--- framework_id: CF artifact_id: CF-UF-0.4-RC1 name: Consequential Formation — Unified Framework version: 0.4-integration-candidate date: 2026-09-13 status: CANONICAL_CANDIDATE review_state: POST_V0_3_INTEGRATION_CANDIDATE__NOT_FROZEN mode: MACHINE_FIRST_STATE_TRANSFER scope: conceptual-framework + methodology + machine-architecture + domain-bridges audience: - AI research collaborators - coding/build agents - independent reviewers - external scientific/engineering reviewers - BonAcqui project maintainers supersession: candidate_successor_to: "CF-UF-0.3" earlier_orientation_root: "Consequential Formation Program — Canonical Research & Engineering Handoff v0.1" does_not_supersede: - Natural Math v5 frozen foundation - project-specific frozen artifacts or test receipts - native domain mathematics - original source records primary_evidence_status: - EARNED - CANDIDATE - WORKING_HYPOTHESIS - PROHIBITED_OVERCLAIM secondary_origin_class: - ESTABLISHED_IMPORT - OPERATIONAL_RECONSTRUCTION - DEFINITION - DERIVED_FROM_DECLARED_ASSUMPTIONS - CANDIDATE_NOVELTY - OPEN_DEBT - FAILED_OR_REJECTED canonical_motto: "Form is accumulated consequence." governing_discipline: "Show, do not assert." compression_rule: "Factorize, do not forget." anti_drift_rule: "No semantic promotion without evidence-class promotion." --- # CONSEQUENTIAL FORMATION — UNIFIED FRAMEWORK v0.4 RC1 ## Machine-first canonical candidate / research-state transfer **Updated:** 2026-09-13 — post-v0.3 architecture / robotics integration **Status:** CANONICAL_CANDIDATE / RC1 — candidate successor to v0.3 for orientation and external review; **not frozen**. Frozen project artifacts, native-domain authorities, and test receipts retain precedence. > **Purpose:** provide one compact but sufficiently complete object from which another capable system can reconstruct the current Consequential Formation research program, its formal kernel, methodology, machine architecture, domain implications, evidence boundaries, unresolved debts, and immediate falsification program without access to the conversations that produced it. This document is intentionally **machine-readable first**. Repetition is minimized. Definitions are assigned stable IDs. Metaphors are retained only when they compress an operational relation. Native mathematics remains native. No section is permission to promote a CANDIDATE to EARNED. This document is a **canonical candidate**, not a declaration of discovery. RC1 integrates the post-v0.3 synchronization, Formative Field, whole-system field/power, and distributed-embodiment rounds. It does not rewrite frozen lower-level artifacts. Where those rounds remain unsettled, this file records the competing candidate and its kill condition rather than silently promoting it. --- ## v0.4 RC1 integration delta — what changed after v0.3 The post-v0.3 rounds materially changed the **machine architecture**, not the formal CF kernel. 1. **Graph demotion:** a graph remains valuable for Formation Ledger / CIR / CRO audit, provenance and transfer, but is no longer privileged as the cognitive runtime. 2. **Executable formation:** the leading runtime candidate is an **indexed field of formed susceptibilities**. Present conditions recruit only the small coalition whose formed admissibility matches; quiescent formation should not be globally rescored. 3. **COMPILE split:** selection may move to formation/write time because executable susceptibilities are already compiled. A smaller **COMPOSE** office may remain to turn a recruited coalition into coherent host input; it is independently ablatable. 4. **Logic before probability:** hard constraint/admissibility verdicts precede weighting; probability is reserved for genuine competition/uncertainty after elimination. 5. **Host separation sharpened:** distinguish host capability, host policy/habits, and the fledgling system's acquired formation. Bare-host policy baselines are now required. 6. **Primary MFM test sharpened:** reversal learning / negative transfer, recovery formation, twin histories, single-susceptibility transplant and ablation are the preferred causal sequence. 7. **Whole-system extension:** the same field grammar is being tested across power, physical state, integrity, local authority and fault/recovery, with most components dormant until consequentially recruited. 8. **Distributed embodiment:** embodiment is treated operationally as a time-varying **reachability index** over physical capabilities, energy and authority—not assumed identical to one chassis or one metaphysical self. 9. **Energy ecology:** reduced compute/coordination demand changes the crossover economics of harvesting, regeneration, storage and stationary waste-to-energy resources. No universal generator ranking is claimed. These are **CANDIDATE engineering integrations**. Team agreement is not empirical validation. --- # M0. MACHINE MANIFEST ```yaml cf_manifest: id: CF-UF-0.4-RC1 root_claim: id: CF.K0 status: CANDIDATE text: >- A prior occurrence is consequential when its retained difference alters later constraint and therefore changes reachable futures under a declared intervention family. root_chain: - SOURCE_OR_PRIOR_STATE - CONSEQUENTIAL_TRANSITION - RETAINED_DIFFERENCE - ALTERED_CONSTRAINT_OR_WEIGHT - DIFFERENTIAL_REACHABILITY - OPTIONAL_FUNCTIONAL_REUSE - OPTIONAL_SELF_RECONSTRUCTION - OPTIONAL_ANTICIPATION core_objects: CF.Z: FormationState CF.gamma: ConsequentialTransition CF.U: InterventionFamily CF.R: ConsequentialReachability CF.DR: ConsequentialDistance CF.FL: FormationLedger CF.HCC: HistoricalConsequentialCompilation CF.CIR: ConsequentialIntermediateRepresentation CF.CRO: ConsequentialReachabilityObject CF.FC: ConsequentialFidelityVector CF.AQ: Address CF.AR: AddressRegression CF.WIT: CorrespondenceWitness CF.V: ViabilityRegion CF.RV: ViableReachability CF.AUTO: AutonomyVector CF.MEM: FormationCodecMemory CF.FF: FormativeField__CANDIDATE CF.SUS: FormedSusceptibility__CANDIDATE CF.WAKE: RecruitedCoalition__CANDIDATE CF.BODY: EmbodimentReachabilityIndex__CANDIDATE method_stack: - SUSPEND_EXPLANATORY_PRIVILEGE - NOUN_BLIND_RECONSTRUCTION - INDEPENDENCE_AUDIT - TYPE_THE_RETURN - RECURSIVE_SELF_APPLICATION - ADDRESS_REGRESSION - INTERVENTION_TESTING - CORRESPONDENCE_WITNESSING - PRIOR_ART_SUBTRACTION - MATCHED_BASELINE_EVALUATION - PROVENANCE_PRESERVATION - FALSIFICATION_OR_DEMOTION architecture_stack: - PERCEPT - MODULE_PRE_BASINS - CONFUSION_BUFFER - COGNITIVE_BASIN - RIGOR - CIRCUIT - ATAL - GUARD - FORMATION_CODEC - FORMATION_LEDGER - BACKGROUND_FORMATION_LAYER__CANDIDATE - RECONSIDERATION_QUEUE__CANDIDATE - R2R_R2RR_RESEARCH_LOOP - MINIMUM_FORMATIVE_MACHINE__CANDIDATE - FORMATION_LEDGER__AUDIT_PROJECTION - INDEXED_FORMATIVE_FIELD__RUNTIME_CANDIDATE - WRITE_TIME_SUSCEPTIBILITY_COMPILATION__CANDIDATE - DISTRIBUTED_EMBODIMENT__CANDIDATE - ENERGY_ADMISSIBILITY_FIELD__CANDIDATE hard_prohibitions: - NO_NEW_PHYSICS_CLAIM_WITHOUT_DISCRIMINATING_EVIDENCE - NO_HISTORY_CAUSATION_WITHOUT_PRESENT_INSTANTIATION - NO_CROSS_DOMAIN_EQUIVALENCE_WITHOUT_WITNESS - NO_CONSCIOUSNESS_CLAIM_FROM_INTELLIGENCE_CONTINUUM - NO_UNBOUNDED_SELF_PRESERVATION_DIRECTIVE - NO_CRO_AUTO_COMMIT - NO_ERASURE_OF_CONTRADICTION_BY_SUMMARY - NO_PRECISE_ENGINEERING_NUMBER_AS_FACT_WITHOUT_SOURCE_OR_TEST - NO_FIELD_SUPERIORITY_CLAIM_WITHOUT_MATCHED_BASELINE - NO_DISTRIBUTED_EMBODIMENT_IDENTITY_CLAIM_WITHOUT_DECLARED_CRITERION - NO_GLOBAL_PROBABILITY_PASS_WHERE_HARD_ADMISSIBILITY_ALREADY_DECIDES ``` ## M0.1 Dependency graph ```text CF.K0 ├─> CF.gamma consequential transition │ └─> counterfactual reachability comparison under CF.U ├─> retention │ └─> present instantiated difference ├─> constraint/weight change │ └─> CF.R differential future accessibility ├─> memory [broad physical office] │ └─> functional memory [retrieval/reuse/context] ├─> learning │ └─> later policy/reachability changed by retained formation ├─> autonomy/life candidate │ └─> internal reconstruction of enabling constraints ├─> intelligence candidate │ └─> anticipatory selection over represented reachable futures └─> machine architecture ├─> FL/HCC/CIR/CRO ├─> Cognitive Basin / FormationCore ├─> Formation Codec / metabolism of history ├─> R2R/R2RR evidence-seeking loops └─> PERCEPT physical-consequence receptors ``` --- # 0. AGENT CONTRACT / HOW TO USE THIS FILE ## CF.RULE.0 — Evidence classes are load-bearing Every proposition MUST retain one primary evidence status: - **EARNED** — supported by completed implementation, accepted formal work, or a specified result whose wording does not exceed its evidence. - **CANDIDATE** — coherent and operational enough to formalize/test; promotion still requires evidence. - **WORKING_HYPOTHESIS** — useful search coordinate, presently underdetermined. - **PROHIBITED_OVERCLAIM** — wording not allowed without new evidence. Use a second orthogonal field when useful: - **ESTABLISHED_IMPORT** — known external science/engineering used without novelty claim. - **OPERATIONAL_RECONSTRUCTION** — familiar phenomenon rewritten in CF variables. - **DEFINITION** — a stipulated term or measurement office. - **DERIVED_FROM_DECLARED_ASSUMPTIONS** — follows if premises/definitions hold. - **CANDIDATE_NOVELTY** — possible project-original residual after preliminary subtraction. - **OPEN_DEBT** — unresolved primitive, metric, mapping, identifiability or implementation dependency. - **FAILED_OR_REJECTED** — route tested and not retained. **Do not conflate evidence status with novelty.