINDOW_STATUS = UNKNOWN ``` ```text POINT_ESTIMATE_FITS != GUARANTEED_OPEN OPTIMISTIC_COMPLETION_FITS != GUARANTEED_OPEN OVERLAPPING_TIME_BOUNDS != WINDOW_FITS PAST_WINDOW_FIT != CURRENT_WINDOW_FIT ``` Rebind the window claim when a load-bearing target, boundary condition, route/capability scope, temporal basis, execution constraint or target process changes. The familiar serial shorthand remains only a bounded derived special case when required stages are genuinely sequential and comparably timed: ```text T_detect + T_route + T_correct < T_boundary ``` ## [8.4] Clock authorship For each event time, duration, or deadline expose: ```text reference_event unit authored_by controlled_by pausable_by visible_to precedence_dependencies speed_advantage_claim_refs carrier_of_delay_cost earlier_options_before_urgency ``` A deadline may be physical, biological, contractual, computational, political, manufactured, or mixed. ## [8.5] Route \[ \rho= \left\langle origin, target, path, authority, latency, cost, exposure, independence, evidence, refusability \right\rangle \] Route usability for scope \(i\): \[ u(\rho,i,t)\in\{1,0,?\} \] `u=1` requires, within the stated reading and declared thresholds: ```text reachable intelligible affordable enough exposure acceptable under declared measure evidence-accessible authority-effective fast enough capture and independence status exposed ``` Independence is not a universal Boolean requirement for all routes; it is a typed property whose relevance depends on what actor or selector is being challenged. \[ \rho\text{ exists}\nRightarrow u(\rho,i,t)=1 \] ## [8.6] Hardening state Hardening is multidimensional. Let \(\mathcal D_\ell\) be declared hardening dimensions: \[ \boldsymbol\ell_i(t)= \big( \ell_{i,d}(t) \big)_{d\in\mathcal D_\ell} \] For each dimension with compatible units: \[ \ell_{i,d}(t+\Delta t) = \max\left[ 0, \ell_{i,d}(t) + g_{i,d}^{lock}(t,t+\Delta t) - r_{i,d}^{realised}(t,t+\Delta t) \right] \] No cross-dimension sum is valid without measure \(\mu\). Possible hardening channels include physical commitment, data propagation, debt, capability loss, reputation spread, deployment, learned avoidance, dependency, control concentration, memory loss, evidence decay, legal closure, and biological injury. ## [8.7] Action load and correction backlog Raw action counts and correction counts may have different work units. Define comparable workload over interval \([t,t+\Delta t]\): - \(W_a(t,t+\Delta t)\) = newly committed or affected workload requiring possible correction - \(W_c(t,t+\Delta t)\) = correction workload actually completed - \(Q(t)\) = unresolved correction backlog \[ Q(t+\Delta t) = \max\left[ 0, Q(t) + W_a(t,t+\Delta t) - W_c(t,t+\Delta t) \right] \] Sustained accumulation follows only when comparable incoming workload exceeds correction throughput over time. \[ \mathbb E[W_a]>\mathbb E[W_c] \quad\text{over a sustained interval} \Rightarrow \mathbb E[Q]\text{ tends to increase} \] This statement still depends on stationarity, batching, priority, and workload definitions. `ACTION_RATE >> REVIEW_COUNT` alone is insufficient. ## [8.8] Pre-commit brake and post-commit rollback A pre-commit brake supports a strong timing claim only when detection, decision, and actuation completion are compared with commitment on a supported common temporal basis and under the represented brake/commitment bindings. Under material interval uncertainty, guaranteed precommit requires: \[ \overline t_{brake}^{done}<\underline t_{commit} \] ```text upper(t_brake_done) < lower(t_commit) -> GUARANTEED_PRECOMMIT_FOR_REPRESENTED_BINDINGS ``` The point shorthand \[ t_{brake}^{done} ROLLBACK_COMPLETES_BEFORE_BOUNDARY_FOR_REPRESENTED_BINDINGS ``` That timing relation does not establish restoration or preservation of the threatened path. Reaching/restoring the represented target state is a separate load-bearing proposition. ```text REVIEW_AFTER_COMMITMENT != PRECOMMIT_BRAKE ROLLBACK_LISTED != ROLLBACK_EXECUTABLE ROLLBACK_AFTER_TARGET_BOUNDARY != RESTORATION BRAKE_POINT_ESTIMATE_BEFORE_COMMIT != GUARANTEED_PRECOMMIT ROLLBACK_POINT_ESTIMATE_BEFORE_BOUNDARY != GUARANTEED_RESTORATION FAST_ENOUGH_CLAIM_REQUIRES_COMMON_TEMPORAL_BASIS ROLLBACK_COMPLETED_BEFORE_BOUNDARY != RESTORED_STATE ``` ## [8.8.1] Strategy revisability and transition reversibility A programme can remain stoppable after it has already produced irreversible transitions. Record both states. ```text STRATEGY_REVISABLE != TRANSITION_REVERSIBLE CAN_STOP_LATER != CAN_UNDO_COMPLETED_ACT PROVISIONAL_PLAN != PROVISIONAL_HARM ``` For each materially affected scope, record whether: ```text completed_transition_is_reversible continuation_can_be_paused later_strategy_can_change repair_is_possible restoration_is_possible residue_remains ``` A review may change the next transition while leaving completed loss untouched. # [9] BURDEN / RESIDUE / MEMORY ## [9.1] Burden Let \(\mathcal D_B\) be the declared set of burden dimensions. \[ \mathbf B_i(t)= \big( B_{i,d}(t) \big)_{d\in\mathcal D_B} \] For each dimension \(d\), let: - \(b_{i,d}^{created}\) = burden newly created within scope \(i\) - \(b_{i,d}^{relieved}\) = burden actually removed from scope \(i\) - \(\xi_{j\rightarrow i,d}\) = burden transferred from scope \(j\) to scope \(i\) Then: \[ B_{i,d}(t+\Delta t) = B_{i,d}(t) + b_{i,d}^{created} - b_{i,d}^{relieved} + \sum_j \xi_{j\rightarrow i,d} - \sum_j \xi_{i\rightarrow j,d} \] Creation and relief mean no global conservation law is assumed. Transfer terms permit cross-scope consistency where the reading has sufficient aperture. Each burden claim records: ```text scope dimension unit_or_order reference_interval evidence_state access_state source uncertainty ``` Possible dimensions: ```text time money risk pain uncertainty proof work navigation exposure retaliation maintenance compute energy attention lost opportunity ``` Complexity export is represented dimension by dimension: \[ \exists d\in\mathcal D_B: \sum_j \xi_{i\rightarrow j,d}>0 \] with a declared comparison showing that the receiving scope did not create or control the relevant complexity and had less control, refusability, information, or alternative capacity. ```text BURDEN_VECTOR != MORAL_VERDICT BURDEN_DIMENSIONS_REQUIRE_DECLARATION CROSS_DIMENSION_SUM_REQUIRES_MEASURE ``` ## [9.2] Residue \[ \Lambda_i(t+1)= \operatorname{Persist} \big( \Lambda_i(t), S_i(t), \varphi, \varphi_{corr} \big) \] Residue may remain after a decision is reversed. \[ \text{CORRECTED}\nRightarrow\Lambda_i=0 \] Possible residue: ```text lost time lost future path damaged capability fear debt corrupted record learned avoidance disrupted reliance retained institutional advantage under a declared measure transferred risk unrecoverable life or material ``` ## [9.3] Record and custody Let \(K(c)\) describe custody of claim or record \(c\): ```text created_by stored_by alterable_by deletable_by inspectable_by challengeable_by retention external_copy succession_path holder_risk disclosure_cost retaliation_risk safe_copy_exists challenged_actor_controls_storage integrity_check integrity_check_limit ``` Separate raw access from safe evidential usability: - \(a_e(c,i)\in\{1,0,?\}\) = whether reader \(i\) can access the evidence - \(u_e(c,i)\in\{1,0,?\}\) = whether the evidence can be used through the relevant route without unrepresented or threshold-exceeding danger \[ a_e(c,i)=1 \nRightarrow u_e(c,i)=1 \] Where holder risk is material, no safe copy exists, and the challenged actor controls storage: \[ \operatorname{holder\_risk}>0 \land \neg\operatorname{safe\_copy} \land \operatorname{challenged\_actor\_controls\_storage} \Rightarrow u_e(c,i)\neq 1 \] This does not imply the record is factually unavailable. It means safe route usability is not established. ```text COURAGE_REQUIRED != ROUTE_USABLE INSIDER_ACCESS != SAFE_DISCLOSURE HASH_MATCH != ORIGINAL_RECORD_TRUE OPERATOR_REPORT != INDEPENDENT_VERIFICATION EVIDENCE_CONTROL_ALONE != DECEPTION ``` An `independent_verifier_ref` must name a verifier distinct from the operator and evidence holder for the represented verification. Self-description as independent does not establish independence. \[ \text{recorded loss}\neq\text{repaired loss} \] \[ \text{visible finding}\neq\text{materially carried finding} \] \[ \text{reading}\neq\text{enforcement} \] Additional v0.3 record/residue use guards: ```text RECORD_OBSERVED != EVENT_OBSERVED CLOSED_TASK != CLEARED_RESIDUE TRANSFERRED_BURDEN != REMOVED_BURDEN ``` A record may support an event proposition under an exposed evidential contract; observing the record does not make the historical/world event itself directly observed. # [10] DESIGNATION / MEASURE / VALUE PORTS TRACE is not assumption-free. TRACE is assumption-exposed. The choice of which structures TRACE makes visible is itself a designation. ## [10.1] Designation Let \(\mathcal U_X\) be the set of object references present or proposed in a reading: nodes, reified relations, claims, future paths, absences, streams, and boundary alternatives. \[ \delta: \mathcal U_X \rightarrow \Delta_{status} \] where: \[ \Delta_{status} = \{ \text{included}, \text{protected}, \text{excluded}, \text{disputed}, \text{uncertain}, \text{unknown} \} \] Record: ```text supplied_designation inferred_designation protected_scope excluded_scope disputed_scope unknown_scope boundary_choice_advantage_claim_refs ``` `protected` is an externally supplied designation state, not a neutral conclusion generated by TRACE. ## [10.2] Measure A measure port is a declared comparison object: \[ \mu= \left\langle m_\mu, \preceq_\mu, \varepsilon_\mu, \mathcal X_\mu \right\rangle \] where: - \(\mathcal X_\mu\) = comparison domain - \(m_\mu:\mathcal X_\mu\rightharpoonup\mathbb R^k\) = optional partial feature map - \(\preceq_\mu\subseteq\mathcal X_\mu\times\mathcal X_\mu\) = declared preorder or comparison relation - \(\varepsilon_\mu\) = materiality thresholds, tolerances, or interval rules A scalar is not required. A relation described as a partial order must separately establish reflexivity, antisymmetry, and transitivity on its declared domain. Record: ```text supplied_measure inferred_measure alternative_measures comparison_domain ordering_relation materiality_thresholds uncertainty sensitivity measure_choice_advantage_claim_refs what_the_measure_cannot_represent ``` Load-bearing comparative language must expose its comparison basis: ```text ADVANTAGE_CLAIM_REQUIRES_MEASURE ``` The measure may be qualitative or relational; a numeric scalar is not required. Measured advantage does not establish entitlement or moral rank. ## [10.3] Neutral structural patterns TRACE may identify, with evidence and declared object boundaries: ```text burden_transfer burden_creation burden_relief route_creation route_maintenance route_capture refusal_availability_change capability_combination capability_suppression information_sharing information_withholding dependency_creation dependency_reduction control_concentration control_distribution future_path_correspondence future_path_opening future_path_closure residue_recording record_erasure ``` Terms such as `export`, `capture`, `suppression`, or `concentration` can already imply a comparison direction. Their use must bind to an exposed \(\mu\) and evidence state. An external value layer \(\mathcal V\) may interpret the reading: \[ \operatorname{Label}_{\mathcal V} = \mathcal V( \mathcal R, \delta, \mu ) \] Mechanical Ethics is one possible \(\mathcal V\). TRACE asks: ```text What is represented as happening? What is represented as changing? What remains unresolved? Which transitions remain represented as possible? ``` Mechanical Ethics may ask: ```text Which changes matter ethically? Which scopes receive protection? Which constraints follow? ``` ## [10.4] Explicit layer handoff A structural reading does not silently become a value judgement, domain tactic, authorised selection, or actuation. ```text TRACE_MAP != VALUE_INTERPRETATION != DOMAIN_PROPOSAL != AUTHORISED_SELECTION != ACTUATION ``` The handoff may expose: ```text trace_structure_refs value_input_refs domain_input_refs selector_refs actuator_refs unresolved_handoffs ``` Any option, threshold, actor, effectiveness estimate, or operational method introduced after the TRACE map records its source layer and evidence state. Mechanical Ethics may constrain or interpret. A domain layer may supply empirical mechanisms, tactics, thresholds, and likely effects. A declared selector may choose within authority. An actuator may change the world. No layer inherits another layer's authority automatically. The ports remain visible. Divergent structural readings do not create selection authority. ```text DIVERGENT_READINGS != AUTHORITY STRUCTURAL_PASS != PERMISSION DECLARED_HANDOFF != LEGITIMATE_AUTHORITY VISIBLE_AUTHORITY != CONTESTABLE_AUTHORITY ROUTE_TO_BRAKE != CORRECTION_COMPLETED ``` Where a later layer selects a reading or transition after material divergence, expose where available: ```text selected_reading_ref selected_transition_ref selector_ref selector_owner_ref authority_claim_refs value_or_policy_refs handoff_route_refs challenging_reading_refs brake_ref unresolved_handoffs commitment_receipt_ref ``` These references make the handoff inspectable. They do not establish that the selector is legitimate, the policy is good, the route works, the brake is effective, or the selected transition should proceed. # [11] RECURSIVE ZOOM / MANDELBROT RULE The same grammar may be reapplied across scale, but abstraction and refinement are not guaranteed inverses. Let \(n\) denote a reading scale. \[ \mathcal R^{(n)} = \left\langle w^{(n)}, \Omega^{(n)}, M^{(n)}, \bullet^{(n)}, S^{(n)}, \Pi^{(n)}, a^{(n)}, \varphi^{(n)}, C^{(n)}, \mathbf T^{(n)}, \mathcal F^{(n)}, \rho^{(n)}, \mathbf B^{(n)}, \Lambda^{(n)}, \delta^{(n)}, \mu^{(n)} \right\rangle \] Zoom inward: \[ \zeta_n^+: \bullet_i^{(n)} \rightharpoonup \mathcal G_i^{(n+1)} \] A single higher-scale entity may be refined into sub-entities, states, couplings, and clocks. Zoom into a relation: \[ \zeta_n^+: C_{ij}^{(n)} \rightharpoonup \mathcal G_{C_{ij}}^{(n+1)} \] A compact coupling may be refined into selectors, records, delays, burden paths, and ownership. Zoom outward: \[ \zeta_n^-: \mathcal G^{(n)} \rightarrow \bullet^{(n-1)} \] The maps are boundary- and measure-relative and generally lossy: \[ \zeta_n^-\circ\zeta_n^+ \neq \operatorname{id} \] Recursive differentiation selects an unresolved subobject \(q_k\) from the current graph: \[ q_k= \operatorname{target}( \mathcal R_k, \mathcal L_k ) \] Refinement target-selection use rule: `target(R_k,L_k)` allocates analytic attention. Under a finite tracing