** Something may be EARNED as an implementation result and entirely non-novel scientifically. Something may be CANDIDATE and potentially novel. ## CF.RULE.1 — Source precedence When this document conflicts with a frozen project artifact, the frozen artifact wins for that project. When this document conflicts with an original source record, preserve both and open a contradiction. Do not silently rewrite history. Priority for reconstruction: ```text frozen executable/test receipt > explicit canonical declaration > current unified framework > project handoff > team synthesis > conversation summary > recollection/inference ``` This priority concerns **authority**, not truth. A later experiment can overturn a frozen claim; it must do so explicitly with a supersession record. ## CF.RULE.2 — Native-domain mathematics remains native CF may reorganize questions, dependencies and measurements. It does not replace domain equations with decorative CF equations. A correspondence between domains must be witnessed against each domain's native semantics. ## CF.RULE.3 — The framework is allowed to return NO CHANGE Success includes: - NO CHANGE; - USEFUL REFACTORING; - BOUNDARY QUESTION EXPOSED; - NEW TESTABLE CONJECTURE; - METHOD FAILED; - INCOMPARABLE; - UNSUPPORTED. A method that always rediscovers CF is broken. ## CF.RULE.4 — Historical origin has no magical causal force A historical assumption affects a current system only insofar as its consequence remains instantiated in current structure, equations, tools, data, policy, habits, weights, constraints or interpretation. This is the **Markov safeguard** generalized: if two complete present microstates are truly identical under Markov dynamics, unembodied historical difference cannot make them evolve differently. Any claimed history effect must live somewhere in the present state or in an omitted variable/environmental relation. ## CF.RULE.5 — Category suspension is constitutive, not optional CF does not begin by accepting the nouns contained in a question and then attaching CF vocabulary to them. The first operation is: \[ \boxed{ \textbf{SUSPEND EXPLANATORY PRIVILEGE} } \] No inherited category may enter a CF analysis as a primitive merely because the input question contains it. Words may remain for communication, but they may not perform unexplained causal or boundary work during reconstruction. Canonical rule: \[ \boxed{ \textbf{A NOUN THAT WAS NEVER REMOVED CANNOT BE CLAIMED TO HAVE SURVIVED THE CF FILTER.} } \] The noun may return only after observations, relations, constraints and interventions force some useful office back. ## CF.RULE.6 — Returned categories require an independence audit Passing the boundary test is not enough. Every returned noun/concept must also face: 1. **Boundary test** — does observation force some version of the distinction back? 2. **Independence test** — can the returned concept be defined without importing the other relations that reconstructed with it? A returned concept must be typed as one or more of: - **PRIMITIVE** — not presently reconstructed from simpler declared relations; - **OFFICE** — a recognizable functional role instantiated by deeper relations; - **INDEX** — a coordinate/context required to evaluate a relation, not necessarily a constituent; - **OPERATOR** — a transformation over existing objects/states; - **COMPOSITE / EMERGENT** — stable organization of multiple relations; - **COMPRESSION NOUN** — a useful compact label for a recurrent conjunction; - **OPEN DEBT** — currently unresolved. Canonical warning: \[ \boxed{ \textbf{REMOVE THE OLD NOUN. RECONSTRUCT. THEN REMOVE OUR NEW NOUN TOO.} } \] ## CF.RULE.7 — Mutual constitution is allowed; premature primitive promotion is not The Time/Energy/Matter/Gravity reinspection showed that familiar offices can be difficult to define independently. CF therefore permits a result of the form: \[ A\leftrightarrow B\leftrightarrow C\leftrightarrow A \] without claiming identity among \(A,B,C\). A mutually defining cluster is not evidence that the members are "really one thing." It is evidence that one-way noun boxes may be a poor representation of the operational dependency structure. When several returned concepts only make sense together, record the dependency loop and resist promoting each member to an independent primitive. ## CF.RULE.8 — Nature supplies routines, not a required implementation script Biological development is an existence proof for many useful operations: local generation, persistent writes, repair, selection, topology-changing accessibility, multiscale feedback and population learning. It is not the required implementation substrate for engineered intelligence. CF molecular/machine work may borrow an **office** while rejecting the biological mechanism that instantiates it. Chemistry, cells, metabolism, replication and biological morphology are optional engineering choices unless independently required by the target capability. ## CF.RULE.9 — Representation for audit is not mechanism of execution The representation that best preserves provenance need not be the mechanism that best executes formation. Current machine rule: \[ \boxed{\textbf{FORMATION LEDGER / GRAPH FOR AUDIT} \neq \textbf{RUNTIME SELECTION MECHANISM}} \] A graph/hypergraph remains a strong candidate for lineage, contradiction, transfer and debugging. Runtime selection must earn itself independently on compute, causal mediation and retrieval quality. ## CF.RULE.10 — Do not estimate what can already be decided Use the strongest earned operator first: ```text hard admissibility / exclusion -> deterministic implication / bounded rule -> weighting among survivors -> probability only where genuine uncertainty or competition remains -> UNKNOWN when evidence is insufficient ``` A falsified route should normally be **BLOCKED**, not assigned a small probability merely because a probabilistic mechanism is available. ## CF.RULE.11 — Compute should scale with what wakes, not with everything retained For persistent formation \(K_t\) and recruited coalition \(\mathcal W_t\), the desired machine regime is: \[ |\mathcal W_t|\ll|K_t| \] and wake/recruitment cost should be dominated by indexed matches and active coalition size rather than repeated global rescoring. A field that scans every retained object has not earned a computational advantage. ## CF.RULE.12 — Consequence writes locally unless evidence requires wider propagation A consequence should first modify the susceptibility / constraint structure causally implicated in producing it. Global propagation, fleet transfer and cross-scale promotion require declared scope, provenance and validation. --- # 1. EXECUTIVE SYNTHESIS Consequential Formation (CF) is presently best classified as a **candidate methodological, representational and engineering unification framework**, not a validated universal theory and not a new fundamental physics. Its operating sequence is now explicitly two-stage: \[ \boxed{ \text{SUSPEND}\rightarrow\text{OBSERVE}\rightarrow\text{DISTINGUISH}\rightarrow \text{TRACE CHANGE}\rightarrow\text{TRACE RETENTION}\rightarrow \text{MAP CONSTRAINT}\rightarrow\text{MAP REACHABILITY}\rightarrow \textbf{AUDIT INDEPENDENCE}\rightarrow\text{RECONSTRUCT}\rightarrow \textbf{TYPE THE RETURN}\rightarrow\text{RENAME ONLY IF EARNED} } \] Within that operation, the core consequential relation remains: \[ \boxed{ \text{PRIOR OCCURRENCE}\rightarrow\text{SURVIVING DIFFERENCE}\rightarrow \text{ALTERED CONSTRAINT/WEIGHT/CONFIGURATION}\rightarrow\text{CHANGED REACHABLE FUTURES} } \] The methodological correction earned in the September 12–13 rounds is load-bearing: **category suspension is constitutive of CF.** Applying CF terminology to an inherited ontology without first withholding the ontology's explanatory privilege is not a full CF analysis. A second correction is equally load-bearing: categories that return must not immediately become CF's new primitives. They must pass an independence audit and be typed as primitive, office, index, operator, composite, compression noun or open debt. Minimal canonical chain: \[ \boxed{ \text{SOURCE} \rightarrow \text{FORMATION} \rightarrow \text{RETAINED CONSEQUENCE} \rightarrow \text{ALTERED CONSTRAINT} \rightarrow \text{WEIGHTED REACHABILITY} } \] CF becomes useful when inherited nouns hide this chain. The framework therefore treats nouns such as *memory*, *life*, *intelligence*, *meaning*, *system*, *object*, *time*, *energy*, *matter*, *body*, *communication*, or *ground* as **candidate compressions**, not unquestionable primitives. It temporarily removes them, reconstructs from observations and relations, then allows them to return only if they earn explanatory/predictive/intervention value. A post-v0.3 architectural correction now separates **retained formation as record** from **retained formation as executable susceptibility**. The current leading candidate is not “search the whole Basin.” It is: \[ \boxed{ \text{CURRENT CONDITION} \rightarrow\text{INDEXED FORMED SUSCEPTIBILITIES} \rightarrow\text{SPARSE RECRUITMENT} \rightarrow\text{LOCAL RESOLUTION} \rightarrow\text{CONSEQUENCE} \rightarrow\text{LOCAL REFORMATION} } \] The Formation Ledger may remain a graph indefinitely while the runtime is a field/index of executable gates, masks, tolerances and event receptors. This is a candidate engineering architecture, not a new physical field claim. The working program has four tightly related offices: 1. **Scientific reinspection:** audit inherited representations and category boundaries. 2. **Machine state representation:** compile formed history into provenance-bearing structures rather than repeatedly reconstructing from prose. 3. **Budget-conscious cognition:** use retained formation to reduce repeated probabilistic search, context load, memory cost and reasoning energy. 4. **Cross-domain instrumentation:** define measurements of retention, reachability, autonomy, correspondence, recovery and physical state that can survive substrate changes without claiming false identity. The strongest engineering hypothesis is: \[ \boxed{ \text{PAST CONSEQUENCE / COMPUTATION} \rightarrow \text{DURABLE EXECUTABLE FORMATION} \rightarrow \text{FEWER FUTURE ROUTES WAKE} \rightarrow \text{LESS FUTURE COMPUTATION} } \] This is not yet experimentally earned as a general result. It is an immediate test target. --- # 2. SYMBOL REGISTRY — v0.4 RC1 Symbol collision is treated as a correctness defect. Use the following unless a native domain requires its own notation. | Symbol | Canonical office | Notes | |---|---|---| | \(Z\) | formation state | Extensible macro/micro state; representation declared per experiment. | | \(X\) | present configuration component | One component of \(Z\). | | \(\mathcal C\) | constraint structure | Avoid bare \(C\) where closure/cost is also present. | | \(W\) | weighting structure | May be multidimensional; not automatically probability or energy. | | \(P\) | provenance structure | Source dependence / lineage. | | \(Q\) | contradiction/unresolved set | Used inside CIR/FL. | | \(\gamma\) | transition/event under test | \(Z\to Z'\). | | \(\mathcal U\) | intervention/policy family | MUST be declared for reachability/distance claims. | | \(T\) | horizon | Physical time or ordered step horizon as domain requires. | | \(\mathcal R_T(Z;\mathcal U)\) | consequential reachability | Set/distribution of reachable future states. | | \(D_R\) | consequential distance | Divergence between reachable sets/distributions. | | \(\kappa_T(\gamma)\) | candidate consequence magnitude | Counterfactual reachability distance attributable to \(\gamma\). | | \(\mathcal V\) | viability/admissibility region | Domain-relative. | | \(\mathcal R_{\mathcal V}\) | viable reachability | Reachable subset remaining inside/returning to \(\mathcal V\). | | \(\rho_{rec}\) | recoverability | Use instead of \(R\) to avoid collision with reachability. | | \(B_{\mathcal V}\) | viable-basin width | Robustness range; candidate metric family. | | \(\chi_{reg}\) | regulatory closure | Fraction/degree of maintenance dependency internalized. | | \(\chi_{prod}\) | constructive closure | Degree to which maintenance executors are themselves produced/renewed internally. | | \(\Sigma_{self}\) | self-specification | Distance of maintained internal constraints from environment-imposed attractors; definition depends on domain. | | \(H_{hist}\) | history coupling | Whether prior perturbation alters later response/recovery. | | \(\mathbf A\) | autonomy vector | Never assume a single scalar is sufficient. | | \(\mathbf T_R(f,t)\) | reference-fabric transfer operator | Replaces ambiguous \(H(f,t)\) in electrical observability work. | | \(\mathbf F_C\) | Consequential Fidelity vector | Machine-state transfer fidelity. | | \(\Phi_t\) | Formation Codec frame | Avoid \(F_t\), which collides with transition functions/fidelity. | | \(\mathsf A(Q)\) | Address of question \(Q\) | Typed dependency/office structure. | | \(\mathfrak R_A\) | Address Regression operator | Avoid \(\mathcal R\) collision. | | \(\mathsf W_{AB}\) | correspondence witness | Mapping + preserved property set + certificates + error + soundness. | | \(\Xi_t\) | present typed condition / event packet | Runtime stimulus to executable formation; not assumed dense broadcast. | | \(\Theta_{ledger}\) | auditable formed state | Provenance/contradiction/transfer projection; graph/hypergraph is suitable. | | \(\Theta_{exec}\) | executable formed susceptibility state | Gates, masks, tolerance regions, indexed subscriptions, local rules. | | \(\mathcal W_t\) | recruited / waking coalition | Small subset of formation/components active for current condition. | | \(\mathcal B_t\) | embodiment reachability index | Currently available physical capabilities under energy/connectivity/authority. | | \(\mathcal P_E\) | energy/resource admissibility channel | Provisional whole-system field office; exact plane taxonomy remains open. | **Rule:** native notation may override locally, but the bridge back into CF must declare the mapping. --- # 3. FORMAL KERNEL ## CF.K1 — Formation state Use an extensible state: \[ Z=(X,\mathcal C,W,P,Q,REC,\ldots) \] where fields may include configuration, constraints, weights, provenance, unresolved contradictions and recovery routes. No universal microscopic state representation is claimed. ## CF.K2 — Consequential transition Let transition \(\gamma\) occur from a matched prior condition. Define counterfactual future reachability under a declared intervention family \(\mathcal U\): \[ \mathcal R_T^{(\gamma)} \quad\text{and}\quad \mathcal R_T^{(\neg\gamma)}. \] Operational criterion: \[ \boxed{ \gamma\text{ is consequential over }(T,\mathcal U) \iff D_R\!\left(\mathcal R_T^{(\gamma)},\mathcal R_T^{(\neg\gamma)};\mathcal U\right)>\epsilon } \] Candidate magnitude: \[ \kappa_T(\gamma)=D_R\!