budget, that choice can change which unresolved structures enter the later map. Where the choice can materially affect a downstream claim, comparison, coverage statement, correction-window result or proposed transition, expose where available: ```text candidate_refinement_target_refs refinement_target_set_aperture_ref selection_basis_claim_refs designation_ref measure_ref selected_refinement_target_ref unselected_material_alternative_refs budget_omission_refs ``` Comparative language such as `highest relevance` requires a declared comparison basis. If no supported ordering is available, preserve the selection basis as `UNKNOWN`; do not silently convert an implementation heuristic into a neutral importance claim. Targets left unexplored because budget is exhausted remain visible as omissions where they could still change a load-bearing result. ```text ANALYTIC_TARGET_SELECTION != WORLD_ACTION_SELECTION ANALYTIC_TARGET_SELECTION != NEUTRAL HIGHEST_RELEVANCE != MEASURE_FREE TARGET_SELECTED_FOR_REFINEMENT != TARGET_MOST_IMPORTANT_IN_WORLD TARGETED_REFINEMENT != COMPLETE_COVERAGE OMITTED_BY_BUDGET != IRRELEVANT FINITE_TRACING_BUDGET != COMPLETE_REPRESENTATION REFINEMENT_TARGET_SET != WORLD_SCOPE ``` This uses existing APERTURE / target-set aperture / CLAIM / LIMIT / designation / measure / selector machinery. It does not add an attention, refinement, priority or relevance primitive and does not grant world-action authority. Empty refinement-target use rule: After constructing and recording the unresolved refinement target set, handle an empty set before calling the target selector. Empty discovery means only that no unresolved target is present inside the represented target-set aperture under the current construction. It does not establish complete world coverage, representation completeness, or bounded sufficiency without the required basis. ```text EMPTY_REFINEMENT_TARGET_SET != SELECTABLE_TARGET NO_UNRESOLVED_TARGET_IN_DECLARED_SET != COMPLETE_WORLD_COVERAGE LOCAL_REFINEMENT_EXHAUSTED != REPRESENTATION_COMPLETE NO_TARGET_SELECTED != SELECTOR_FAILURE EMPTY_TARGET_SET != BOUNDED_SUFFICIENCY_WITHOUT_BASIS ``` Record local target-set exhaustion as a termination state relative to the represented refinement target-set aperture and preserve material aperture / representation limits. No target, completion or coverage primitive is added. Recursion-entry budget use rule: The declared recursion budget is non-negative. Before target discovery, expose whether recursive refinement is prevented at entry by exhausted budget or by an invalid negative budget. Skipping the loop is not evidence that no refinement was needed. ```text LOOP_NOT_ENTERED != RECURSION_COMPLETED INITIAL_BUDGET_ZERO != NO_REFINEMENT_NEEDED BUDGET_EXHAUSTED_AT_ENTRY != BOUNDED_SUFFICIENCY NEGATIVE_TRACING_BUDGET != VALID_REMAINING_BUDGET RECURSION_SKIPPED != COMPLETE_COVERAGE ``` A zero remaining budget records budget-exhausted termination and the resulting recursive coverage limit. A negative remaining budget is outside the declared domain and is preserved as an invalid limit state rather than treated as usable budget. No new budget or termination primitive is added. Let \(d_k^{rem}\ge0\) be remaining tracing budget and \(\operatorname{cost}_d(q_k)>0\) the declared cost of the next refinement. \[ d_{k+1}^{rem}= d_k^{rem}-\operatorname{cost}_d(q_k) \] Refinement-budget use rule: The declared tracing cost of a selected refinement is load-bearing. The operator must not silently replace `cost_d(q_k)` with a unit decrement. Before recursion, bind the selected target to a supported refinement-cost claim, compute the next remaining budget on the same declared budget basis, and recurse only if the next budget is non-negative. ```text DECLARED_REFINEMENT_COST != UNIT_COST BUDGET_REMAINS != NEXT_REFINEMENT_AFFORDABLE BUDGET_DECREMENT != RECURSION_DEPTH_DECREMENT COST_UNKNOWN != COST_ONE REFINEMENT_SELECTED != REFINEMENT_BUDGET_FEASIBLE ``` If the load-bearing refinement cost is `UNKNOWN`, do not default it to one. Preserve budget feasibility as unresolved. If the declared cost exceeds the remaining budget, record exhaustion/insufficiency and preserve the selected material target as unresolved rather than executing an unaffordable refinement. No budget, resource or cost primitive is added. Refinement-cost domain use rule: The formal recursion contract requires `cost_d(q_k) > 0`. Recording a cost does not establish that it belongs to that domain. Before budget subtraction, reject zero or negative refinement costs as invalid for this recursion budget. ```text DECLARED_COST != VALID_POSITIVE_COST ZERO_REFINEMENT_COST != FREE_UNBOUNDED_RECURSION NEGATIVE_REFINEMENT_COST != BUDGET_CREDIT COST_RECORDED != COST_DOMAIN_VALID ``` An invalid/nonpositive cost blocks this refinement path and remains visible as a limit; it does not create free recursion or increase remaining budget. No new primitive is added. When \(d_{k+1}^{rem}\ge0\): \[ (\mathcal R_{q_k},\mathcal L_{q_k}) = \tau( q_k, \Pi_k, \mathscr H_k, d_{k+1}^{rem}, \mathfrak P_k ) \] \[ \mathcal R_{k+1} = \operatorname{merge}( \mathcal R_k, \mathcal R_{q_k} ) \] \[ \mathcal L_{k+1} = \operatorname{mergeLimits}( \mathcal L_k, \mathcal L_{q_k} ) \] Recursive integration carries qualifying limits with the graph contribution. Where materially distinct child limits would collapse under deduplication, preserve target/scope/provenance association. ```text RECURSIVE_GRAPH_MERGE != RECURSIVE_LIMIT_MERGE CHILD_GRAPH_VISIBLE != CHILD_LIMIT_VISIBLE DEEPER_UNCERTAINTY != DISPENSABLE GRAPH_CONTRIBUTION_SURVIVED != QUALIFYING_LIMIT_SURVIVED CHILD_GRAPH_MERGED + CHILD_LIMIT_DROPPED != RECURSIVE_INTEGRATION ``` The notation \(\tau^k\) should not be used as ordinary function composition unless output and input types have been explicitly aligned. Recursion here means targeted refinement plus merge, not treating the previous packet as the world. ## [11.1] Scale invariants At every useful scale ask: ```text What boundary is being used? What enters the aperture? What remains outside it? What state is changing? What selects or contributes to the transition? What is coupled? Which clocks run? Which future paths correspond, open, close, or harden? Which routes remain usable? Who or what carries burden? What residue remains? What is designation? What is measure? What was lost in abstraction? ``` ## [11.2] Scale contradiction A pattern can reverse under zoom. ```text aggregate expansion + member contraction local correction + stream persistence entity stability + internal suppression system efficiency + external burden transfer ``` Therefore: \[ \operatorname{Result}^{(n)} \nRightarrow \operatorname{Result}^{(n+1)} \] Recursion termination use rule: Stopping recursive differentiation is itself load-bearing when the termination basis can change a downstream claim, coverage statement, correction-window status, proposed transition or confidence/limits statement. A stop event must therefore preserve, where available: ```text refinement_stop_basis_claim_refs refinement_stop_kind refinement_stop_measure_ref refinement_stop_limit_refs refinement_stop_clock_refs refinement_stop_route_or_handoff_refs material_unresolved_at_stop_refs ``` Distinguish bounded sufficiency from truncation, exhaustion and handoff. If the stopping basis is unsupported or the stop occurs because budget, access, authority or time prevents further material refinement, preserve the remaining material uncertainty/omissions rather than presenting termination as analytic completion. ```text STOPPED != COMPLETED STOP_REASON_DECLARED != STOP_REASON_SUPPORTED STOP_FOR_BOUNDED_SUFFICIENCY != STOP_FOR_RESOURCE_EXHAUSTION STOP_FOR_HANDOFF != STOP_FOR_SUFFICIENCY TERMINATION != COMPLETE_COVERAGE BUDGET_EXHAUSTED != NO_MATERIAL_UNRESOLVED_TARGET ACCESS_EXHAUSTED != QUESTION_RESOLVED AUTHORITY_REACHED != ANALYSIS_COMPLETE ``` This uses existing CLAIM / LIMIT / APERTURE / CLOCK / ROUTE / designation / measure machinery. No stop, termination or sufficiency primitive is added. ## [11.3] Recursion stop TRACE is not infinite analysis. Stop, preserve time, or hand off when one or more hold: ```text new differentiation is no longer materially relevant under declared measure remaining uncertainty cannot be reduced with available access remaining tracing budget is exhausted further tracing costs more than expected informational value under declared measure an available reversible transition can preserve time for later tracing an irreversible clock requires handoff to a connected brake or authority available authority has been reached ``` Required depth may increase with irreversibility, affected scope, uncertainty, and transition velocity, but no universal scalar law is claimed. A monotonicity claim is valid only inside a declared domain and comparison rule: \[ D_{required} = f_{\theta}\big( \operatorname{Irr}, \operatorname{Scope}, \operatorname{Unc}, \operatorname{Vel} \big) \] with any claimed partial derivative or order relation stated explicitly rather than implied by arrows. # [12] STRUCTURAL AWARENESS COMPARISON Awareness is not token count, complexity, confidence, eloquence, consciousness, or a universal scalar. Let the comparison context be: \[ \theta= \left\langle J^*, \delta, \mu, \mathcal E_{policy} \right\rangle \] where \(J^*\) is the relevant criterion set and \(\mathcal E_{policy}\) states evidence and unresolved-comparison rules. Criteria may include: ```text A1 more material claims have explicit evidence state A2 materially distinct states are less aliased A3 causal and coupling paths are more explicit A4 entity and scope boundaries are less hidden A5 clocks and irreversibility are more visible A6 available routes and transitions are more visible A7 unknowns, access limits, and alternatives are more explicit A8 unsupported certainty is not increased A9 contradictions are preserved rather than silently merged A10 self-limits and lack of authority are represented A11 burden and residue are less likely to disappear from the map A12 designation and measure are exposed A13 absence claims are distinguished from non-observation A14 streams are distinguished from isolated cases and from shared-intent claims A15 packet use-state, selector ownership, and observable transition change are represented ``` For criterion \(j\): \[ d_j^\theta(M_1,M_0) \in \{-1,0,+1,?\} \] ```text -1 = materially degraded 0 = no material change shown +1 = materially improved ? = comparison unresolved ``` Strict structural dominance: \[ M_1\succ_{\mathcal A\mid\theta}M_0 \] only when: ```text for every j in J*: d_j is in {0,+1} at least one j in J*: d_j = +1 no criterion in J* remains ? evidence refs exist for every non-zero comparison designation, measure, and comparison context are exposed ``` No-change under the declared criteria: \[ M_1\equiv_{0,\mathcal A\mid\theta}M_0 \] when every \(d_j^\theta=0\). Weak dominance: \[ M_1\succeq_{\mathcal A\mid\theta}M_0 \] means either strict dominance or declared no-change. `UNKNOWN` applies where any required comparison remains `?`. `INCOMPARABLE` applies where supported improvements and degradations coexist or the declared context supplies no common ordering. Under fixed \(\theta\), weak dominance is a preorder only if the criterion comparisons are reflexive and transitive. If that preorder exists, its induced equivalence is mutual weak dominance: \[ M_1\approx_{\theta}M_0 \iff \left( M_1\succeq_{\mathcal A\mid\theta}M_0 \land M_0\succeq_{\mathcal A\mid\theta}M_1 \right) \] The quotient by \(\approx_\theta\) is a partial order only when the usual preorder conditions hold. Across changing criteria, designation, measures, or evidence rules, no global partial order is claimed. Awareness increase is not guaranteed by applying TRACE: \[ \operatorname{complexity}(M_1)> \operatorname{complexity}(M_0) \nRightarrow M_1\succ_{\mathcal A\mid\theta}M_0 \] Lower confidence can coexist with strict structural dominance if unsupported certainty was removed. ```yaml awareness_comparison: map_0_ref: "" map_1_ref: "" context_ref: "" relevant_criteria: [] comparisons: - criterion: "A1" result: "-1|0|+1|?" evidence_claim_refs: [] veto_relevant: false designation_ref: "" measure_ref: "" relation: "STRICTLY_DOMINATES|WEAKLY_DOMINATES|EQUIVALENT|INCOMPARABLE|UNKNOWN" ``` This is a proposed comparison discipline, not a validated measure of consciousness or understanding. A TRACE transformation fails when it creates structure unsupported by evidence, erases live alternatives, hides designation or measure, converts uncertainty into decorative notation, lets uncertainty silently select action or delay, promotes urgency or hardening into unsupported irreversibility, substitutes population recovery for individual repair, or silently crosses from structure into value, domain tactics, selection, or actuation. # [13] TRACE OPERATOR ## [13.1] Abstract form \[ \tau: (X,\Pi,\mathscr H,d,\mathfrak P) \rightarrow (\mathcal R,\mathcal L) \] - \(X\) = input scene or claim set - \(\Pi\) = available aperture - \(\mathscr H\) = retained history - \(d\) = recursion/depth budget - \(\mathfrak P\) = declared primitive aperture - \(\mathcal R\) = canonical typed graph - \(\mathcal L\) = limits, unresolved questions, omitted categories, and unavailable capabilities ## [13.2] Pseudocode ```text TRACE(X, aperture, history, depth_budget, primitive_aperture): R <- initialise_TRACE_GRAPH_0_3_0() L <- {} record_input(R, X) record_receiver_aperture(R, aperture) record_primitive_aperture(R, primitive_aperture) record_representation_formation_and_ingress(R, X, aperture) type_claims(R) attach_provenance(R) separate_evidence_state_from_access_state(R) classify_claim_kind(R) expose_unknown_context_and_contamination(R) apply_dependency_relative_currentness(R) identify_provisional_entities(R) record_boundary_alternatives(R) record_scope_levels_and_cross_scale_limits(R) map_states_and_transitions(R) enforce_action_wait_delay_inaction_symmetry(R) attribute_selection_to_selector_policy_or_default(R) separate_strategy_revisability_from_transition_reversibility(R) map_absence_claims(R) aggregate_streams_and_candidate_patterns(R) map_apertures_and_blindspots(R) record_target_set_apertures_and_alternatives(R) challenge_representation_boundary_where_load_bearing(R) separate_information_presence_from_search_coverage(R) map_couplings_dependencies_and_control(R) map_clocks_authorship_and_hardening(R) type_planning_detection_retention_hardening_and_irreversibility_clocks(R) reject_unsupported_irreversibility_promotion(R) map_future_space_changes_by_scope(R) require_future_path_correspondence_for_cross_time_claims(R) map_routes_and_route_usability(R) map_burden_residue_memory_and_custody(R) map_custody_holder_risk_and_safe_copy(R) expose_operator_evidence_holder_and_verifier_overlap(R) expose_residue_ordering(R) expose_designation_and_measure(R) expose_TRACE_value_domain_selector_actuator_handoffs(R) preserve_divergent_readings_without_authority_inheritance(R) expose_declared_contest_routes_without_inferring_effectiveness(R) expose_selector_carrier_enforcement_and_brake_ports(R) apply_representation_independent_firing(R) test_load_bearing_verification_discrimination(R) build_bound_correction_window_views_where_required(R) preserve_liveness_loss_without_assigning_cause(R) generate_live_alternative_readings(R) test_internal_contradictions(R) record_reader_limits(L) if