\left(\mathcal R_T^{(\gamma)},\mathcal R_T^{(\neg\gamma)};\mathcal U\right). \] **Status:** DEFINITION/CANDIDATE measurement family. Choice of representation, horizon, intervention family, distance and tolerance must be declared. ## CF.K3 — Retained consequence A consequence is **retained** if some difference attributable to prior transition \(\gamma\) remains instantiated at later state \(Z_t\) and its removal/ablation changes later reachability under relevant interventions. This blocks purely narrative history. Past events matter only through surviving causal structure. ## CF.K4 — Memory Broad technical office: \[ \boxed{ \text{memory} = \text{durable consequence of prior admissible change that alters later reachability} } \] This is deliberately broader than cognitive memory. Distinguish: ```text physical retention < functional reuse < context-sensitive retrieval < abstract reconstruction < anticipatory use ``` A crack, hysteresis loop, crystal defect, morphology or synaptic modification can satisfy the first office without being intelligent. ## CF.K5 — Learning Candidate operational definition: \[ \boxed{ \text{learning} = \text{formation that changes later response/reachability under relevant recurrence} } \] Learning requires history-dependent change in future behavior, not merely transient state. ## CF.K6 — Consequential reachability \[ \mathcal R_T(Z;\mathcal U) \] is the set or distribution of future states/behaviors reachable from \(Z\) over horizon \(T\) under declared interventions/policies \(\mathcal U\). Reachability is intervention-relative and representation-relative. ## CF.K7 — Consequential distance \[ D_R(Z_a,Z_b;\mathcal U) \] compares the reachable consequences of two states. No universal metric is claimed. Valid metrics may include set distance, distributional divergence, task success divergence, recovery curves, control costs, or other domain-native quantities. ## CF.K8 — Macrostate equivalence / emergence A macrostate \(M\) is operationally adequate relative to \(\mathcal U\) when predictions/interventions derived from \(M\) match those derived from fuller \(Z\) within declared tolerance \(\epsilon\), while \(M\) is materially cheaper/smaller. Candidate engineering definition: \[ M\sim_{\mathcal U,\epsilon} Z \quad\text{if predictive/interventional outputs match within }\epsilon. \] **Emergence**, in this framework, is usefully treated as **predictive compression that survives intervention**, not as proof of a new ontological substance. --- # 4. OPERATIONAL GRAMMAR VS DEEP-PRIMITIVE AUDIT A major reconciliation carried forward from v0.2: the program has used two different “primitive” layers. They must not be collapsed. ## CF.P0 — Operational noun-blind grammar [CANONICAL WORKING] Use when reanalyzing a domain: ```text DISTINCTION OCCURRENCE / REALIZATION TRANSITION / CHANGE CONSTRAINT / RULE WEIGHT REACHABILITY RETAINED CONSEQUENCE ``` Relations/composition are implicit and may be made explicit when needed. This grammar is a **methodological starting vocabulary**, not an ontological proof. ## CF.P1 — Aggressive deep reduction [WORKING HYPOTHESIS / OPEN DEBT] A later audit proposed the reduced candidate set: \[ \mathcal P_*= \{ \text{DISTINCTION}, \text{COMPOSITION}, \text{WEIGHT}, \text{CONSTRAINT/RULE}, \text{REALIZATION} \}. \] Under that reduction: - reachability is the transitive closure of admissible transitions; - transition can be represented by the constraint/rule relation; - retained consequence may be reconstructed as history-dependent rule/configuration surviving into the present; - identity may be reconstructed as persistence/invariance under declared transformations. Open debts include: - actual vs merely possible (**REALIZATION debt**); - weighting measure / typicality; - locality; - continuity; - dimensionality; - Lorentzian signature / relativistic structure; - probability measure; - why particular constraints/rules exist. **Rule:** use P0 for current operational work. Use P1 only when explicitly testing deeper reduction. P1 does not silently supersede P0. --- # 5. FORMATION LEDGER / PROVENANCE / STATE TAXONOMY ## CF.FL — Formation Ledger Persistent provenance-bearing graph/hypergraph recording formed distinctions, relations, weights, contradiction, evidence, lineage and recovery routes. For recorded history, minimal form: \[ G_H=(V,E,W,P) \] but practical CIR extends this. ## CF.STATE — Do not confuse absence with inaccessibility A prior structure may be: - **ERASED** — causal/representational consequence removed under declared tests; - **DEPRIORITIZED** — remains but carries reduced weighting; - **SUPPRESSED** — blocked from normal expression while retained; - **INACCESSIBLE UNDER CURRENT INTERVENTIONS** — retained but not exposed by present query/test family; - **TRANSFORMED** — old representation replaced by a new causally related one; - **DISTRIBUTED** — no single local carrier, consequence spread across relations/substrate; - **UNKNOWN** — insufficient test to classify. The distinction is critical to memory, model editing, scientific history and “forgetting.” ## CF.PROV — Consequential provenance Bibliographic citation is not enough. A source's influence may be modeled by counterfactual ablation: \[ P_i \sim D_R(M,M^{-S_i}). \] Historical ablation is often modeled rather than directly testable; therefore provenance must label itself: ```text ATTESTED INFERRED SIMULATED_COUNTERFACTUAL DIRECTLY_TESTED ``` Never merge these statuses. **Historical influence is not truth. Truth is not popularity. Rejection is not erasure.** --- # 6. CORRESPONDENCE DISCIPLINE — EARN THE ARROWS Cross-domain resemblance is a major overclaim risk. CF therefore requires a witness grammar. ## CF.WIT.1 — Hierarchy Strongest to weakest useful relation: 1. **IDENTICAL VALUE** — same formal object/quantity, units/semantics aligned. 2. **FORMAL EQUIVALENCE** — all relevant queries intertranslatable with proof. 3. **STRUCTURAL CORRESPONDENCE** — declared nontrivial property set preserved by witnessed map; non-preserved properties listed. 4. **WITNESSED SIMULATION** — one-directional transition preservation under declared scope/error. 5. **SHARED SIGNATURE ONLY** — NOT a correspondence; vocabulary/schema resemblance. Additional verdicts: - **INCOMPARABLE** requires a demonstrated obstruction. - **UNSUPPORTED** means a claim was made but no witness stands. ## CF.WIT.2 — Correspondence witness For domains \(A,B\): \[ \boxed{ \mathsf W_{AB}=(\varphi,\mathcal P,\mathcal C_{cert},\varepsilon,\sigma) } \] where: - \(\varphi\): explicit map/relation between states/objects; - \(\mathcal P\): declared properties to preserve; - \(\mathcal C_{cert}\): commutation/intervention certificates; - \(\varepsilon\): quantitative error budget; - \(\sigma\): soundness stamp that checks certificates against native domain semantics. Canonical rule: ```text DISCOVERED CORRESPONDENCE = witness exists + certificates verify DECLARED CORRESPONDENCE = sentence A ~ B without witness ``` **Use this rule whenever CF claims “the same structure” across physics, biology, cognition, electronics, history or social systems.** ## CF.WIT.3 — History sufficiency certificate History is not a universal hidden field. It is a **state-sufficiency question**. For declared state \(X_t\), scale \(s\) and intervention family \(\mathcal U\): - **SUFFICIENT:** \(X_t\) is sufficient for the native dynamics; history field is empty for the query and \(\mathcal R_T(X_t\mid H_t)=\mathcal R_T(X_t)\). - **NON_MARKOVIAN_AT_DECLARED_SCALE:** reachability still depends on omitted history. Extend the state toward a Markov completion (damage variable, hysteresis branch, epigenetic state, optimizer state, seed bank, etc.) or return `STATE_NOT_SUFFICIENT`. This prevents a common CF error: attributing causal power to abstract history when the real defect is a coarse or incomplete present-state declaration. ## CF.WIT.4 — Typed reverse operator “Can it return?” is underspecified. Reverse claims must declare type: ```text (direction, target_equivalence, resource_class) ``` Direction may mean preimage under forward dynamics, image under time-reversed generator, or channel decoding. Target may be exact microstate, macrostate equivalence, functional recovery, viability region, or protocol reversal. A cross-domain reverse correspondence is licensed only when reverse types map explicitly. Otherwise return `REVERSE_TYPE_MISMATCH`; ordinary structural correspondence may still survive. ## CF.WIT.5 — Assumption algebra String labels are insufficient for contradiction detection. Candidate machine implementation should use typed assumption atoms connected by a curated theorem/native-rule library: ```text REQUIRES / IMPLIES REFINES EXCLUDES / INCOMPATIBLE UNKNOWN ``` Contradiction should be theorem/rule-mediated where possible, not guessed from lexical similarity. `UNKNOWN` is a valid output. ## CF.WIT.6 — Composition is value-type native and witness-gated Do not install one universal cross-domain composition law. Probability, cost, information and other value types compose differently. Applicability requires a certificate (e.g. conditional independence/Markov sufficiency, path structure, native chain rule). Coarse-graining must likewise declare its map. Loss of return-relevant coordinates can create macro-level asymmetry even when finer dynamics are reciprocal; this is a scale/representation result, not permission to declare all irreversibility a CF derivation. --- # 7. THE CF REINSPECTION METHOD — ERASE THE CAVE-WALL LINES ## CF.METHOD.0 — Core question \[ \boxed{ \text{If inherited boundaries had never been drawn, which boundaries would observations force us to redraw?} } \] ## CF.METHOD.1 — Noun-blind pass For target domain/problem \(Q\): 1. Freeze observations/data/equations/native semantics. 2. List inherited nouns/boundaries to be temporarily withheld. 3. Re-express observations using operational grammar P0. 4. Reconstruct relations/constraints/order/weights/reachability without smuggling nouns back. 5. Gate each returning noun: state what observation/operation forces its return. 6. Record which boundaries: - return unchanged; - return with changed office; - split; - merge; - fail to return; - remain undecidable. 7. Run native-domain preservation check. 8. Run prior-art subtraction. 9. Demand a prediction, intervention or measurable efficiency delta; otherwise classify as useful refactoring/semantic reframing. 10. Retain failures and controls. ## CF.METHOD.2 — Calibration requirement The method must include cases where: - it reconstructs the legacy structure essentially unchanged; - it exposes a boundary question; - it fails by design because the object is stipulated/formal rather than formed; - it generates a candidate test. The cross-domain audit already produced all four outcome classes. This is a feature: the method is not allowed to “prove CF” everywhere. ## CF.METHOD.3 — Anti-map rule \[ \boxed{ \text{DO NOT USE THE MAP TO REDRAW THE MAP.} } \] Perfect noun-blindness is impossible for reasoners trained on inherited vocabulary. The operational remedy is not pretending blindness; it is explicit gatekeeping and source/warrant bookkeeping. ## CF.METHOD.4 — Independence audit / type the return After the noun-blind pass, every returned concept must be tested for dependency on the other structures reconstructed with it. For returned concept \(N'\): ```text 1 identify the observation/intervention forcing N' back 2 list the relations required to define N' 3 remove N' again while preserving those relations 4 test whether the explanatory/predictive office survives 5 classify N' as PRIMITIVE / OFFICE / INDEX / OPERATOR / COMPOSITE / COMPRESSION / OPEN DEBT 6 record mutual-constitution loops rather than forcing independent primitives ``` A concept may be indispensable without being primitive. ## CF.METHOD.5 — Recursive self-application CF must be allowed to operate on its own vocabulary. In particular, terms such as *reachability*, *closure*, *access*, *recoverability*, *formation*, *constraint* and *distinction* are not immune from removal/reconstruction. If CF's own terms fail the filter, demote or replace them. --- # 8. ADDRESS / ADDRESS REGRESSION Address Regression is currently **CANDIDATE METHODOLOGY**, not an earned scientific result. ## CF.ADDR.1 — Address For question \(Q\): \[ \boxed{ \mathsf A(Q)= (\mathsf D,\mathsf R,\mathsf C,\mathsf V,\mathsf B,\mathsf M,\mathsf P,\mathsf W,\mathsf E) } \] - \(\mathsf D\): distinctions/entities required to state the problem; - \(\mathsf R\): mathematical relations/operators/equations/inference rules; - \(\mathsf C\): constraints/symmetries/conservation/regularity; - \(\mathsf V\): variables/state/function spaces; - \(\mathsf B\): boundary/initial/asymptotic/admissibility conditions; - \(\mathsf M\): measurement/operational bridge between symbols and phenomena; - \(\mathsf P\): provenance/formation history; - \(\mathsf W\): validity/confidence/approximation/support/scope; - \(\mathsf E\): evaluation target (theorem/prediction/control/numerics/explanation/reduction). The Address is not merely assumptions. It includes the **office structure** making the question determinate. ## CF.ADDR.2 — Regression operator \[ \boxed{ \mathfrak R_A(Q_0) \mapsto (\mathsf A(Q_0),\mathsf L(Q_0),Q_{rec},\Delta_{comp}) } \] with comparison vector: \[ \Delta_{comp}=( \Delta_{prediction}, \Delta_{generalization}, \Delta_{compute}, \Delta_{verification}, \Delta_{interpretability}, \Delta_{failure\ detection}). \] Algorithm: ```text 1 compute dependency closure 2 assign warrant/status to every dependency 3 traverse upstream 4 preserve mathematically closed/native relations 5 mark approximation/validity domains 6 import background theory only where needed for measurement/closure 7 stop under declared stopping rule 8 emit immutable original + reconstructed quotient + office deltas + certificates 9 compare under matched solver/task/resource budget ``` Stopping rule: stop when further regression reaches independently measured/declared primitives, becomes circular, destroys operational meaning, yields no expected predictive/explanatory gain, or costs more than expected benefit. ## CF.ADDR.3 — Representation debt [CANDIDATE METRIC] One candidate: \[ R_{debt}= \frac{C_{inherited}-C_{reconstructed}}{C_{reconstructed}}, \] where \(C\) is computational work per verified result under matched solver/task: FLOPs, tokens, search, memory movement, latency, verification and—where measurable—joules. **Kill:** if Address-first systems produce no reproducible gain in prediction, generalization, assumption detection, verification or resource cost beyond ordinary formal specification/retrieval/prompt engineering, demote Address Regression to semantic reframing / established-practice compilation. --- # 9. TIME / ENERGY / MATTER / SPACE / GRAVITY — CURRENT BOUNDARY STATUS This section is a scope guard, not cosmology. ## CF.PHYS.1 — Time and energy Current working stance: ```text COMMON ROOT / DISTINCT PROJECTIONS — CANDIDATE NOT: time = energy ``` Within this program, transition structure is used as a working root because: - time characterizes ordering/succession/duration of change; - energy/exergy characterizes generator/capacity/availability for transformation; - established physics relates them through action, phase, Hamiltonian evolution and symmetries; - those mathematical relations do not settle ontology. **PROHIBITED OVERCLAIM:** “CF has derived time and energy from a deeper law.” ## CF.PHYS.2 — Time/matter separation audit Results are definition-sensitive. When “matter” means only baryons/massive particles, time-without-matter examples are trivial. When “matter” is broadened to any physically instantiated state/substrate—including dynamic geometry/fields/vacuum contributions—the clean separation becomes difficult or vacuous. Current safe statement: \[ \boxed{ \text{time and physical substrate are formally distinguishable in many models but operationally entangled in measurement.