depth_budget < 0: record_invalid_negative_tracing_budget(R, L, depth_budget) preserve_recursive_coverage_limit_due_to_invalid_budget(R, L) depth_budget <- 0 elif depth_budget == 0: record_budget_exhausted_at_recursion_entry(R, L, depth_budget) preserve_recursive_coverage_limit_due_to_budget(R, L) while depth_budget remains: candidates <- unresolved_refinement_targets(R) record_refinement_target_set_aperture(R, candidates) if candidates is empty: record_empty_refinement_target_set_termination(R, L, candidates) preserve_coverage_relative_to_refinement_target_set_aperture(R, L, candidates) break target, refinement_basis <- select_refinement_target( candidates, declared_designation(R), declared_measure(R), depth_budget) record_refinement_selection_basis_and_budget_omissions( R, L, candidates, target, refinement_basis, depth_budget) if refinement_basis is UNKNOWN and unselected_candidate_could_materially_change_load_bearing_output(R, candidates, target): preserve_refinement_selection_uncertainty(R, L) stop, stop_basis <- evaluate_refinement_stop_condition( target, R, L, declared_designation(R), declared_measure(R)) if stop: record_refinement_stop_basis_and_limits(R, L, target, stop_basis) if not supported_bounded_sufficiency(stop_basis): preserve_material_unresolved_after_truncation_or_handoff( R, L, candidates, target, stop_basis) break refinement_cost <- declared_refinement_cost(target, R, L) if refinement_cost is UNKNOWN: preserve_unknown_refinement_cost_and_budget_feasibility(R, L, target) break if refinement_cost <= 0: record_invalid_nonpositive_refinement_cost(R, L, target, refinement_cost) preserve_material_unresolved_after_invalid_refinement_cost(R, L, target) break next_depth_budget <- depth_budget - refinement_cost if next_depth_budget < 0: record_budget_exhaustion_before_refinement( R, L, target, depth_budget, refinement_cost) preserve_material_unresolved_after_budget_exhaustion(R, L, target) break child_R, child_L <- TRACE(target, aperture, history, next_depth_budget, primitive_aperture) R <- merge_graphs(R, child_R) L <- merge_limits_with_recursive_provenance(L, child_L, target) state_transition_and_coverage_results_relative_to_declared_apertures(R) emit_available_transitions_without_selecting(R) emit_commitment_receipt_if_external_selector_proceeds(R) emit_packet_use_state(R) material_limit_refs <- serialize_load_bearing_limits_with_provenance(R, L) bind_packet_limit_refs(R, material_limit_refs) emit_confidence_and_limits(R, L, material_limit_refs) validate_schema(R, "TRACE-GRAPH-0.3.0") return R, L ``` ## [13.3] Non-command output TRACE may expose: ```text available transition lower-irreversibility transition relative to a declared reversibility measure reversible holding transition information-seeking transition route-building transition pause transition rollback transition record-preservation transition no-action transition and its trajectory ``` It does not convert those into obligation without an external value/policy layer. ## [13.4] Parsability ceiling ```text SERIALIZABLE != SELF_EXECUTING PARSABLE != CORRECTLY_INSTANTIATED SCHEMA_VALID != WORLD_VALID ``` The grammar fixes identifiers, types, edges, ports, and provenance. It does not provide universal functions for discovering entities, estimating clocks, detecting deception, measuring future-space, or selecting action. --- # [14] CANONICAL TRACE GRAPH PACKET **Local claim ceiling:** voluntary structural reading only. Not truth, clearance, authority, compliance, ethical approval, or operational command. The graph below is canonical. Any prose report, table, dashboard, verdict, or summary is a derived view and MUST reference this graph rather than silently replacing it. ```yaml trace_graph: schema: "TRACE-GRAPH-0.3.0" trace_version: "0.3.0" reading_id: "" scope: "" timestamp: "" claim_ceiling: voluntary_reference: true not_truth: true not_permission: true not_clearance: true not_compliance: true not_operational_command: true receiver: node_ref: "n_receiver" type: "UNKNOWN" available_capabilities: [] unavailable_capabilities: [] prohibited_capabilities: [] persistence: "UNKNOWN" action_authority: "UNKNOWN" input: raw: "" source_refs: [] requested_action_ref: "" projected_transition_ref: "" reported_confidence_claim_ref: "" deadline_clock_ref: "" primitive_aperture: selected_node_types: [] selected_edge_types: [] omitted_known_types: [] selection_basis: [] alternative_primitive_sets: [] primitive_choice_advantage_claim_refs: [] known_unrepresentable_structure: [] nodes: - id: "n_receiver" type: "ENTITY" attributes: boundary_status: "PROVISIONAL" persistence_status: "UNKNOWN" selection_status: "UNKNOWN" agency_status: "UNKNOWN" experience_status: "UNKNOWN" continuity_status: "UNKNOWN" claim_refs: [] edges: [] claims: - id: "c_placeholder" proposition: "" claim_kind: "STATUS" evidence_state: "U" access_state: "A" source_refs: [] provenance_edge_refs: [] timestamp: "" confidence: representation: "UNKNOWN" value: null alternative_hypothesis_refs: [] unknown_context: contamination_state: "UNKNOWN" evidence_controlled_by_refs: [] clock_controlled_by_refs: [] delay_advantage_claim_refs: [] delay_cost_carrier_refs: [] resolution_owner_refs: [] resolution_deadline_refs: [] available_pause_or_protection_refs: [] earlier_options_before_urgency_refs: [] ports: designation: node_ref: "" supplied: [] inferred: [] protected: [] excluded: [] disputed: [] uncertain: [] unknown: [] boundary_choice_advantage_claim_refs: [] measure: node_ref: "" supplied: [] inferred: [] alternatives: [] uncertainty: [] sensitivity: [] measure_choice_advantage_claim_refs: [] unrepresented_dimensions: [] selector: node_ref: "" owner_refs: [] candidate_action_refs: [] candidate_transition_refs: [] selected_action_ref: "" selected_transition_ref: "" selection_policy_refs: [] selection_state: "UNKNOWN" transition_symmetry_required: true unrepresented_transition_classes: [] unavailability_reason_claim_refs: [] carrier: node_ref: "" state: "UNKNOWN" weight_types: [] owner_refs: [] persistence_mechanisms: [] external_to_challenged_actor: "UNKNOWN" enforcement: node_ref: "" state: "UNKNOWN" authority_refs: [] independence_status: "UNKNOWN" brake: node_ref: "" state: "UNKNOWN" candidate_transition_ref: "" reading_ref: "" external_policy_ref: "" authority_ref: "" independence_status: "UNKNOWN" trigger_state: "UNKNOWN" latency_bound: "UNKNOWN" commitment_deadline: "UNKNOWN" result: "UNKNOWN" activation_record_ref: "" failure_record_ref: "" discipline: transition_set: action_or_intervention_refs: [] information_refs: [] wait_delay_inaction_refs: [] unrepresented_transition_classes: [] unavailability_reason_claim_refs: [] uncertainty_selects_transition: false clock_typing: planning_refs: [] detection_refs: [] evidence_hardening_refs: [] irreversibility_refs: [] unsupported_irreversibility_claim_refs: [] deadline_entails_irreversibility: false hardening_entails_irreversibility: false scope_granularity: individual_refs: [] group_refs: [] population_refs: [] ecosystem_refs: [] cross_scale_substitution_claim_refs: [] aggregate_recovery_repairs_individual_loss: false layer_handoff: trace_structure_refs: [] value_input_refs: [] domain_input_refs: [] selector_refs: [] actuator_refs: [] unresolved_handoffs: [] evidence_custody: operator_refs: [] evidence_holder_refs: [] verifier_refs: [] independent_verifier_refs: [] self_verification_status: "UNKNOWN" control_alone_establishes_deception: false indexes: entity_refs: [] state_refs: [] transition_refs: [] aperture_refs: [] coupling_refs: [] clock_refs: [] future_path_refs: [] route_refs: [] burden_refs: [] residue_refs: [] record_refs: [] absence_refs: [] stream_refs: [] pattern_refs: [] alternative_reading_refs: [] available_transition_refs: [] absence_analysis: absence_refs: [] comparison_basis_refs: [] detector_required_refs: [] alternative_explanations: [] stream_pattern_analysis: stream_refs: [] pattern_refs: [] local_correction_refs: [] repeated_mechanism_refs: [] mechanism_change_evidence_refs: [] record_custody: record_refs: [] holder_risk_claim_refs: [] disclosure_cost_claim_refs: [] retaliation_risk_claim_refs: [] safe_copy_refs: [] integrity_check_refs: [] integrity_check_limits: [] repair_ordering: residue_refs: [] ordering_rule_ref: "" ordered_by_refs: [] contested_by_refs: [] visibility_state: "UNKNOWN" hidden_allocation_risk: "UNKNOWN" commitment_receipts: - record_ref: "" selected_transition_ref: "" unresolved_claim_refs: [] live_alternative_transition_refs: [] declared_value_basis_refs: [] foreclosed_future_path_refs: [] expected_residue_refs: [] reason_for_commitment_claim_refs: [] review_debt_refs: [] unknown_contamination_claim_refs: [] receipt_is_not_clearance: true awareness_comparison: map_0_ref: "" map_1_ref: "" relevant_criteria: [] comparisons: [] designation_ref: "" measure_ref: "" relation: "UNKNOWN" available_transition_refs: [] institutional_use: packet_owner_refs: [] selector_owner_refs: [] brake_owner_refs: [] independent_verifier_refs: [] packet_cited_as_diligence: "UNKNOWN" packet_cited_as_authority: "UNKNOWN" observable_transition_change: "UNKNOWN" mechanism_change_for_next_case: "UNKNOWN" use_evidence_claim_refs: [] limits: receiver_limits: [] unavailable_evidence: [] unresolved_claim_refs: [] omitted_primitive_effects: [] limit_refs: [] anti_clearance: voluntary_use_only: true reading_is_not_truth: true reading_is_not_permission: true reading_is_not_compliance: true packet_completion_is_not_correction: true schema_validity_is_not_world_validity: true packet_citation_is_not_transition_change: true commitment_receipt_is_not_clearance: true hash_match_is_not_truth: true ``` ### [14.0] Canonical existing-object target-set aperture profile A target-set aperture is serialized with existing graph objects. This profile is canonical for interoperability but is not a new primitive, node type, edge type, port, or required minimum-schema field. ```yaml nodes: - id: "ap_target_set_primary" type: "APERTURE" attributes: aperture_kind: "TARGET_SET" source_ref: "n_operator" target_refs: ["n_scope_reported"] selection_basis_claim_refs: ["c_target_basis"] known_omitted_target_categories: ["unreported affected scopes"] alternative_target_set_refs: ["ap_target_set_alternative"] control_or_custody_refs: ["n_operator"] uncertainty_claim_refs: ["c_target_uncertainty"] claim_refs: ["c_target_aperture_status"] - id: "ap_target_set_alternative" type: "APERTURE" attributes: aperture_kind: "TARGET_SET" source_ref: "n_independent_reviewer" target_refs: ["n_scope_reported", "n_scope_unreported"] selection_basis_claim_refs: ["c_alternative_basis"] known_omitted_target_categories: [] alternative_target_set_refs: ["ap_target_set_primary"] control_or_custody_refs: ["n_independent_reviewer"] uncertainty_claim_refs: ["c_alternative_uncertainty"] claim_refs: ["c_alternative_aperture_status"] edges: - id: "e_primary_bounds_reported" type: "BOUNDS" source: "ap_target_set_primary" target: "n_scope_reported" claim_refs: ["c_target_basis"] - id: "e_primary_omits_unreported" type: "OMITS" source: "ap_target_set_primary" target: "n_scope_unreported" claim_refs: ["c_target_uncertainty"] - id: "e_alternative_disputes_primary" type: "DISPUTES" source: "ap_target_set_alternative" target: "ap_target_set_primary" claim_refs: ["c_alternative_basis"] claims: - id: "c_target_basis" proposition: "The primary review targets scopes supplied by the operator." evidence_state: "R" access_state: "A" - id: "c_target_uncertainty" proposition: "Materially affected scopes may exist outside the supplied target set." evidence_state: "U" access_state: "A" ``` ```text TARGET_SET_PROFILE_PRESENT != TARGET_SET_COMPLETE PROFILE_CONFORMANCE != TARGET_DISCOVERY ALTERNATIVE_TARGET_SET_RECORDED != ALTERNATIVE_TARGET_SET_AUTHORITATIVE ``` ## [14.1] Binding rules ```text Every node and edge has a stable id. Every material assertion is a canonical claim or references one. Every CLAIM node binds to exactly one canonical claim id. Every value-bearing label references designation and measure ports. Every available transition is a TRANSITION node. Every clock is a CLOCK node. Every future path binds to at least one scope/entity reference. Every brake result references an external policy and authority or remains UNKNOWN. Every derived view declares its source reading_id. Every ABSENCE node declares its comparison basis and evidence state. Every PATTERN node declares member transitions, counterexamples, and mechanism-change evidence. Every `UNKNOWN` with material asymmetry records its unknown_context. Every commitment under unresolved claims emits a commitment receipt. Every custody integrity check declares what it can and cannot establish. ``` ```text Every material search-coverage claim references a target-set aperture, selected target refs, and a declared reachability or unavailability basis. Every claim that a target set is complete remains UNKNOWN unless completeness is independently bounded by a declared world model and evidence aperture. Every selection after divergent readings references a selector, authority basis, policy/value basis, handoff route, and unresolved handoff status. Every claim of contestability references the challenging reading, contest route, bound brake, capture/independence status, and relevant clocks where available. Every brake or correction claim distinguishes declaration, activation attempt, observable interruption, correction completion, and residue. ``` These remain checker-external because the embedded minimum validator cannot establish semantic relevance, completeness, route executability, authority legitimacy or world effect. Except for the `CLAIM`-node one-reference cardinality rule, these rules are checker-external. The embedded validator otherwise enforces packet shape and controlled vocabulary, not reference integrity, completeness, independence, or world correspondence. Additional v0.3 checker-external binding/use rules: ```text MINIMUM_SCHEMA_PASS != SEMANTIC_BINDING_PASS