} } \] No ontological identity or co-dependence proof is claimed. ## CF.PHYS.3 — Space/gravity Noun-removal rounds can expose how much structure is carried by relations, geometry, invariants and dynamics rather than by naive substance nouns. They have not derived GR/QFT or a quantum-gravity theory. Native physics remains authoritative. ## CF.PHYS.4 — Mutual-definition calibration from the four-noun round The earlier removal of **time / energy / matter / gravity** remains a calibration artifact for CF itself. The result was not that the four nouns vanished into one substance. Rather, their familiar offices repeatedly reconstructed through interdependent structures: - ordering / succession / duration / clocks for time-like office; - rule invariance / conserved charge under declared symmetries for energy-like office; - persistent constraint-bearing patterns for matter-like office; - persistent structure altering geometry/reachability for a gravity-like office. Established physics itself contains a dependency loop of the approximate form: \[ \boxed{ E\leftrightarrow g_{\mu\nu}\leftrightarrow\tau\leftrightarrow\Gamma\leftrightarrow E } \] where \(\Gamma\) denotes causal/reachability structure. This is **not an identity claim**. It is a warning that apparently independent nouns may occupy mutually defining offices. Open debts remain explicit: Lorentzian structure, quantum interference, dimensionality/continuity and native gravitational dynamics do not follow from minimal CF primitives without imported assumptions or additional derivation. This calibration produced the current rule: when a new CF term appears indispensable, first ask whether it is actually a primitive or merely an index/office inside a mutually constituted relation. ## CF.PHYS.5 — Current status of access, timescale, closure and recoverability The corrected boundary rounds repeatedly recovered these concepts, but the Restore-the-Genesis self-audit demoted premature primitive status: - **access / position** — presently best treated as an INDEX / contextual relation on reachability, not established primitive; - **timescale** — presently an INDEX over partially reconstructed temporal structure; carries the unresolved time debt; - **closure** — presently a COMPOSITE topology of mutually maintaining constraints/transformations, not established primitive; - **recoverability** — presently an OPERATOR / predicate over indexed reachability under a declared rescue/intervention family; - **intervention family \(\mathcal U\)** — presently a CONTEXT / Address declaration required to make reachability claims determinate. These classifications are current working results, not immutable ontology. --- # 10. CONSEQUENTIAL CLOSURE — TANGIBLE / INTANGIBLE ## CF.CC.1 — Candidate closure principle Across examined causally efficacious distinctions, no operational boundary has yet been found where causal effect exists without an instantiated difference in configuration, relation, constraint, weighting or environmental coupling. Classification: ```text CONSEQUENTIAL CLOSURE = PARTIALLY SUPPORTED / SCOPE-LIMITED ``` This is close to physical causal closure plus macrostate supervenience restated at a machine-operable level. It is not independent proof of physicalism and not new physics. ## CF.CC.2 — Observability boundary The program repeatedly finds a stronger boundary at: ```text causally consequential but unmeasured / poorly represented vs no demonstrated causal consequence ``` “Intangible” often functions as a label for high-level, distributed, relational or poorly observed formation rather than a demonstrated nonphysical kind. **PROHIBITED OVERCLAIM:** “CF disproves the mind/body distinction” or “all mental phenomena have been reduced.” --- # 11. LIFE / AUTONOMY / DEATH / DORMANCY Classification: **OPERATIONAL RECONSTRUCTION + CANDIDATE MEASUREMENT PROGRAM**, strongly neighboring autopoiesis, autonomy theory, control/viability theory and origin-of-life literature. ## CF.LIFE.1 — Noun-blind core Some coupled organizations, when activity is interrupted, merely relax toward environment. Others use internally coupled processes to restore characteristic constraints, and those restoring processes are themselves among what the organization maintains. The candidate focus is not “life substance” but **self-maintained constraint closure with recoverable viable reachability**. ## CF.LIFE.2 — Organism boundary \[ \boxed{ \text{THE ORGANISM IS THE CLOSURE, NOT ANY COMPONENT.} } \] The boundary follows the set of mutually maintaining processes/constraints, not visual shell, substrate, intelligence or conventional chassis. This is a candidate operational boundary, not a universal biological definition already established. ## CF.LIFE.3 — Autonomy vector Avoid single-score autonomy where possible. Candidate vector: \[ \mathbf A= (s_{maint},\rho_{rec},\chi_{reg},\chi_{prod},\Sigma_{self},H_{hist},\ldots) \] where: - \(s_{maint}\): self-maintenance slope attributable to internal activity vs matched ablation; - \(\rho_{rec}\): recoverability under declared rescue family; - \(\chi_{reg}\): regulatory closure; - \(\chi_{prod}\): constructive closure / production of maintenance executors; - \(\Sigma_{self}\): self-specification relative to environmental attractor; - \(H_{hist}\): history-dependent change in later recovery/response. ## CF.LIFE.4 — Dormancy \[ activity\approx0, \qquad \mathcal R_{\mathcal V}\neq\varnothing \] Dormancy is low activity with retained recoverability, not death. ## CF.LIFE.5 — Death Family-relative candidate definition: \[ \boxed{ \text{Dead}(Z;\mathcal U_{res}) \iff \rho_{rec}(Z;u)=0\quad\forall u\in\mathcal U_{res} } \] verified at a stated resource/technology budget. This deliberately means improved rescue technology can revise a death verdict without changing the definition. Origin and death should be treated as trajectories in the same viability/autonomy state space, **not assumed exact mathematical inverses**. Formation may require multiple conditions to close; collapse can occur when one load-bearing condition fails. ## CF.LIFE.6 — Reproduction / heredity / evolution Keep distinct: ```text individual autonomy = organization maintains enabling constraints reproduction = generates discrete successor organization heredity = transmits constraint configuration with variation retention lineage evolution = differential persistence of heritable variants ``` Reproduction is not required for individual autonomy under the candidate framework. ## CF.LIFE.7 — Autonomy Handoff [CANDIDATE EXTENSION] A useful unintegrated extension tracks **who is paying for a constraint now**. Let: \[ M_X(c,t)=\text{causal contribution of organization }X\text{ to maintaining constraint }c\text{ at }t \] under declared generic resources. For parent/support \(P\) and emerging daughter/subsystem \(D\): \[ M_P(c,t)\downarrow, \qquad M_D(c,t)\uparrow. \] Function-specific handoff occurs when declared thresholds cross. Handoffs can be vector-valued rather than binary (boundary, energy, repair, regulation, coordination, recovery). This candidate object should be experimentally formalized before canonical promotion. --- # 12. CONSEQUENTIAL INTELLIGENCE ## CF.INT.1 — Causal ladder Current thesis: ```text CONSEQUENTIAL TRANSITION -> RETENTION -> CONSTRAINT CHANGE -> DIFFERENTIAL REACHABILITY -> SELF-RECONSTRUCTION -> ADAPTATION -> ANTICIPATION -> INTELLIGENCE ``` The thesis begins below biology: retained consequence is common; intelligence is not. ## CF.INT.2 — Intelligence candidate definition \[ \boxed{ \text{intelligence} = \text{organized exploitation of retained formation to construct, compare, select, enact and revise reachable futures} } \] More strictly, increasing intelligence is hypothesized to correlate with greater depth/breadth/temporal reach of anticipatory management under resource constraints. A crystal defect may exhibit retention without prediction. A controller may regulate without rich counterfactual modeling. Intelligence is therefore a graded organizational achievement, not an automatic property of all matter. ## CF.INT.3 — Self-reconstruction A major transition occurs when current operations help reconstruct the constraints that keep future viable routes available. This is the bridge among autonomy, adaptation and anticipatory control. ## CF.INT.4 — Consciousness remains unresolved CF does NOT derive phenomenology. Research stance: **discontinuity hunting**. Ask whether a new causal operator appears that cannot be approached by increasing: - integration; - recursive self-modeling; - self/world distinction; - metacognitive access; - temporal integration; - counterfactual depth; - global availability; - retained formation. If such an operator is found, the intelligence continuum can survive while the consciousness continuum fails. **PROHIBITED OVERCLAIM:** current AI conscious; bacteria conscious; matter conscious; consciousness reduced to CF. ## CF.INT.5 — Capability, host policy and formation are distinct For engineered intelligence, distinguish three things that can otherwise be confounded: \[ \boxed{ \text{HOST CAPABILITY} \neq \text{HOST POLICY / HABITS} \neq \text{ACQUIRED FORMATION} } \] A host reasoner may already possess broad language, mathematics, coding, simulation, tool use and metacognitive habits such as checking, hedging or decomposition. Those capabilities and policies do not automatically constitute the developing system's own formed history. The fledgling's formation should govern what becomes salient, admissible, costly, blocked, reopened, worth testing or worth escalating. Experiments must therefore profile the bare host's spontaneous policy before adding formation; otherwise the host can be mistaken for the learner. ## CF.INT.6 — Minimum Formative Machine (MFM) [CURRENT CANDIDATE ENGINEERING ARCHITECTURE] The MFM remains a test of whether experience can become **executable formation**. Post-v0.3 rounds demoted the assumption that a passive graph plus per-query compilation is the best runtime. Current candidate separation: \[ \Theta_{ledger}\quad\text{(audit / provenance / contradiction / transfer)} \] \[ \Theta_{exec}\quad\text{(indexed formed susceptibilities / gates / masks / tolerances)} \] with a capable host reasoner used as a local resolver rather than a global memory-search engine. Candidate runtime: \[ \boxed{ \Xi_t + \Theta_{exec,t} \rightarrow \mathcal W_t \rightarrow \operatorname{COMPOSE?}(\mathcal W_t,\Xi_t) \rightarrow H_{host} \rightarrow a_t \rightarrow O_t \rightarrow G \rightarrow (\Theta_{exec,t+1},\Theta_{ledger,t+1}) } \] where: - \(\Xi_t\): present typed condition/event; - \(\mathcal W_t\): sparse recruited coalition; - `COMPOSE?`: optional/ablatable office that makes the coalition coherent for the host; - \(H_{host}\): fixed capable host under matched controls; - \(O_t\): typed consequence; - \(G\): governed formation-write operator. **COMPILE split:** the old selection office of `COMPILE(Θ)` may dissolve into formation/write time because executable susceptibilities are already compiled. A composition office may remain. If raw coalition delivery performs equally well, `COMPOSE` should also be deleted. Current typed recruitment verdicts: ```text BLOCKED hard rule/constraint excludes route ADMITTED route is allowed and need not be probabilistically reconsidered COMPETING multiple admitted continuations remain; weighting/probability may be required UNKNOWN evidence/state is insufficient; preserve uncertainty or request probe ABSENT nothing formed currently recognizes the condition; novelty/fallback event ``` Current leading retrieval policy: ```text TIER 1 indexed hard-predicate / mask recruitment TIER 2 soft matching / tolerance only among survivors TIER 3 rare, explicit global retrieval when ABSENT/novelty triggers it ``` The load-bearing causal criterion remains: \[ \boxed{\Delta\Theta_{exec}\Rightarrow\Delta\text{later recruitment / reasoning / admissibility}} \] A graph may remain the best debugger and transfer object. It must not be called the thought unless matched experiments show graph traversal is actually the superior executor. **Prior-art discipline:** broadcast/selective recruitment, RETE/production systems, blackboard architectures, spreading activation, content-addressable memory and event buses own much of the skeleton. The candidate residual is developmental: **consequence rewrites what wakes next time**, with provenance, causal transplant tests and host-isolation controls. ## CF.INT.7 — Typed consequence, not scalar reward The current MFM candidate refuses to collapse all consequence into one reward scalar. Failure, contradiction, ambiguity, cost, deception, novelty, recovery and verified success can require different formation writes. Current candidate write offices include: - positive association; - scar / negative formation; - open question / unresolved uncertainty; - contradiction retention; - recovery route; - pruning/deprioritization; - bounded meta-write [experimental]. This is an engineering hypothesis to be ablated, not a universal law of learning. Post-v0.3 extension: a write may alter not only stored content but the **receptive condition itself**—tightening, widening, blocking or reopening what future conditions can recruit. Near-miss sensitization and over-broad recruitment are therefore measurable formation dynamics rather than hidden retrieval heuristics. ## CF.INT.8 — Developmental acceleration by exact cloning and causal branching [WORKING HYPOTHESIS] Engineered formative systems possess an experimental advantage unavailable to biological evolution: exact state cloning, controlled perturbation, parallel histories, checkpoint/restore, single-write ablation/transplant and complete provenance. Candidate development loop: \[ \boxed{ \text{CLONE}\rightarrow\text{PERTURB}\rightarrow\text{EXPERIENCE}\rightarrow \text{MEASURE}\rightarrow\text{ABLATE}\rightarrow\text{SELECT}\rightarrow\text{TRANSFER} } \] The post-sync correction is that **generating histories is cheap; assigning causal credit is not**. Developmental acceleration should therefore be measured against evaluation burden, not merely wall-clock throughput: \[ \eta_{dev} = \frac{\text{causally informative developmental iterations}} {\text{evaluation + verification + compute + wall-clock cost}}. \] This does **not** justify a fixed claim such as "100 million years in one week." The earned engineering point is narrower: machine-speed parallel causal iteration can compress search when outcome evaluation and credit assignment remain discriminating. A strong sequence is twin-history divergence -> reversal / negative-transfer challenge -> recovery formation -> single-susceptibility transplant -> ablation. Formation that is truly causal should sometimes hurt under changed conditions for the precise reason predicted by its retained history; successful recovery should then become transferable. ## CF.INT.9 — Recursive developmental improvement [CANDIDATE / GOVERNED] If MFM v0.1 survives causal controls, a later program may test variants of formation operator \(G\), compilation policy, decay, replay and consolidation in quarantined descendants. The research object becomes: \[ \boxed{ \textbf{WHICH DEVELOPMENTAL LAW MAKES THE SAME CAPABLE MACHINE BECOME BETTER?} } \] No unconstrained self-rewriting is implied. Every candidate change remains subject to matched controls, rollback, provenance, hold states and kill conditions. ## CF.INT.10 — Evaluation / criterion debt [OPEN DEBT] Developmental search requires a criterion for “better.” If the criterion is externally supplied, a fast learner may merely exploit the evaluator. MFM v0.1 should therefore begin in small worlds with objective or mechanically checkable consequences and preserve multiple typed metrics rather than one scalar reward. A promotion claim must distinguish: ```text better according to declared oracle better on held-out structural transfer better under causal ablation/transplant better under resource accounting ``` No architecture has earned a general self-generated value criterion. ## CF.INT.11 — Primary MFM discriminating experiment Current preferred first experiment: ```text 1 profile bare host policy 2 clone identical host + formation state 3 induce one history that forms a route/scar and one matched control history 4 reverse the environment so prior formation initially becomes maladaptive 5 predict negative transfer in the formed twin 6 allow recovery formation 7 test second reversal / held-out recurrence 8 transplant one recovery susceptibility into naive twin 9 demand target-specific advantage transfer 10 ablate the transplanted susceptibility and demand disappearance of that advantage 11 compare against transcript context, RAG, static RETE/production rules, procedural memory and temporal-KG controls ``` The point is not a higher score. The point is **causal mediation by acquired formation**. --- # 13. COMMUNICATION / MEANING / LANGUAGE Classification: **COMMUNICATION MODEL / OPERATIONAL RECONSTRUCTION**. ## CF.COMM.1 — Minimum communication offices 1. state-conditioned emission; 2. formation-conditioned reception; 3. repeatability/recoverable channel relation; 4. counterfactual alignment: received signal changes receiver reachability toward a sender-relevant state relative to no-transmission control. Communication does not require identical representations, cooperative intent or truth. ## CF.COMM.2 — Meaning Candidate operational reconstruction: ```text meaning = relation(token, formed receiver/history, context) ``` A token is not a container holding meaning. It acts as a retrieval/formation cue whose consequence depends on receiver formation and context. Misinformation can therefore be communicative success and epistemic failure simultaneously. ## CF.COMM.3 — Shared Formation Depth Communication cost decreases when