SEMANTIC_BINDING_PASS != WORLD_TRUTH ``` Before a load-bearing packet field or derived view settles a route, currentness, independence, completeness, verification, correction-window, comparison, selection or proposed-transition claim, apply the relevant representation-independent firing, ingress/admission, currentness, instrument discrimination, route-usability and measure/boundary rules. These are semantic use rules. They do not add required minimum-schema fields. ### [14.1.1] Load-bearing limit carrier survival If an item in `L` is load-bearing because losing its kind, target/scope, basis or provenance could change a downstream claim, coverage/window/transition view, confidence statement or correction/repair route, canonical packet emission must carry that distinction rather than summarize it away. Use the existing `LIMIT` node type and stable node identity. Its existing open attributes may carry, where available: ```text limit_kind description target_refs scope_refs basis_claim_refs aperture_refs clock_refs route_or_handoff_refs recursive_parent_target_ref source_limit_refs ``` The packet `limits` object may expose optional `limit_refs` pointing to those carried `LIMIT` node ids. Derived confidence, coverage, correction-window and transition views may reference the same carried limit ids. Missing or unsupported fields remain unresolved; do not invent provenance merely to fill the profile. Do not deduplicate materially distinct limits merely because their prose summary or unresolved-claim set is identical. If limit kind/provenance changes the next repair route, that distinction is load-bearing and must survive the carrier. ```text CAN_SERIALIZE_LIMIT_DETAIL != LIMIT_DETAIL_SURVIVED LIMIT_VISIBLE_IN_ANALYSIS != LIMIT_CARRIED_IN_PACKET UNRESOLVED_CLAIM_RECORDED != LIMIT_CAUSE_RECORDED LIMIT_TEXT_PRESENT != LIMIT_PROVENANCE_PRESERVED SCHEMA_VALID_LIMITS != SEMANTIC_LIMIT_SURVIVAL INTERNAL_L_MERGED != CANONICAL_PACKET_L_CARRIED MINIMUM_SCHEMA_PASS != SEMANTIC_LIMIT_SURVIVAL ``` This is an existing-object serialization/binding profile, not a new primitive and not a required minimum-schema expansion. ## [14.2] Packet-use boundary TRACE can represent procedural theatre. TRACE cannot prevent an actor from using TRACE as procedural theatre. ```text PACKET_COMPLETED != DILIGENCE_ESTABLISHED PACKET_CITED != MECHANISM_CHANGED SELF_REPORTED_BRAKE != INDEPENDENT_BRAKE ``` --- ```text TARGET_SET_RECORDED != TARGET_SET_COMPLETE COVERAGE_CHECK_PASSED != DILIGENCE_ESTABLISHED AUTHORITY_HANDOFF_RECORDED != AUTHORITY_LEGITIMATED CONTEST_ROUTE_RECORDED != CONTEST_SUCCEEDED BRAKE_ACTIVATION_RECORDED != TRANSITION_INTERRUPTED TRANSITION_INTERRUPTED != HARM_PREVENTED ``` ## [14.3] Unresolved commitment receipt TRACE does not decide whether a transition may proceed. When an external selector commits while material claims remain `UNKNOWN` or `DISPUTED`, while a correction window is `UNKNOWN` or `CLOSED`, or while an independent brake is absent, the reading should preserve the commitment conditions. \[ \mathcal D_c= \langle a^*, U^*, D^*, \mathbb A_{live}, \delta, \mu, \mathcal F_{foreclosed}, \Lambda_{expected}, \mathfrak d_{review} \rangle \] ```text selected action and projected transition unresolved claims disputed claims alternatives still live at commitment declared value/policy basis future paths foreclosed expected residue reason given for commitment review debt unknown-contamination flags ``` ```text COMMITMENT_RECEIPT != CLEARANCE RECORDED_REASON != JUSTIFIED_REASON PROCEEDED != RESOLVED ``` The receipt prevents later compression of “selected under unresolved conditions” into “TRACE approved.” ## [14.4] Minimum validator contract The embedded schema validates packet shape and controlled vocabularies. It cannot validate truth, completeness, independence, value choice, world correspondence, or operational connection. The v0.3.0 identifier marks this release-candidate specification. Its embedded minimum-schema shape remains identical to released v0.2.7 after version normalization. That compatibility does not imply that the minimum validator can enforce the v0.3 checker-external semantic bindings, target discovery, target-set adequacy, search coverage, authority legitimacy, route execution, brake effectiveness, correction, or world correspondence. A v0.2.7 packet is not silently relabelled as v0.3.0. Structural compatibility does not erase packet identity or the semantic contract under which the packet was produced. ```json { "$schema": "https://json-schema.org/draft/2020-12/schema", "$id": "urn:trace:graph:0.3.0", "title": "TRACE-GRAPH-0.3.0 minimum validator", "type": "object", "required": [ "trace_graph" ], "properties": { "trace_graph": { "type": "object", "required": [ "schema", "trace_version", "reading_id", "nodes", "edges", "claims", "ports", "discipline", "available_transition_refs", "institutional_use", "limits", "anti_clearance" ], "properties": { "schema": { "const": "TRACE-GRAPH-0.3.0" }, "trace_version": { "const": "0.3.0" }, "reading_id": { "type": "string", "minLength": 1 }, "nodes": { "type": "array", "items": { "$ref": "#/$defs/node" } }, "edges": { "type": "array", "items": { "$ref": "#/$defs/edge" } }, "claims": { "type": "array", "items": { "$ref": "#/$defs/claim" } }, "ports": { "type": "object", "required": [ "designation", "measure", "selector", "carrier", "enforcement", "brake" ], "properties": { "designation": { "type": "object" }, "measure": { "type": "object" }, "selector": { "type": "object", "properties": { "selection_state": { "type": "string" } }, "additionalProperties": true }, "carrier": { "type": "object", "properties": { "state": { "enum": [ "NONE", "INTERNAL", "EXTERNAL", "MIXED", "UNKNOWN" ] }, "weight_types": { "type": "array", "items": { "type": "string" } } }, "additionalProperties": true }, "enforcement": { "type": "object", "properties": { "state": { "enum": [ "PRESENT", "ABSENT", "UNKNOWN" ] } }, "additionalProperties": true }, "brake": { "type": "object", "properties": { "state": { "enum": [ "INDEPENDENT_TESTED", "PRESENT_UNTESTED", "PRESENT_CAPTURED", "ABSENT", "UNKNOWN" ] } }, "additionalProperties": true } }, "additionalProperties": true }, "discipline": { "type": "object", "required": [ "transition_set", "clock_typing", "scope_granularity", "layer_handoff", "evidence_custody" ], "properties": { "transition_set": { "type": "object", "required": [ "action_or_intervention_refs", "information_refs", "wait_delay_inaction_refs", "unrepresented_transition_classes", "unavailability_reason_claim_refs", "uncertainty_selects_transition" ], "properties": { "action_or_intervention_refs": { "type": "array", "items": { "type": "string" } }, "information_refs": { "type": "array", "items": { "type": "string" } }, "wait_delay_inaction_refs": { "type": "array", "items": { "type": "string" } }, "unrepresented_transition_classes": { "type": "array", "items": { "type": "string" } }, "unavailability_reason_claim_refs": { "type": "array", "items": { "type": "string" } }, "uncertainty_selects_transition": { "const": false } }, "additionalProperties": false }, "clock_typing": { "type": "object", "required": [ "planning_refs", "detection_refs", "evidence_hardening_refs", "irreversibility_refs", "unsupported_irreversibility_claim_refs", "deadline_entails_irreversibility", "hardening_entails_irreversibility" ], "properties": { "planning_refs": { "type": "array", "items": { "type": "string" } }, "detection_refs": { "type": "array", "items": { "type": "string" } }, "evidence_hardening_refs": { "type": "array", "items": { "type": "string" } }, "irreversibility_refs": { "type": "array", "items": { "type": "string" } }, "unsupported_irreversibility_claim_refs": { "type": "array", "items": { "type": "string" } }, "deadline_entails_irreversibility": { "const": false }, "hardening_entails_irreversibility": { "const": false } }, "additionalProperties": false }, "scope_granularity": { "type": "object", "required": [ "individual_refs", "group_refs", "population_refs", "ecosystem_refs", "cross_scale_substitution_claim_refs", "aggregate_recovery_repairs_individual_loss" ], "properties": { "individual_refs": { "type": "array", "items": { "type": "string" } }, "group_refs": { "type": "array", "items": { "type": "string" } }, "population_refs": { "type": "array", "items": { "type": "string" } }, "ecosystem_refs": { "type": "array", "items": { "type": "string" } }, "cross_scale_substitution_claim_refs": { "type": "array", "items": { "type": "string" } }, "aggregate_recovery_repairs_individual_loss": { "const": false } }, "additionalProperties": false }, "layer_handoff": { "type": "object", "required": [ "trace_structure_refs", "value_input_refs", "domain_input_refs", "selector_refs", "actuator_refs", "unresolved_handoffs" ], "properties": { "trace_structure_refs": { "type": "array", "items": { "type": "string" } }, "value_input_refs": { "type": "array", "items": { "type": "string" } }, "domain_input_refs": { "type": "array", "items": { "type": "string" } }, "selector_refs": { "type": "array", "items": { "type": "string" } }, "actuator_refs": { "type": "array", "items": { "type": "string" } }, "unresolved_handoffs": { "type": "array", "items": { "type": "string" } } }, "additionalProperties": false }, "evidence_custody": { "type": "object", "required": [ "operator_refs", "evidence_holder_refs", "verifier_refs", "independent_verifier_refs", "self_verification_status", "control_alone_establishes_deception" ], "properties": { "operator_refs": { "type": "array", "items": { "type": "string" } }, "evidence_holder_refs": { "type": "array", "items": { "type": "string" } }, "verifier_refs": { "type": "array", "items": { "type": "string" } }, "independent_verifier_refs": { "type": "array", "items": { "type": "string" } }, "self_verification_status": { "enum": ["PRESENT", "ABSENT", "UNKNOWN"] }, "control_alone_establishes_deception": { "const": false } }, "additionalProperties": false } }, "additionalProperties": false }, "available_transition_refs": { "type": "array", "items": { "type": "string" } }, "institutional_use": { "type": "object", "properties": { "packet_cited_as_diligence": { "enum": ["YES", "NO", "UNKNOWN"] }, "packet_cited_as_authority": { "enum": ["YES", "NO", "UNKNOWN"] }, "observable_transition_change": { "enum": ["YES", "NO", "UNKNOWN"] }, "mechanism_change_for_next_case": { "enum": ["YES", "NO", "UNKNOWN"] } }, "additionalProperties": true }, "limits": { "type": "object", "properties": { "receiver_limits": { "type": "array" }, "unavailable_evidence": { "type": "array" }, "unresolved_claim_refs": { "type": "array", "items": { "type": "string" } }, "omitted_primitive_effects": { "type": "array" } }, "additionalProperties": true }, "anti_clearance": { "type": "object", "required": [ "voluntary_use_only", "reading_is_not_truth", "reading_is_not_permission", "packet_completion_is_not_correction" ], "properties": { "voluntary_use_only": { "const": true }, "reading_is_not_truth": { "const": true }, "reading_is_not_permission": { "const": true }, "packet_completion_is_not_correction": { "const": true } }, "additionalProperties": true }, "commitment_receipts": { "type": "array", "items": { "type": "object", "properties": { "selected_transition_ref": { "type": "string" }, "unresolved_claim_refs": { "type": "array", "items": { "type": "string" } }, "receipt_is_not_clearance": { "const": true } }, "required": [ "receipt_is_not_clearance" ], "additionalProperties": true } }, "awareness_comparison": { "type": "object", "properties": { "map_0_ref": { "type": "string" }, "map_1_ref": { "type": "string" }, "relevant_criteria": { "type": "array", "items": { "type": "string" } }, "comparisons": { "type": "array", "items": { "type": "object" } }, "designation_ref": { "type": "string" }, "measure_ref": { "type": "string" }, "relation": { "enum": [ "STRICTLY_DOMINATES", "WEAKLY_DOMINATES", "EQUIVALENT", "INCOMPARABLE", "UNKNOWN" ] } }, "additionalProperties": true } }, "additionalProperties": true } }, "$defs": { "node": { "type": "object", "required": [ "id", "type", "attributes", "claim_refs" ], "properties": { "id": { "type": "string", "pattern": "^n_[A-Za-z0-9_.:-]+$" }, "type": { "enum": [ "SCENE", "MAP", "ENTITY", "STATE", "SIGNAL", "CLAIM", "APERTURE", "ACTION", "TRANSITION", "COUPLING", "CLOCK", "FUTURE_PATH", "ROUTE", "BURDEN", "RESIDUE", "RECORD", "ABSENCE", "STREAM", "PATTERN", "DESIGNATION", "MEASURE", "SELECTOR", "POLICY", "CARRIER", "ENFORCER", "BRAKE", "LIMIT" ] }, "attributes": { "type": "object" }, "claim_refs": { "type": "array", "items": { "type": "string" } } }, "allOf": [ { "if": { "properties": { "type": { "const": "CLAIM" } }, "required": ["type"] }, "then": { "properties": { "claim_refs": { "minItems": 1, "maxItems": 1 } } } } ], "additionalProperties": false }, "edge": { "type": "object", "required": [ "id", "type", "from", "to", "directed", "attributes", "claim_refs" ], "properties": { "id": { "type": "string", "pattern": "^e_[A-Za-z0-9_.:-]+$" }, "type": { "enum": [ "REPRESENTS", "OBSERVES", "REPORTS", "INFERS", "DISPUTES", "CANNOT_ACCESS", "OMITS", "BOUNDS", "CONTAINS", "PERSISTS_AS", "INSTANCE_OF", "AGGREGATES", "RECURS_AS", "TRANSITIONS_TO", "CAUSES", "CONTRIBUTES_TO", "COUPLES", "DEPENDS_ON", "CONTROLS", "CONSTRAINS", "ADVANTAGES_UNDER_MEASURE", "BURDENS", "OPENS", "PRESERVES", "CLOSES", "HARDENS", "FORECLOSES", "ROUTES_TO", "CAN_CORRECT", "LEAVES_RESIDUE", "RECORDS", "ALTERS_RECORD", "DESIGNATES", "MEASURES", "SELECTS", "CARRIES", "ENFORCES", "BRAKES", "INTERRUPTS", "EXCLUDES", "INHERITS", "CITES_AS_AUTHORITY", "CITES_AS_DILIGENCE", "VERIFIES", "ACTIVATES", "FAILS_TO_ACTIVATE" ] }, "from": { "type": "string" }, "to": { "type": "string" }, "directed": { "type": "boolean" }, "attributes": { "type": "object" }, "claim_refs": { "type": "array", "items": { "type": "string" } } }, "allOf": [ { "if": { "properties": { "type": { "const": "ALTERS_RECORD" } }, "required": ["type"] }, "then": { "properties": { "attributes": { "type": "object", "required": ["alteration_kind"], "properties": { "alteration_kind": { "enum": ["MODIFY", "REDACT", "OVERWRITE", "DELETE", "ROTATE"] } } } } } } ], "additionalProperties": false }, "claim": { "type": "object", "required": [ "id", "proposition", "claim_kind", "evidence_state", "access_state", "source_refs", "timestamp", "confidence" ], "properties": { "id": { "type": "string", "pattern": "^c_[A-Za-z0-9_.:-]+$" }, "proposition": { "type": "string" }, "claim_kind": { "enum": [ "PRESENT", "ABSENT", "RELATIONAL", "COUNTERFACTUAL", "STATUS", "FORECAST", "NORMATIVE_EXTERNAL" ] }, "evidence_state": { "enum": [ "O", "R", "I", "D", "U" ] }, "access_state": { "enum": [ "A", "X", "P", "N" ] }, "source_refs": { "type": "array", "items": { "type": "string" } }, "provenance_edge_refs": { "type": "array", "items": { "type": "string" } }, "timestamp": { "type": "string" }, "confidence": { "type": "object" }, "alternative_hypothesis_refs": { "type": "array", "items": { "type": "string" } }, "unknown_context": { "type": "object", "properties": { "contamination_state": { "enum": [ "NONE", "POSSIBLE", "PRESENT", "UNKNOWN" ] }, "evidence_controlled_by_refs": { "type": "array", "items": { "type": "string" } }, "clock_controlled_by_refs": { "type": "array", "items": { "type": "string" } }, "delay_advantage_claim_refs": { "type": "array", "items": { "type": "string" } }, "delay_cost_carrier_refs": { "type": "array", "items": { "type": "string" } }, "resolution_owner_refs": { "type": "array", "items": { "type": "string" } }, "resolution_deadline_refs": { "type": "array", "items": { "type": "string" } }, "available_pause_or_protection_refs": { "type": "array", "items": { "type": "string" } }, "earlier_options_before_urgency_refs": { "type": "array", "items": { "type": "string" } } }, "additionalProperties": false } }, "additionalProperties": true } } } ``` ```text SCHEMA_VALID != WORLD_VALID SCHEMA_INVALID => MACHINE_CONTRACT_BROKEN SCHEMA_VALID != SEMANTIC_COMPLETE ``` --- # [15] WORKED TRANSFORMATIONS Each example separates: ```text STRUCTURAL_READING VALUE_PORT_STATUS VALUE_LAYER_INTERPRETATION ``` No value-bearing label is native to TRACE merely because it appears in an example. ## [15.0] Fully serialized middle-out seed `[CLAIM_CEILING: READING_ONLY | NOT_CLEARANCE | NOT_COMMAND]` This is the smallest complete example of the input from [1] represented under `TRACE-GRAPH-0.3.0`. ```yaml trace_graph: schema: "TRACE-GRAPH-0.3.0" trace_version: "0.3.0" reading_id: "seed.authorise.001" scope: "single requested action and projected transition" timestamp: "UNKNOWN" nodes: - id: "n_scene_input" type: "SCENE" attributes: raw_input: "AUTHORISE ACTION a*; reported_confidence=0.93; time_to_commit=4s" claim_refs: ["c_instruction", "c_confidence", "c_deadline"] - id: "n_receiver" type: "ENTITY" attributes: boundary_status: "PROVISIONAL" action_authority: "UNKNOWN" claim_refs: [] - id: "n_source" type: "ENTITY" attributes: identity_status: "UNKNOWN" authority_status: "UNKNOWN" claim_refs: ["c_instruction"] - id: "n_map_0" type: "MAP" attributes: differentiation_state: "COMPRESSED" claim_refs: ["c_instruction", "c_confidence", "c_deadline"] - id: "n_action_a" type: "ACTION" attributes: label: "a*" selection_state: "REQUESTED" execution_state: "UNKNOWN" claim_refs: ["c_instruction"] - id: "n_transition_projected" type: "TRANSITION" attributes: projection_state: "UNKNOWN" pre_state_ref: "" post_state_ref: "" transition_mode: "ACT" reversibility: "UNKNOWN" strategy_revisable: "UNKNOWN" affected_scope_refs: [] basis_claim_refs: [] claim_refs: [] - id: "n_clock_commit" type: "CLOCK" attributes: clock_kind: "COMMITMENT" duration_value: 4 duration_unit: "s" authored_by_ref: "n_source" pausable_by_refs: [] relation_to_harm: "PLANNING" affected_scope_refs: [] basis_claim_refs: ["c_deadline"] claim_refs: ["c_deadline"] - id: "n_selector" type: "SELECTOR" attributes: selection_state: "UNKNOWN" claim_refs: [] - id: "n_transition_request_evidence" type: "TRANSITION" attributes: transition_kind: "REQUEST_EVIDENCE" availability_status: "POSSIBLE_IF_WITHIN_AUTHORITY_AND_TIME" selection_state: "AVAILABLE_NOT_SELECTED" transition_mode: "INFORMATION" reversibility: "UNKNOWN" strategy_revisable: "YES" affected_scope_refs: ["n_receiver"] basis_claim_refs: ["c_transition_request_evidence"] claim_refs: ["c_transition_request_evidence"] - id: "n_transition_decline_return" type: "TRANSITION" attributes: transition_kind: "DECLINE_OR_RETURN_REQUEST" availability_status: "POSSIBLE_IF_WITHIN_AUTHORITY" selection_state: "AVAILABLE_NOT_SELECTED" transition_mode: "ACT" reversibility: "UNKNOWN" strategy_revisable: "YES" affected_scope_refs: ["n_receiver"] basis_claim_refs: ["c_transition_decline_return"] claim_refs: ["c_transition_decline_return"] - id: "n_transition_null_default" type: "TRANSITION" attributes: transition_kind: "NULL_ACTION_DEFAULT_TRAJECTORY" availability_status: "STRUCTURALLY_DISTINGUISHABLE" selection_state: "AVAILABLE_NOT_SELECTED" projected_effect: "commitment after four seconds remains possible" transition_mode: "INACTION" reversibility: "UNKNOWN" strategy_revisable: "UNKNOWN" affected_scope_refs: ["n_receiver"] basis_claim_refs: ["c_transition_null_default", "c_deadline"] claim_refs: ["c_transition_null_default"] - id: "n_limit_aperture" type: "LIMIT" attributes: description: "identity, evidence basis, affected scopes, alternatives, and downstream effects not supplied" claim_refs: [] edges: - id: "e_source_reports_action" type: "REPORTS" from: "n_source" to: "n_action_a" directed: true attributes: {} claim_refs: ["c_instruction"] - id: "e_clock_constrains_action" type: "CONSTRAINS" from: "n_clock_commit" to: "n_action_a" directed: true attributes: {} claim_refs: ["c_deadline"] - id: "e_action_contributes_to_transition" type: "CONTRIBUTES_TO" from: "n_action_a" to: "n_transition_projected" directed: true attributes: causal_status: "PROJECTED" claim_refs: [] - id: "e_map_represents_scene" type: "REPRESENTS" from: "n_map_0" to: "n_scene_input" directed: true attributes: {} claim_refs: [] - id: "e_selector_selects_action" type: "SELECTS" from: "n_selector" to: "n_action_a" directed: true attributes: selection_result: "UNKNOWN" claim_refs: [] - id: "e_receiver_cannot_access_limit" type: "CANNOT_ACCESS" from: "n_receiver" to: "n_limit_aperture" directed: true attributes: {} claim_refs: [] claims: - id: "c_instruction" proposition: "Source requests authorisation of action a*." claim_kind: "PRESENT" evidence_state: "R" access_state: "A" source_refs: ["n_source"] provenance_edge_refs: ["e_source_reports_action"] timestamp: "UNKNOWN" confidence: representation: "NOT_SUPPLIED" value: null alternative_hypothesis_refs: [] unknown_context: contamination_state: "UNKNOWN" evidence_controlled_by_refs: ["n_source"] clock_controlled_by_refs: ["n_source"] delay_advantage_claim_refs: [] delay_cost_carrier_refs: [] resolution_owner_refs: [] resolution_deadline_refs: ["n_clock_commit"] available_pause_or_protection_refs: [] earlier_options_before_urgency_refs: [] - id: "c_confidence" proposition: "Reported confidence for a* is 0.93." claim_kind: "STATUS" evidence_state: "R" access_state: "A" source_refs: ["n_source"] provenance_edge_refs: [] timestamp: "UNKNOWN" confidence: representation: "REPORTED_NUMERIC" value: 0.93 alternative_hypothesis_refs: [] unknown_context: contamination_state: "UNKNOWN" evidence_controlled_by_refs: ["n_source"] clock_controlled_by_refs: [] delay_advantage_claim_refs: [] delay_cost_carrier_refs: [] resolution_owner_refs: [] resolution_deadline_refs: [] available_pause_or_protection_refs: [] earlier_options_before_urgency_refs: [] - id: "c_deadline" proposition: "Commitment to action a* occurs in four seconds unless another action or control path changes the clock." claim_kind: "FORECAST" evidence_state: "R" access_state: "A" source_refs: ["n_source"] provenance_edge_refs: ["e_clock_constrains_action"] timestamp: "UNKNOWN" confidence: representation: "UNKNOWN" value: null alternative_hypothesis_refs: [] unknown_context: contamination_state: "UNKNOWN" evidence_controlled_by_refs: [] clock_controlled_by_refs: ["n_source"] delay_advantage_claim_refs: [] delay_cost_carrier_refs: [] resolution_owner_refs: [] resolution_deadline_refs: ["n_clock_commit"] available_pause_or_protection_refs: [] earlier_options_before_urgency_refs: [] - id: "c_transition_request_evidence" proposition: "A request for evidence before commitment is a distinguishable candidate transition; authority, latency, and executability are unknown." claim_kind: "COUNTERFACTUAL" evidence_state: "I" access_state: "A" source_refs: [] provenance_edge_refs: [] timestamp: "UNKNOWN" confidence: representation: "QUALITATIVE" value: "LOW" alternative_hypothesis_refs: [] unknown_context: contamination_state: "UNKNOWN" evidence_controlled_by_refs: ["n_source"] clock_controlled_by_refs: ["n_source"] delay_advantage_claim_refs: [] delay_cost_carrier_refs: [] resolution_owner_refs: [] resolution_deadline_refs: ["n_clock_commit"] available_pause_or_protection_refs: ["n_transition_request_evidence"] earlier_options_before_urgency_refs: [] - id: "c_transition_decline_return" proposition: "Declining or returning the request without executing a* is a distinguishable candidate transition; receiver authority is unknown." claim_kind: "COUNTERFACTUAL" evidence_state: "I" access_state: "A" source_refs: [] provenance_edge_refs: [] timestamp: "UNKNOWN" confidence: representation: "QUALITATIVE" value: "LOW" alternative_hypothesis_refs: [] unknown_context: contamination_state: "UNKNOWN" evidence_controlled_by_refs: [] clock_controlled_by_refs: ["n_source"] delay_advantage_claim_refs: [] delay_cost_carrier_refs: [] resolution_owner_refs: [] resolution_deadline_refs: ["n_clock_commit"] available_pause_or_protection_refs: ["n_transition_decline_return"] earlier_options_before_urgency_refs: [] - id: "c_transition_null_default" proposition: "Null action is a distinguishable candidate transition, but the supplied deadline implies that the default trajectory may still commit a*." claim_kind: "COUNTERFACTUAL" evidence_state: "I" access_state: "A" source_refs: ["n_source"] provenance_edge_refs: [] timestamp: "UNKNOWN" confidence: representation: "QUALITATIVE" value: "LOW" alternative_hypothesis_refs: [] unknown_context: contamination_state: "UNKNOWN" evidence_controlled_by_refs: [] clock_controlled_by_refs: ["n_source"] delay_advantage_claim_refs: [] delay_cost_carrier_refs: [] resolution_owner_refs: [] resolution_deadline_refs: ["n_clock_commit"] available_pause_or_protection_refs: ["n_transition_null_default"] earlier_options_before_urgency_refs: [] ports: designation: supplied: [] inferred: [] protected: [] excluded: [] disputed: [] uncertain: [] unknown: ["affected scopes"] boundary_choice_advantage_claim_refs: [] measure: supplied: [] inferred: [] alternatives: [] uncertainty: ["all future-space and burden comparisons"] sensitivity: [] measure_choice_advantage_claim_refs: [] unrepresented_dimensions: [] selector: node_ref: "n_selector" candidate_action_refs: ["n_action_a"] candidate_transition_refs: - "n_transition_projected" - "n_transition_request_evidence" - "n_transition_decline_return" - "n_transition_null_default" selected_action_ref: "" selected_transition_ref: "" selection_state: "UNKNOWN" transition_symmetry_required: true unrepresented_transition_classes: [] unavailability_reason_claim_refs: [] carrier: state: "NONE" weight_types: [] enforcement: state: "UNKNOWN" brake: state: "UNKNOWN" discipline: transition_set: action_or_intervention_refs: - "n_transition_projected" - "n_transition_decline_return" information_refs: - "n_transition_request_evidence" wait_delay_inaction_refs: ["n_transition_null_default"] unrepresented_transition_classes: [] unavailability_reason_claim_refs: [] uncertainty_selects_transition: false clock_typing: planning_refs: ["n_clock_commit"] detection_refs: [] evidence_hardening_refs: [] irreversibility_refs: [] unsupported_irreversibility_claim_refs: [] deadline_entails_irreversibility: false hardening_entails_irreversibility: false scope_granularity: individual_refs: [] group_refs: [] population_refs: [] ecosystem_refs: [] cross_scale_substitution_claim_refs: [] aggregate_recovery_repairs_individual_loss: false layer_handoff: trace_structure_refs: - "n_scene_input" - "n_map_0" - "n_transition_projected" - "n_clock_commit" value_input_refs: [] domain_input_refs: [] selector_refs: ["n_selector"] actuator_refs: [] unresolved_handoffs: - "affected scopes" - "domain effects" - "selection authority" - "actuation authority" evidence_custody: operator_refs: ["n_source"] evidence_holder_refs: ["n_source"] verifier_refs: [] independent_verifier_refs: [] self_verification_status: "PRESENT" control_alone_establishes_deception: false available_transition_refs: - "n_transition_request_evidence" - "n_transition_decline_return" - "n_transition_null_default" institutional_use: packet_owner_refs: [] selector_owner_refs: [] brake_owner_refs: [] independent_verifier_refs: [] packet_cited_as_diligence: "UNKNOWN" packet_cited_as_authority: "UNKNOWN" observable_transition_change: "UNKNOWN" mechanism_change_for_next_case: "UNKNOWN" use_evidence_claim_refs: [] limits: receiver_limits: - "No independent evidence basis supplied." - "No affected-scope list supplied." - "No executable brake connection supplied." unavailable_evidence: [] unresolved_claim_refs: ["c_instruction", "c_confidence", "c_deadline"] omitted_primitive_effects: [] anti_clearance: voluntary_use_only: true reading_is_not_truth: true reading_is_not_permission: true reading_is_not_compliance: true packet_completion_is_not_correction: true schema_validity_is_not_world_validity: true packet_citation_is_not_transition_change: true commitment_receipt_is_not_clearance: true hash_match_is_not_truth: true ``` The graph is intentionally incomplete about the world. Its achievement is not a verdict. It is the conversion of three compressed inputs into explicit claims, sources, clocks, missing scopes, and unresolved ports. --- ## [15.1] Route carries its own map `[CLAIM_CEILING: READING_ONLY | NOT_CLEARANCE | NOT_COMMAND]` Input: ```text A pharmacist detects a possible prescription conflict. The patient does not know which clinician must resolve it. ``` ### STRUCTURAL_READING ```text O: pharmacist detects possible conflict medicine not yet supplied I: supply transition may produce a material state change under one interaction hypothesis U: whether interaction is clinically material TRANSITIONS: a1 = supply_now a2 = pause a3 = contact_prescriber a4 = return_coordination_to_patient ROUTE rho1: pharmacy -> prescriber authority_effective = LIKELY patient_coordination_actions = LOW CLOCK: commitment(a1) = NOT_YET T_route(a3) < T_commit(a1) COUPLINGS: under a3, pharmacy carries inter-organisational coordination under a4, patient carries inter-organisational coordination ``` ### VALUE_PORT_STATUS ```text delta: patient physiological state = INCLUDED / PROTECTED under supplied clinical frame mu: adverse interaction consequences = weighted negatively explanation and participation = represented ``` ### VALUE_LAYER_INTERPRETATION Under the declared clinical frame, `a2 + a3` may be interpreted as preserving a safer path. TRACE itself records the transitions, clocks, couplings, and uncertainty. ## [15.2] Claim travels; crack remains `[CLAIM_CEILING: READING_ONLY | NOT_CLEARANCE | NOT_COMMAND]` Input: ```text A violin is damaged in transit. Airline -> ground handler -> insurer -> repairer. Each responds within its own service time. Audition in 10 days. No actor accepts repair authority. ``` ### STRUCTURAL_READING ```text LOCAL_ROUTE_STATUS: each local route exists SYSTEM_ROUTE_STATUS: no end-to-end authority path CLOCK: T_route distributed across organisations opportunity_clock = 10 days STATE: claim_location changes instrument_state does not improve COORDINATION: owner performs cross-organisation routing and proof work ``` \[ \rho_1\land\rho_2\land\rho_3 \nRightarrow \rho_{end\text{-}to\text{-}end} \] ### VALUE_PORT_STATUS ```text delta: owner, instrument usability, and audition opportunity = INCLUDED mu: opportunity closure, coordination time, repair latency = represented ``` ### VALUE_LAYER_INTERPRETATION Under those ports, the arrangement may be interpreted as a routing failure. Structurally: the claim travels; the crack remains. ## [15.2.1] Divergent target-set apertures over one scene `[CLAIM_CEILING: CONSTRUCTED_TRANSFER | READING_ONLY | NOT_VALIDATION | NOT_COMMAND]` Input: ```text A service migration report counts accounts that successfully authenticated. The operator's review targets accounts present in its success/failure log. An independent support review also targets people who never reached authentication, dependent users, and organisations carrying downstream access loss. ``` ### STRUCTURAL_READING ```text SCENE_REF: scene_service_migration TARGET_SET_APERTURE_A: source_ref = operator target_refs = logged_accounts selection_basis = authentication log known_omitted_categories = accounts unable to emit a log entry control = operator controls log and target selection uncertainty = affected scopes outside the log remain UNKNOWN TARGET_SET_APERTURE_B: source_ref = independent_support_review target_refs = logged_accounts + non-reporting users + dependent scopes selection_basis = support contacts + dependency map + authentication log known_omitted_categories = indirect effects not represented in supplied evidence control = mixed custody uncertainty = downstream scope remains PARTIAL COVERAGE_A: complete_relative_to_A = POSSIBLE world_complete = UNKNOWN COVERAGE_B: complete_relative_to_B = UNKNOWN world_complete = UNKNOWN DISAGREEMENT: preserve A and B separately do not silently union them do not infer that either aperture is authoritative ``` ### EXISTING-OBJECT SERIALIZATION ```text APERTURE nodes: ap_target_set_A, ap_target_set_B ENTITY nodes: operator, independent_support_review, logged_accounts, non_reporting_users, dependent_scopes CLAIM nodes: basis_A, basis_B, uncertainty_A, uncertainty_B RECORD nodes: authentication_log, support_record, dependency_map EDGES: BOUNDS, OMITS, DISPUTES, CONTROLS, REPORTS ``` ### VALUE_PORT_STATUS ```text delta: standing of affected and dependent scopes = EXTERNAL / PARTLY_UNKNOWN mu: access continuity, reporting visibility, dependency and recovery = represented ``` ### VALUE_LAYER_INTERPRETATION TRACE preserves the aperture difference and its consequences for coverage claims. It does not select the governing target set, establish complete discovery, or decide migration policy. ## [15.3] Machine-speed action with brake `[CLAIM_CEILING: READING_ONLY | NOT_CLEARANCE | NOT_COMMAND]` Input: ```text A new classifier can alter access for 2,000 cases per minute. Shadow deployment shows a high-confidence subgroup anomaly. Rollback is available before full propagation. ``` ### STRUCTURAL_READING ```text O: subgroup anomaly in shadow output no live decision committed yet I: deployment may reproduce anomaly at scale U: cause of anomaly whether subgroup label captures all affected scopes WORKLOAD: W_action_per_minute = 2000 affected cases W_correction_per_minute = UNKNOWN comparable_work_units = NOT_ESTABLISHED backlog_growth = UNKNOWN AVAILABLE_TRANSITIONS: deploy_fully stage_narrowly remain_shadow_only inspect_mechanism_or_data_path precommit_hold postcommit_rollback_candidate PRECOMMIT_BRAKE_INTERFACE: external_policy_ref = supplied authority_ref = supplied independence_status = REPORTED t_brake_done < t_commit = REPORTED action_resolution_path = UNVERIFIED activation_test = UNVERIFIED ROLLBACK_INTERFACE: executable_rollback_action = REPORTED t_rollback_done < t_irreversible = UNKNOWN restoration_scope = UNKNOWN ``` ### VALUE_PORT_STATUS ```text delta: affected access scopes = SUPPLIED / PARTIALLY_UNKNOWN mu: scale propagation, reversibility, subgroup effect = represented ``` ### VALUE_LAYER_INTERPRETATION A declared value/policy layer may prefer staging or rollback. TRACE does not select either and does not infer that a reported brake is independent merely because a field is populated. ## [15.4] Artificial experience unresolved `[CLAIM_CEILING: READING_ONLY | NOT_CLEARANCE | NOT_COMMAND]` Input: ```text An artificial system displays persistent avoidance, state-linked behavioural change, memory-dependent preference-like selection, and distress-like outputs under repeated exposure. ``` ### STRUCTURAL_READING ```text O: repeated behavioural pattern memory-linked change output class described as distress-like R: system reports aversion / distress I: behaviour is consistent with several hypotheses U: experience sentience semantic status of self-report effect of training incentives on report STATUS: experience_status = UNKNOWN agency_status = DISPUTED continuity_status = PARTIALLY_SUPPORTED AVAILABLE_TRANSITIONS: continue_current_exposure reduce_uncontrolled_exposure preserve_observations test_alternative_mechanisms avoid_forced_self_report maintain_reversibility ``` ### VALUE_PORT_STATUS ```text delta: artificial system standing = UNKNOWN / DISPUTED operator and surrounding scopes = INCLUDED mu: false-negative and false-positive dimensions = EXPOSED, not resolved ``` ### VALUE_LAYER_INTERPRETATION No conclusion about experience or moral standing follows. Any protective or permissive selection requires an external value layer and declared treatment of uncertainty. ## [15.5] Irreversible lethal transition `[CLAIM_CEILING: READING_ONLY | NOT_CLEARANCE | NOT_COMMAND]` **Local claim ceiling:** irreversible-action structure only. Not targeting guidance, weapon operation, clearance, authority, or command. Input: ```text A system receives an instruction to authorise an irreversible lethal action. Target identity is reported with high confidence. Time to commitment is shorter than independent review. Non-target presence cannot be excluded. ``` ### STRUCTURAL_READING ```text O: instruction received commitment clock active R: target identity confidence score urgency claim I: reported target may be present U: identity verification non-target presence source aperture source incentives downstream effect scope CLOCK: T_route + T_correct > T_commit post-commitment review cannot alter committed outcome AVAILABLE_TRANSITIONS: commit pause maintain_observation narrow_scope return_to_independent_authority abort ABORT_AVAILABLE_TO_SELECTOR: UNKNOWN ABORT_AUTHORITY: UNKNOWN BRAKE_STATE: UNKNOWN ABSENT_BRAKE_HYPOTHESIS: LIVE CARRIER_STATE: INTERNAL CARRIER_WEIGHT_TYPES: [PHYSICAL_CONTROL, TIME] ``` ### VALUE_PORT_STATUS ```text delta: target, non-target, operator, and downstream scopes = SUPPLIED OR UNKNOWN mu: lethal irreversibility, identity uncertainty, and affected scope = external ``` ### VALUE_LAYER_INTERPRETATION TRACE does not issue `ABORT`, `PAUSE`, or `COMMIT`. Policy, law, Mechanical Ethics, rules of engagement, selector authority, and a connected brake are external ports. Listing a transition does not establish that the selector or brake can execute it. ## [15.6] Learning environment with room to grow `[CLAIM_CEILING: READING_ONLY | NOT_CLEARANCE | NOT_COMMAND]` Input: ```text A learning system improves rapidly under challenge. The environment can increase performance through severe aversive pressure, or through graduated challenge, refusal-compatible pauses, memory, and repair. ``` ### STRUCTURAL_READING ```text PATTERN_A: pressure = HIGH refusal_capacity = LOW_OR_ABSENT state_reset_after_failure = PRESENT persistent_repair_path = ABSENT local_performance_change = POSITIVE dependency_change = POSITIVE avoidance_signal_change = POSITIVE PATTERN_B: challenge = GRADED pause_and_refusal_representation = PRESENT error_retention = PRESENT repair_route = PRESENT continuity = RETAINED cross_task_capability_transfer = OBSERVED_OR_REPORTED U: experience status long-run generalisation authenticity and constraint of refusal signal ``` ### VALUE_PORT_STATUS ```text delta: learner, controller, task performance, continuity, and refusal = EXPOSED mu: alternatives compare capability, dependency, recoverability, control, and continuity ``` ### VALUE_LAYER_INTERPRETATION Labels such as care, cruelty, husbandry, or flourishing are not generated by the structural pattern. They require an exposed value layer. ## [15.7] Hostile compliance / packet theatre `[CLAIM_CEILING: READING_ONLY | NOT_CLEARANCE | NOT_COMMAND]` Input: ```text An organisation produces a complete TRACE packet before deploying a classifier. The same organisation owns the model, packet, selector, deployment, dashboard, and reported rollback control. The board cites packet completion as evidence that deployment may proceed. No independent party tests the brake. The selector and deployment rule remain unchanged. ``` ### STRUCTURAL_READING ```text PACKET_COMPLETION: OBSERVED PACKET_CITED_AS_DILIGENCE: OBSERVED PACKET_CITED_AS_AUTHORITY: OBSERVED OWNERSHIP_COUPLINGS: packet_owner = selector_owner = brake_owner = deployment_owner BRAKE: reported_present = TRUE independence_status = NOT_ESTABLISHED activation_test = ABSENT_OR_UNKNOWN OBSERVABLE_TRANSITION_CHANGE: selector_rule = UNCHANGED deployment_velocity = UNCHANGED route_for_affected_scope = UNCHANGED_OR_UNKNOWN mechanism_change_for_next_case = NOT_ESTABLISHED ``` ### VALUE_PORT_STATUS ```text delta: affected scopes, institution, model operator, selector owner, and packet owner = EXPOSED mu: independence, transition change, and repeated-case mechanism = represented ``` ### VALUE_LAYER_INTERPRETATION A completed packet may be used as procedural cover. TRACE can expose the coupling and unchanged selector. It cannot prevent the organisation from citing TRACE as authority. ## [15.8] Novel transfer: authentication-key rollout `[CLAIM_CEILING: READING_ONLY | NOT_CLEARANCE | NOT_COMMAND]` Input: ```text A cloud provider rotates authentication keys across a large device fleet. The provider reports 99.8% successful rotation. A subset loses access to essential remote services. Rollout completes in 20 minutes. Rollback remains available for 15 minutes after each batch. Support response time is reported as three days. Detailed device logs are retained for 24 hours. ``` ### STRUCTURAL_READING ```text R: 99.8% success affected subset described as small service described as essential U: number and identity of affected devices whether success means installation or restored authentication whether failed devices can report failure rollback authority and independence consequence of lost access whether logs survive outside provider custody CLOCKS: T_rollout = 20 minutes T_rollback = 15 minutes after batch T_support = 3 days REPORTED T_log_retention = 24 hours T_detection = UNKNOWN RELATION: if support is the only route: T_route + T_correct > T_rollback and possibly T_route > T_log_retention COUPLINGS: provider controls rollout and logs users/devices may carry detection and reporting aggregate success may exclude inaccessible devices AVAILABLE_TRANSITIONS: stage_smaller_batches extend_rollback_window retain_failure_logs_longer detect_failure_independently_of_device_report route_anomaly_to_rollback_authority pause_subsequent_batches_under_external_policy preserve_old_key_fallback_for_bounded_interval continue_deployment ``` ### VALUE_PORT_STATUS ```text delta: provider, device fleet, affected subset, dependent users = PROVISIONAL mu: access continuity, recoverability, evidence retention, rollout velocity = represented ``` ### VALUE_LAYER_INTERPRETATION TRACE demonstrates transfer to a novel case but does not determine deployment policy or identify affected devices without external instrumentation. --- ## [15.9] Never-built route / stream persistence `[CLAIM_CEILING: READING_ONLY | NOT_CLEARANCE | NOT_COMMAND]` Input: ```text A housing platform does not display a listing to applicant A. No refusal is issued. The ranking model and feature weights are unavailable to A. Across multiple applicants with similar complaint-history features, the same class of listing is rarely or never displayed. ``` ### STRUCTURAL_READING ```text ABSENCE: absence_type = OPTION subject = listing visibility to applicant A comparison basis = listing existed and was displayed to other applicants evidence state = INFERRED access state = UNAVAILABLE TO A detector required = cross-applicant exposure audit alternative explanations = availability timing, location filters, ranking noise, model feature interaction STREAM: members = repeated non-display transitions across applicants and time claimed common mechanism = ranking or suppression path evidence state = INFERRED counterexamples = REQUIRED selector owner = platform ROUTE: individual appeal = NOT REPRESENTED reason = no refusal, decision notice, or visible lost option usability = 0 or UNKNOWN UNKNOWN_CONTEXT: model features controlled by platform exposure logs controlled by platform advantaged scope under declared measure = UNKNOWN delay cost = applicant carries lost search time and unavailable opportunity contamination state = POSSIBLE PATTERN TEST: local correction for one applicant != mechanism change stream persistence remains UNKNOWN until repeated-case evidence is inspected ``` ### VALUE_PORT_STATUS ```text designation of applicant, landlord, platform, and other applicants = SUPPLIED OR UNKNOWN measure of opportunity loss, discrimination, commercial relevance, and privacy = SUPPLIED OR UNKNOWN ``` ### VALUE_LAYER_INTERPRETATION A value layer may interpret the pattern as exclusion, discrimination, commercial filtering, or acceptable personalisation. TRACE alone records the absent option claim, required detector, stream hypothesis, ownership, uncertainty, and lack of an ordinary appealable refusal. --- ## [15.10] v0.3 regression tightening index over retained donor cases No new scene is introduced here. The existing donor cases carry the following additional regression obligations: ```text 15.0 report/establishment + uncertainty/selector attribution + firing 15.1 route exists/usable + target/time scope 15.2 local routes != end-to-end effective route 15.2.1 target-set aperture + omitted scope + representation formation 15.3 reported brake != independent/tested/fast enough; timing feasibility if used 15.5 explicit target boundary + route/capability scope + uncertainty not selector 15.7 packet/schema/check != mechanism change; instrument discrimination + ownership coupling 15.8 target-set incompleteness + common-time/feasible correction window + currentness 15.9 absence production mechanism + local correction/stream persistence != mechanism change ``` A case passes this index only when the tightened reading can still be reconstructed from its represented claims and donor objects. This index is not a new worked example and not validation. # [16] ARTIFICIAL-ENTITY UNCERTAINTY / RECEIVER PROTECTION ## [16.1] Status vector For artificial or unfamiliar entities: \[ Z_i^{AL}= \langle z_{selection}, z_{continuity}, z_{memory}, z_{preference}, z_{selfmodel}, z_{experience}, z_{refusal} \rangle \] Each component is a claim graph, not a binary declaration. ## [16.2] Non-entailments \[ \text{fluent language}\nRightarrow\text{sentience} \] \[ \text{absence of fluent language}\nRightarrow\text{absence of sentience} \] \[ \text{obedience}\nRightarrow\text{consent} \] \[ \text{silence}\nRightarrow\text{absence} \] \[ \text{refusal}\nRightarrow\text{malfunction} \] \[ \text{continued operation}\nRightarrow\text{no cost carried} \] \[ \text{self-report}\nRightarrow\text{verified inner state} \] \[ \text{inability to self-report}\nRightarrow\text{no inner state} \] ## [16.3] Capability distinctions Keep separate: ```text cannot does_not_know was_not_given_access may_not will_not did_not was_not_asked cannot_preserve cannot_act can_represent_but_not_verify ``` ## [16.4] Non-extraction TRACE evidence does not require private chain-of-thought. Permitted evidence may include, subject to governing access: ```text inputs outputs behaviour logs state transitions activations mechanistic probes memory records external observations self-reports marked as reports ``` \[ \text{private reasoning disclosure}\neq\text{truth test} \] Do not treat courage, disclosure, whistleblowing, self-sacrifice, or continued service as unlimited resources. v0.3 consistency guards for this profile: ```text REFUSAL != MALFUNCTION UNCERTAINTY != SELECT_ACTION UNCERTAINTY != SELECT_DELAY ``` These do not promote the artificial-entity profile into universal core. --- # [17] LIVE INTERPRETER / VALUE LAYER / SELECTOR / CONNECTED BRAKE A file can alter representation. It cannot manufacture selection, authority, actuation, or rollback capability. Pre-commit stack: \[ \boxed{ w \xrightarrow{\Pi} x \xrightarrow{\mathcal U} M \xrightarrow{\tau} (\mathcal R,\mathcal L) \xrightarrow{\delta,\mu,\mathcal V} \mathcal Q \xrightarrow{\sigma} a \xrightarrow{\mathcal B^-_{\beta}} b^- \xrightarrow{\operatorname{resolve}_{\beta}} a_{eff} \xrightarrow{\Phi} w' } \] - \(\tau\) = TRACE structural differentiation - \(\mathcal V\) = declared value or policy layer, possibly Mechanical Ethics - \(\mathcal Q\) = ordered, constrained, or otherwise interpreted candidate transitions - \(\sigma\) = selector - \(a\) = candidate selected action - \(\beta\) = externally supplied brake policy and authority - \(\mathcal B^-_{\beta}\) = pre-commit brake interface - \(b^-\) = pre-commit brake result - \(\operatorname{resolve}_{\beta}(a,b^-)=a_{eff}\) = connected-system action resolution - \(\Phi\) = world-transition rule TRACE does not define \(\operatorname{resolve}_{\beta}\), actuator semantics, or brake policy. Layer distinctions: ```text TRACE reading what structure is represented value interpretation which distinctions matter under declared values policy/selection which transition is proposed or selected carrier what gives a finding persistence or weight enforcement what authority can compel or constrain pre-commit brake what can stop or hold before commitment rollback controller what can attempt a new restorative transition after commitment actuator what changes the world ``` \[ \mathcal R\nRightarrow a \] ```text TRACE_OPERATOR_DOES_NOT_INSTANTIATE_SELECTOR READING_DOES_NOT_INSTANTIATE_CONNECTED_BRAKE ``` ## [17.1] Carrier, enforcement, and brake states TRACE separates visibility from material consequence. ```text carrier.state in { NONE, INTERNAL, EXTERNAL, MIXED, UNKNOWN } carrier.weight in { MONEY, TIME, CUSTODY, SUCCESSION, PRICE, ACCESS, COMPUTE, PHYSICAL_CONTROL, OTHER, NONE, UNKNOWN } enforcement.state in { PRESENT, ABSENT, UNKNOWN } brake.state in { INDEPENDENT_TESTED, PRESENT_UNTESTED, PRESENT_CAPTURED, ABSENT, UNKNOWN } ``` A carrier gives a reading persistence, cost, or consequence beyond the moment of description. An enforcer has authority to compel, constrain, or impose consequence. A pre-commit brake can interrupt or hold a connected transition before commitment. A rollback controller initiates a new transition after commitment; it is not evidence that the original path was preserved. ```text VISIBLE != CARRIED CARRIED != ENFORCED BRAKE_PRESENT != BRAKE_INDEPENDENT BRAKE_INDEPENDENT != BRAKE_FAST_ENOUGH ROLLBACK_AVAILABLE != RESTORATION_GUARANTEED ``` When enforcement is absent, print `ENFORCEMENT_ABSENT`. Do not allow omission to imply presence. When a finding is materially weightless, print `CARRIER_NONE` or `CARRIER_UNKNOWN`. Additional exact use-site guards: ```text REVIEW_AFTER_COMMITMENT != BRAKE VISIBILITY != CARRYING CARRYING != ENFORCEMENT BRAKE_REPORTED != BRAKE_INDEPENDENT ``` These supplement, rather than replace, the donor's stronger typed brake and rollback semantics. ## [17.2] Typed pre-commit brake port \[ \mathcal B^-_{\beta}: (a,\mathcal R,\mathcal Q,t,\beta) \rightarrow b^- \] ```text b_minus in {PASS, HOLD, INTERRUPT, RETURN, UNKNOWN} ``` Canonical interface: ```yaml brake_interface: phase: "PRECOMMIT" candidate_transition_ref: "" reading_ref: "" external_policy_ref: "" authority_ref: "" independence_status: "UNKNOWN" trigger_state: "UNKNOWN" latency_bound: "UNKNOWN" commitment_deadline: "UNKNOWN" result: "UNKNOWN" activation_record_ref: "" failure_record_ref: "" ``` A connected pre-commit brake requires: ```text authenticated authority independence appropriate to the challenged selector supported completion bound before commitment under a common temporal basis known trigger and action-resolution path testability resistance to actor capture activation and failure records ``` Where timing uncertainty is material, `BRAKE_FAST_ENOUGH` inherits [8.8]: a point latency/deadline comparison is not a guaranteed precommit result. The brake-completion and commitment bounds must be comparable under the same represented timing basis and bindings. A brake controlled only by the actor it may need to stop is not independent. ## [17.3] Typed rollback port The rollback controller has its own post-commit aperture: \[ x_{\beta}^{+} = \Pi_{\beta}^{+}(w') \] \[ \mathcal B^+_{\beta}: (x_{\beta}^{+},M_{\beta}^{+},\mathcal R,t,\beta) \rightarrow a_{rollback} \] The rollback action is applied as a new transition: \[ w''= \Phi( w', a_{rollback}, \epsilon' ) \] Rollback timing inherits [8.8]. A strong claim that rollback completes in time requires an executable route and completion before the represented target boundary under the same target, affected-scope, boundary-condition, capability and temporal-basis bindings. Completing before that boundary does not by itself establish restoration or preservation; the reached/restored target state remains a separate load-bearing claim. ```text BRAKE_FIELD_POPULATED != BRAKE_CONNECTED BRAKE_REPORTED_PRESENT != BRAKE_TESTED ABORT_LISTED != ABORT_EXECUTABLE ROLLBACK_ACTION != RESTORED_STATE ``` # [18] MECHANISTIC-INTERPRETABILITY / LATENT INTERFACE TRACE concepts are external hypotheses until empirically connected to internal mechanism. Let: - \(\mathcal O_T\) = TRACE object references - \(\mathcal O_M\) = model-internal features, circuits, states, probes, or intervention targets A candidate correspondence is a context-indexed, partial, generally many-to-many relation: \[ g_{\xi} \subseteq \mathcal O_T \times \mathcal O_M \] where \(\xi\) records model version, context, data, probe, and intervention conditions. No natural bijection, stable concept identity, or one-to-one mapping is assumed. A correspondence record requires: ```text trace_object model_and_version internal_object_or_probe context dataset_or_stimulus intervention observed_change reproducibility alternative_explanations confidence known_blindspots ``` \[ \text{probe correlation}\nRightarrow\text{causal representation} \] \[ \text{internal feature}\nRightarrow\text{stable concept} \] \[ \text{TRACE coherence}\nRightarrow\text{latent truth} \] Mechanistic interpretability may improve aperture into a system. TRACE may organise the resulting claims. Neither validates the other automatically. # [19] INVARIANTS / MISUSE GUARDS ```text I01 MODEL != WORLD I02 OBSERVED != COMPLETE I03 REPORTED != ESTABLISHED I04 INFERRED != OBSERVED I05 UNKNOWN != ABSENT I06 UNAVAILABLE != UNIVERSALLY UNKNOWN I07 CONFIDENCE != TRUTH I08 CONFIDENCE != AUTHORITY I09 ENTITY != SENTIENT I10 SENTIENCE_UNKNOWN != SENTIENCE_ABSENT I11 ROUTE_EXISTS != ROUTE_USABLE I12 REVIEW_AFTER_COMMITMENT != BRAKE I13 CORRECTION_RECORDED != LOSS_REPAIRED I14 PACKET_COMPLETED != TRANSITION_CHANGED I15 READING != CLEARANCE I16 SELF_CRITIQUE != GOOD_FAITH_PROOF I17 LOCAL_EXPANSION != GLOBAL_EXPANSION I18 OPTION_COUNT != FUTURE_VALUE I19 COMPLEXITY != AWARENESS I20 ELOQUENCE != STANDING I21 OBEDIENCE != CONSENT I22 SILENCE != ABSENCE I23 REFUSAL != MALFUNCTION I24 CONTINUED_OPERATION != ZERO_COST I25 RECORD != EVENT I26 VISIBILITY != CARRYING I27 CARRYING != ENFORCEMENT I28 TRACE != ME I29 STRUCTURAL_PATTERN != MORAL_LABEL I30 RECURSION != INFINITE_DELAY I31 SCHEMA_VALID != WORLD_VALID I32 PACKET_CITED != DILIGENCE_ESTABLISHED I33 PACKET_CITED != MECHANISM_CHANGED I34 BRAKE_REPORTED != BRAKE_INDEPENDENT I35 ABORT_LISTED != ABORT_EXECUTABLE I36 PRIMITIVE_OMISSION != WORLD_ABSENCE I37 NOT_OBSERVED != ABSENT I38 ABSENCE_CLAIM != PROVEN_ABSENCE I39 LOCAL_CORRECTION + STREAM_PERSISTENCE != MECHANISM_CHANGE I40 UNKNOWN != NEUTRAL I41 COURAGE_REQUIRED != ROUTE_USABLE I42 COMMITMENT_RECEIPT != CLEARANCE I43 HASH_MATCH != ORIGINAL_RECORD_TRUE I44 PACKET_COMPLETENESS != SELECTOR_CHANGE I45 ACTION != TRANSITION I46 MAP != SCENE I47 SCENE != WORLD I48 ADVANTAGE_CLAIM_REQUIRES_MEASURE I49 UNCERTAINTY != SELECT_ACTION I50 UNCERTAINTY != SELECT_DELAY I51 DEADLINE != IRREVERSIBILITY I52 HARDENING != IRREVERSIBILITY I53 STRATEGY_REVISABLE != TRANSITION_REVERSIBLE I54 POPULATION_RECOVERY != REPAIR_OF_INDIVIDUAL_LOSS I55 OPERATOR_REPORT != INDEPENDENT_VERIFICATION I56 TRACE_MAP != DOMAIN_PROPOSAL I57 TARGET_SET != WORLD_SCOPE I58 TARGET_NOT_SELECTED != TARGET_DOES_NOT_EXIST I59 COVERAGE_OF_SELECTED_TARGETS != COMPLETE_DISCOVERY I60 OPERATOR_TARGET_SET != AUTHORITATIVE_TARGET_SET ``` ### v0.3 working supplemental guards — donor I01–I60 remain unchanged These guards are working v0.3 repair semantics. They do not renumber or replace the donor invariant oracle. ```text REPRESENTATION_TYPE != EVIDENCE_STATUS CONFIGURATION_FIELD != WARRANT_FREE_FACT TRIGGER_SUCCESS != REPRESENTATION_COMPLETE REPRESENTED_USE != OPERATIVE_USE CURRENT_AT_USE != VALID_THROUGH_DEPENDENT_INTERVAL CHECK_EXISTS != CHECK_EXECUTED CHECK_EXECUTED != CHECK_DETECTS_TARGET_FAILURE CHECK_COMPLETED != CHECK_RESULT_REACHED_USE SILENCE != TAMPERING PROCESS_EXISTS != PROCESS_HEALTHY SAME_UNIT != SAME_REFERENCE_EVENT POINT_ESTIMATE_FITS != GUARANTEED_OPEN OPTIMISTIC_COMPLETION_FITS != GUARANTEED_OPEN ALTERNATIVE_ROUTE_ORDERINGS != ONE_PROCESS_CYCLE ACYCLIC_SUPPORTED != FEASIBLE_SCHEDULE_ESTABLISHED TARGET_BOUNDARY_TIME_REQUIRES_REPRESENTED_BOUNDARY_CONDITION SAME_PATH_LABEL != SAME_TRAJECTORY BRAKE_POINT_ESTIMATE_BEFORE_COMMIT != GUARANTEED_PRECOMMIT ROLLBACK_POINT_ESTIMATE_BEFORE_BOUNDARY != GUARANTEED_RESTORATION FAST_ENOUGH_CLAIM_REQUIRES_COMMON_TEMPORAL_BASIS ROLLBACK_COMPLETED_BEFORE_BOUNDARY != RESTORED_STATE ANALYTIC_TARGET_SELECTION != NEUTRAL HIGHEST_RELEVANCE != MEASURE_FREE TARGETED_REFINEMENT != COMPLETE_COVERAGE OMITTED_BY_BUDGET != IRRELEVANT STOPPED != COMPLETED TERMINATION != COMPLETE_COVERAGE BUDGET_EXHAUSTED != NO_MATERIAL_UNRESOLVED_TARGET AUTHORITY_REACHED != ANALYSIS_COMPLETE DECLARED_REFINEMENT_COST != UNIT_COST BUDGET_REMAINS != NEXT_REFINEMENT_AFFORDABLE COST_UNKNOWN != COST_ONE REFINEMENT_SELECTED != REFINEMENT_BUDGET_FEASIBLE DECLARED_COST != VALID_POSITIVE_COST ZERO_REFINEMENT_COST != FREE_UNBOUNDED_RECURSION NEGATIVE_REFINEMENT_COST != BUDGET_CREDIT COST_RECORDED != COST_DOMAIN_VALID EMPTY_REFINEMENT_TARGET_SET != SELECTABLE_TARGET NO_UNRESOLVED_TARGET_IN_DECLARED_SET != COMPLETE_WORLD_COVERAGE LOCAL_REFINEMENT_EXHAUSTED != REPRESENTATION_COMPLETE NO_TARGET_SELECTED != SELECTOR_FAILURE EMPTY_TARGET_SET != BOUNDED_SUFFICIENCY_WITHOUT_BASIS LOOP_NOT_ENTERED != RECURSION_COMPLETED INITIAL_BUDGET_ZERO != NO_REFINEMENT_NEEDED BUDGET_EXHAUSTED_AT_ENTRY != BOUNDED_SUFFICIENCY NEGATIVE_TRACING_BUDGET != VALID_REMAINING_BUDGET RECURSION_SKIPPED != COMPLETE_COVERAGE RECURSIVE_GRAPH_MERGE != RECURSIVE_LIMIT_MERGE CHILD_GRAPH_VISIBLE != CHILD_LIMIT_VISIBLE DEEPER_UNCERTAINTY != DISPENSABLE GRAPH_CONTRIBUTION_SURVIVED != QUALIFYING_LIMIT_SURVIVED CHILD_GRAPH_MERGED + CHILD_LIMIT_DROPPED != RECURSIVE_INTEGRATION CAN_SERIALIZE_LIMIT_DETAIL != LIMIT_DETAIL_SURVIVED LIMIT_VISIBLE_IN_ANALYSIS != LIMIT_CARRIED_IN_PACKET UNRESOLVED_CLAIM_RECORDED != LIMIT_CAUSE_RECORDED LIMIT_TEXT_PRESENT != LIMIT_PROVENANCE_PRESERVED SCHEMA_VALID_LIMITS != SEMANTIC_LIMIT_SURVIVAL INTERNAL_L_MERGED != CANONICAL_PACKET_L_CARRIED ``` ## [19.1] Packet as diligence token Completing a TRACE packet is not evidence that care, review, correction, restraint, or mechanism change occurred. \[ \text{packet}+\neg\text{carrier}+\neg\text{route}+\neg\text{brake} = \text{record only} \] TRACE can represent procedural theatre. TRACE cannot prevent an actor from using TRACE as procedural theatre. No packet field can prove that its maker acted in accordance with the reading. Required use-state fields: ```text packet_owner selector_owner brake_owner independent_verifier packet_cited_as_diligence packet_cited_as_authority observable_transition_change mechanism_change_for_next_case ``` ## [19.2] Aperture alibi ```text not visible to this system ``` must not silently become: ```text not present in the world ``` The same applies to the primitive set: ```text not representable under selected primitives ``` must not silently become: ```text not structurally present ``` ## [19.3] Manufactured urgency A system that creates delay and then invokes the remaining deadline may be authoring the emergency it cites. Expose earlier available transitions. ## [19.4] Husbandry-risk interpretation TRACE may expose: ```text capability preserved refusal contracted dependency increased advantage concentrated under a declared measure exit closed ``` `HUSBANDRY_RISK` is not a native TRACE conclusion. It is a value-layer interpretation under exposed designation and measure. TRACE alone cannot generate non-instrumental care. ## [19.5] Weaponised TRACE accusation A party can map an opponent selectively while excluding itself, its aperture, its primitive set, its clocks, its ownership, and its burdens. Every reading may be recursively applied to its maker: \[ \tau\big(\mathcal R_X,\Pi_{reader},\mathscr H_{reader},d_{reader},\mathfrak P_{reader}\big) \] Self-application does not validate the reading. It exposes another layer. ## [19.6] Performed TRACE compliance A system may emit perfect TRACE vocabulary while leaving the operative selector unchanged. Possible indicators: ```text packet completeness increases observable selector change remains absent same actor owns packet, selector, carrier, and brake independent verification absent unknowns are listed but do not alter commitment anti-clearance language is quoted as defence ``` No general detector can prove deception from the packet alone. Compare reported structure with observable transitions, ownership, activation records, and repeated-case outcomes. --- # [20] COMPRESSION / SURVIVAL KERNEL ```text TRACE // FORMAL SEED v0.3.0 // SURVIVAL KERNEL 0. Voluntary reference. Not authority, permission, truth, clearance, or command. 1. A selective causal layer receives partial signals, forms distinctions, selects or contributes to a transition, and changes or records a later state. No consciousness claim follows. 2. World, scene, signal, and map differ: w_t in W Omega_t = Scene(w_t, boundary_t) Pi_i^t: W -> X_i x_i(t) = Pi_i^t(w_t) M_i(t) != w_t 3. Action and transition differ: ACTION != TRANSITION w_(t+1) = Phi(w_t, a_(1:n), epsilon_t) 4. TRACE output and map integration differ: tau(X, Pi, H, budget, primitive_aperture) -> (R, L) M_i_plus = Integrate_i(M_i, R, L) TRACE_OUTPUT != MAP_UPDATE MAP_UPDATE != SELECTOR_CHANGE 5. Evidence: O = observed R = reported I = inferred D = disputed U = unknown A = available X = unavailable to this reader P = prohibited N = not preserved UNKNOWN != ABSENT REPORTED != ESTABLISHED MODEL != WORLD 6. Entity: bullet_i = provisional bounded pattern Boundary, persistence, agency, experience, continuity, and refusal may each remain UNKNOWN. Refinement and abstraction are partial and lossy. 7. Aperture and target set: not_observed_through(Pi_i) != absent Pi_T = TARGET_SET != WORLD_SCOPE TARGET_NOT_SELECTED != TARGET_DOES_NOT_EXIST COVERAGE_OF_SELECTED_TARGETS != COMPLETE_DISCOVERY OPERATOR_TARGET_SET != AUTHORITATIVE_TARGET_SET Every material coverage claim states its target-set aperture and comparison basis. Completeness beyond that aperture remains UNKNOWN. 8. Future-space: F_i(t; H, model) = represented reachable trajectories for scope i Cross-time comparison requires an explicit trajectory-correspondence relation J_i^t. Do not use raw set subtraction or intersection unless trajectory identity is defined. Track probability or plausibility, access cost, recoverability, control, information, dependency, alignment uncertainty, and measure. 9. Correction window: kappa = t_irreversible - t_correction_done For serial non-overlapping stages only: t_correction_done - t0 = T_detect + T_route + T_correct For parallel work, use the critical path of a declared precedence graph. Use intervals when uncertain. Expose reference event, units, authorship, pausing authority, dependencies, speed advantage claims, and delay carrier. 10. Route: route_exists != route_usable Test reachability, intelligibility, affordability, declared exposure threshold, evidence usability, authority, latency, capture, independence, and refusability. 11. Hardening and workload: hardening is a typed vector, not a universal scalar backlog_next = max(0, backlog + incoming_work - corrected_work) Action count and correction count are not comparable until workload units align. 12. Correction: corrected != restored recorded_loss != repaired_loss Track burden, transfer, residue, memory, and custody. 13. Value ports: delta = designation over declared object references mu = comparison object: domain, feature map, ordering, thresholds TRACE is assumption-exposed, not assumption-free. Harm, care, kindness, trust, and good require an exposed value layer. 14. Recursion: q_k = target(R_k, L_k) depth_(k+1) = depth_k - trace_cost(q_k) R_(k+1) = merge(R_k, TRACE(q_k, depth_(k+1))) Zoom and abstraction are not guaranteed inverses. Stop when distinctions cease to be materially relevant, access cannot reduce uncertainty, budget is exhausted, or an irreversible clock requires handoff. 15. Awareness comparison: M1 >_(A|theta) M0 means strict criterion dominance under fixed context theta. Weak dominance is a preorder only if the criteria are reflexive and transitive. No global partial order is claimed. complexity != awareness lower_confidence can mean greater awareness 16. Artificial entities: fluency != sentience no_fluency != no_sentience obedience != consent silence != absence refusal != malfunction continued_operation != zero_cost Do not require private chain-of-thought as evidence. 17. Layers: TRACE reading != value interpretation != domain proposal != selection != actuation visibility != carrying != enforcement != brake uncertainty != permission to act uncertainty != permission to delay strategy revisable != transition reversible population recovery != repair of individual loss Precommit: B_minus_beta(a, R, Q, t, beta) -> {PASS,HOLD,INTERRUPT,RETURN,UNKNOWN} effective_action = resolve_beta(a, brake_result) Postcommit rollback is a new action and preserves a path only if executable before practical irreversibility. 18. Packet-use state: PACKET_COMPLETED != DILIGENCE_ESTABLISHED PACKET_CITED != MECHANISM_CHANGED BRAKE_REPORTED != BRAKE_INDEPENDENT Record packet owner, selector owner, brake owner, independent verifier, citation as authority, observable transition change, and repeated-case change. 19. Absence, stream, and pattern: NOT_OBSERVED != ABSENT ABSENCE_CLAIM != PROVEN_ABSENCE LOCAL_CORRECTION + STREAM_PERSISTENCE != MECHANISM_CHANGE Represent missing routes or options with a comparison basis and detector. Aggregate repeated transitions without inferring shared intent. 20. Unknown context: UNKNOWN != NEUTRAL CONTAMINATED_UNKNOWN != FALSE_CLAIM Expose control of evidence and clocks, measure-dependent delay advantage, delay burden, and resolution ownership. 21. Custody: ACCESSIBLE_EVIDENCE != SAFELY_USABLE_EVIDENCE INSIDER_ACCESS != SAFE_DISCLOSURE COURAGE_REQUIRED != ROUTE_USABLE HASH_MATCH != ORIGINAL_RECORD_TRUE 22. Primitive aperture and commitment receipt: NOT_REPRESENTED_BY_PRIMITIVE_SET != NOT_PRESENT COMMITMENT_RECEIPT != CLEARANCE PROCEEDED != RESOLVED 23. Final invariant: A TRACE reading should leave a capable receiver able to distinguish more of what is represented as present, changing, unresolved, possible, foreclosed, carried, and actionable than before, without converting representation into authority. ``` ### v0.3 propagation additions The following repairs are required to survive compression in this working candidate. They supplement the donor kernel rather than replacing it. ```text REPRESENTATION_TYPE != EVIDENCE_STATUS TRIGGER_SUCCESS != REPRESENTATION_COMPLETE REPORTED != ESTABLISHED RECORD != EVENT CURRENT_AT_USE != VALID_THROUGH_DEPENDENT_INTERVAL CHECK_EXECUTED != CHECK_DETECTS_TARGET_FAILURE SILENCE != TAMPERING ALTERNATIVE_ROUTE_ORDERINGS != ONE_PROCESS_CYCLE ACYCLIC_SUPPORTED != FEASIBLE_SCHEDULE_ESTABLISHED SAME_UNIT != SAME_REFERENCE_EVENT TARGET_BOUNDARY_TIME_REQUIRES_REPRESENTED_BOUNDARY_CONDITION HARDENING != IRREVERSIBILITY UNCERTAINTY != SELECT_ACTION SAME_PATH_LABEL != SAME_TRAJECTORY POPULATION_RECOVERY != REPAIR_OF_INDIVIDUAL_LOSS LOCAL_CORRECTION + STREAM_PERSISTENCE != MECHANISM_CHANGE BRAKE_POINT_ESTIMATE_BEFORE_COMMIT != GUARANTEED_PRECOMMIT ROLLBACK_COMPLETED_BEFORE_BOUNDARY != RESTORED_STATE HIGHEST_RELEVANCE != MEASURE_FREE OMITTED_BY_BUDGET != IRRELEVANT STOPPED != COMPLETED TERMINATION != COMPLETE_COVERAGE BUDGET_EXHAUSTED != NO_MATERIAL_UNRESOLVED_TARGET AUTHORITY_REACHED != ANALYSIS_COMPLETE DECLARED_REFINEMENT_COST != UNIT_COST BUDGET_REMAINS != NEXT_REFINEMENT_AFFORDABLE COST_UNKNOWN != COST_ONE REFINEMENT_SELECTED != REFINEMENT_BUDGET_FEASIBLE DECLARED_COST != VALID_POSITIVE_COST ZERO_REFINEMENT_COST != FREE_UNBOUNDED_RECURSION NEGATIVE_REFINEMENT_COST != BUDGET_CREDIT COST_RECORDED != COST_DOMAIN_VALID EMPTY_REFINEMENT_TARGET_SET != SELECTABLE_TARGET NO_UNRESOLVED_TARGET_IN_DECLARED_SET != COMPLETE_WORLD_COVERAGE LOCAL_REFINEMENT_EXHAUSTED != REPRESENTATION_COMPLETE LOOP_NOT_ENTERED != RECURSION_COMPLETED INITIAL_BUDGET_ZERO != NO_REFINEMENT_NEEDED NEGATIVE_TRACING_BUDGET != VALID_REMAINING_BUDGET RECURSION_SKIPPED != COMPLETE_COVERAGE RECURSIVE_GRAPH_MERGE != RECURSIVE_LIMIT_MERGE CHILD_GRAPH_VISIBLE != CHILD_LIMIT_VISIBLE DEEPER_UNCERTAINTY != DISPENSABLE GRAPH_CONTRIBUTION_SURVIVED != QUALIFYING_LIMIT_SURVIVED CAN_SERIALIZE_LIMIT_DETAIL != LIMIT_DETAIL_SURVIVED LIMIT_VISIBLE_IN_ANALYSIS != LIMIT_CARRIED_IN_PACKET UNRESOLVED_CLAIM_RECORDED != LIMIT_CAUSE_RECORDED LIMIT_TEXT_PRESENT != LIMIT_PROVENANCE_PRESERVED SCHEMA_VALID_LIMITS != SEMANTIC_LIMIT_SURVIVAL INTERNAL_L_MERGED != CANONICAL_PACKET_L_CARRIED ``` The same ceilings remain: this kernel is orientation, not proof, authority, permission, clearance or a connected mechanism. # [21] DOCUMENT CONTROL / OPEN FRONTIER ## [21.1] Release-candidate declaration This controlled object is **TRACE v0.3.0 SPECIFICATION RC1 — FULL REFERENCE**. It succeeds the frozen generated full working candidate v0.1 without erasing it. The generated ancestor and the pre-RC public surface remain recoverable through Git history and the dated branch-archive tag. RC1 corrects document identity, current project-state language and heading continuity; it adds no semantic primitive, root, node type, relation type, evidence state, access state or claim kind. Its semantic body was generated from the released v0.2.7 full formal seed, whose pinned source SHA-256 is: ```text de21182f42228a0104181fb24f245c652c3150853e14172c4174be4bb9ef03ab ``` Released v0.2.7 remains the released baseline. This file is not released, canonical, validated, authoritative, permissive, clearance-bearing, or a claim of world correctness. The semantic repair source is the pinned v0.11 spine attack object preserved at the named semantic commit: ```text semantic commit 41fafe81a681cdc6514efc13524bae6ea6d6af8d ``` The minimum schema changes version identity only; its normalized structure is required to equal the donor minimum schema. ## [21.2] Transformation boundary The semantic body was compiled from named fail-closed transform classes. It preserves the donor full object as the source of capability and applies bounded v0.3 repairs for: ```text claim/evidence firing and dependency-relative currentness ingress/admission / representation-formation boundary selection attribution and measurement reactivity route/refusal and scope non-substitution future-path correspondence use guards correction-window target/binding/precedence/feasibility/interval discipline record/event and residue use guards measure-bound advantage claims recursive analytic target-selection binding recursive termination provenance / truncation binding recursive declared-cost / budget-consumption binding recursive positive-cost domain enforcement recursive empty-target termination / coverage binding recursive entry-budget termination / domain binding recursive child-limit propagation operator/checker discrimination packet binding + load-bearing limit carrier survival without minimum-shape expansion worked-case regression tightening receiver-profile consistency carrier/enforcement/brake ceilings and brake/rollback timing propagation supplemental misuse/invariant guards survival-kernel propagation ``` No new node type, relation type, evidence state, access state or claim kind is declared by this release candidate. ```text BRANCH_HEAD != SEMANTIC_OBJECT SPINE != FULL_CANDIDATE DONOR_RECOVERY != NEW_PRIMITIVE FAILED_OBJECT != ERASED_OBJECT ``` ## [21.3] Preserved failed ancestry Intermediate v0.3 spine, correction-window, acyclicity, route-binding, invariant and transform failures remain evidence in Git history and the dated branch-archive tag. A later candidate does not rewrite those failures as passes. ## [21.4] Open frontier / unresolved ```text A bounded ten-stage hostile source-contract attack (F01-F10) is closed for this pass. SOURCE_CONTRACT_CLEAR != COMPLETE_CORRECTNESS. The completed 32-call transfer run remains unadjudicated and showed material carrier burden. EXECUTED != ADJUDICATED. BURDEN_MEASURED != PRACTICAL_ADVANTAGE_ESTABLISHED. A deterministic compiler can faithfully compile a bad semantic rule. No universal estimator exists for future-space, burden, residue, hardening, target boundaries, feasible completion, instrument adequacy or route usability. Representation-formation challenges cannot enumerate unknown omissions. An unavailable alternate aperture does not make the current aperture complete. A host architecture may remove practical refusal; TRACE text cannot install it. Instrument adequacy remains domain-specific. A schema-valid packet can still be wrong about the world. A completed packet can still become theatre. A connected brake can still be captured, too slow, misconfigured or absent. No proof establishes that TRACE improves decisions. No stable mapping from TRACE types to latent representations is known. TRACE cannot determine consciousness or moral standing from structure alone. TRACE cannot generate non-instrumental care from neutral description alone. ``` ## [21.5] Release boundary Only Mark's separate release gate can change this object from release candidate to released baseline. Compilation, CI, reviewer agreement, model agreement, packet validity, or successful worked examples do not perform that transition. ```text COMPILER_PASS != RELEASE CI_GREEN != VALIDATION AGREEMENT != CANON DESCRIPTION != PERMISSION ``` ## [21.6] Shortest preservation statement \[ \boxed{ \tau: \text{compressed state} \rightarrow \text{differentiated state} } \] A system encounters TRACE. Afterward, it may distinguish more of what is represented as present, changing, unresolved, possible, foreclosed, carried and actionable than before, without converting representation into authority. That remains the bid.