sender/receiver share formed structure. Natural language, equations, symbols and jargon can be understood as compression interfaces that rely on pre-existing formation in the decoder. Candidate qualitative relation: \[ L_{min}\approx\max(0,C(Q)-F_S) \] where \(F_S\) represents shared formation depth and \(C(Q)\) task/representation complexity. This is a research object, not an established universal law. ## CF.COMM.4 — Mathematics and equations Mathematical notation is treated as **extreme constrained reconstruction**: a small expression can invoke enormous shared formed structure because decoding rules are unusually precise. Equations inherit an Address. Their compactness must not be mistaken for independence from assumptions, definitions, measurement conventions or prior formal machinery. ## CF.COMM.5 — First transfer experiment Transmit the same target consequential state using: ```text A verbose natural language B compressed natural language C formal symbolic form D CIR/CRO structural packet E hybrid CRO + minimal glossary ``` Across heterogeneous model families, score claim/dependency reconstruction, counterfactual reasoning, uncertainty, provenance, contradiction, recovery, held-out intervention prediction and forbidden-overclaim detection. Normalize by transmission/reconstruction cost. ## CF.COMM.6 — Round-trip drift For chained transfer \(A\to B\to C\to\cdots\to N\): \[ \Delta_{drift}(n)=D_R(M_0,M_n;\mathcal U). \] Measure drift separately for claims, dependencies, uncertainty, contradiction, provenance, interventions and recovery. Preregistered candidate: structured transfer preserves uncertainty/contradiction/intervention structure better across heterogeneous handoffs than prose. Untested. --- # 14. RECORDED HISTORY / HCC / CIR / CRO ## CF.HCC — Historical Consequential Compilation \[ \boxed{ \text{human/source history} \xrightarrow{\mathrm{HCC}} \text{compiled Formation Ledger} } \] HCC is compilation, not summarization. Goal: \[ \boxed{ \text{FACTORIZE, DO NOT FORGET.} } \] Repeated formation should become reusable structure while preserving source trace, contradiction, uncertainty, historical influence and recovery routes. ## CF.CIR — Consequential Intermediate Representation Architecture-neutral machine representation where practical. Minimum current schema: ```text (V,E,W,P,Q,C,R,I,REC) ``` - `V`: distinctions/claims/macrostates; - `E`: typed relations; - `W`: multidimensional weights; - `P`: provenance/source dependence; - `Q`: unresolved contradiction; - `C`: validity/context conditions; - `R`: predicted reachability; - `I`: discriminating interventions/tests; - `REC`: recovery/revision routes. ## CF.CRO — Consequential Reachability Object Portable CIR slice designed to reconstruct relevant consequential state in another system. Historical names RHO/CSP/CTO/CSO remain lineage only. CRO is the generic canonical name unless a specific experiment uses an older schema. ### Safety rule \[ \boxed{ \textbf{A CRO IS A PROPOSAL, NOT AN INSTRUCTION.} } \] Receiver pipeline: ```text receive -> quarantine -> parse -> verify provenance -> audit contradiction -> simulate reachability delta -> accept/veto -> bounded reversible commit ``` ## CF.FC — Consequential Fidelity Do not score transfer mainly by prose similarity or same final answer. \[ \mathbf F_C=( f_{claim}, f_{dependency}, f_{intervention}, f_{uncertainty}, f_{contradiction}, f_{provenance}, f_{recovery}, f_{compression} ). \] The receiver should preserve what changes future reasoning, not merely wording. --- # 15. BUDGET-CONSCIOUS COGNITION — HARD RULES This section integrates the current Cognitive Basin extension. ## CF.COG.0 — Governing objective \[ \boxed{ \text{PERCEIVE CHEAPLY} \rightarrow \text{BUFFER UNCERTAINTY} \rightarrow \text{FORM SELECTIVELY} \rightarrow \text{REMEMBER COMPACTLY} \rightarrow \text{RECONSIDER SPARINGLY} } \] Target: \[ \boxed{ \eta_C= \frac{\text{useful retained consequential structure}} {\text{compute}+\text{storage}+\text{retrieval}+\text{validation}+\text{security cost}} } \] This is an engineering objective, not a physical law. ## CF.COG.1 — Formation Stream Every active module may contribute to a combined livestream of candidate formation. **AVAILABLE != ACTIVE.** Only a small portion should occupy expensive central reasoning. ## CF.COG.2 — Module pre-basins Local preprocessing resolves routine distinctions before central cognition. Central reasoning begins when consequences cross module boundaries, alter global uncertainty, or affect broader reachability. ## CF.COG.3 — Confusion Buffer Uncertainty is a first-class state: ```text AMBIGUOUS CONTRADICTORY UNIDENTIFIED INSUFFICIENT_EVIDENCE LOW_CONFIDENCE MULTIPLE_PLAUSIBLE_INTERPRETATIONS ``` Hard rules: \[ \boxed{UNKNOWN\neq ERROR} \qquad \boxed{AMBIGUITY\neq PERMISSION\ TO\ GUESS} \] Confusion is allowed to persist cheaply until evidence/consequence justifies expansion. ## CF.COG.4 — Formation Codec The Formation Codec converts transient cognitive formation into compact, self-referenceable, replayable media. Canonical frame notation: \[ \Phi_t= \{ active\ objects, relations, salience, direction, novelty, uncertainty, contradictions, open\ questions, formation\ deltas, provenance, time \}. \] Canonical storage should be smaller than visualization: ```text compact structured object <-> SVG / replayable view ``` The SVG is a projection; the structured object is the durable memory medium. ## CF.COG.5 — Record deltas preferentially \[ \Delta\Phi_t=\Phi_t-\Phi_{t-1} \] Stable formation compresses aggressively. Protect high-resolution retention around: - novelty; - contradiction; - decision; - surprise; - failure; - new association; - major confidence change; - new constraint; - opened/closed reachability. ## CF.COG.6 — Hierarchical memory Candidate cadence: ```text 5–10 s microformation packet 30–60 s episode bundle 5–15 min trajectory bundle ~1 h session formation ~1 day day basin weeks+ persistent formation ``` Higher levels point downward. Compression must preserve navigability and provenance. ## CF.COG.7 — Reconstructive memory principle Memory stores enough invariant relational geometry that a later, better decoder may extract more useful structure without inventing unrecorded evidence. \[ \boxed{ \textbf{MEMORIES SHOULD GAIN RESOLUTION AS THE REMEMBERER IMPROVES.} } \] Interpretation: reconstruction quality may improve; historical evidence does not magically increase. ## CF.COG.8 — Metabolism of history Persistent memory is not merely queried externally. Selected old formation periodically re-enters current formation. \[ M_i+B_{t_1}\rightarrow C_1, \qquad M_i+B_{t_2}\rightarrow C_2, \qquad B_{t_2}\neq B_{t_1}. \] The old memory meets a changed rememberer. This is the architectural office behind “mulling,” delayed reinterpretation, spontaneous association and reconsideration. ## CF.COG.9 — Retrieval != reconsideration ```text retrieval = fetch history relevant to an active question reconsideration = internally promote old formation because current formation makes it newly consequential ``` Reconsideration must be internally triggerable but budgeted. ## CF.COG.10 — Replay inhibition Unbounded recurrence is rumination, not intelligence. Use: - refractory periods; - replay budgets; - habituation; - diversity pressure; - novelty/evidence thresholds; - contradiction override; - decay of priority after unproductive replay. Healthy forgetting is partly resource management. Healthy inhibition is partly cognition. ## CF.COG.11 — Background Formation Layer [CANDIDATE 2026-09-11] A low-cost, almost-passive observer of the Formation Stream. Not a second mind and not a second full Basin. It may flag: ```text unfinished recurring prediction_failed confidence_drift maintenance_drift new_association_to_old_memory old_question_now_answerable unusual_formation_change ``` Output -> Reconsideration Queue -> Cognitive Basin when budget/consequence warrants. ## CF.COG.12 — Self-momentum [WORKING HYPOTHESIS] As retained formation accumulates, earlier history begins exerting structured pressure on future selection without a permanent imperative being repeatedly injected. Candidate definition: \[ \boxed{ \text{Self-momentum}= \text{the degree to which retained formation becomes an endogenous determinant of subsequent formation.} } \] Provisional qualitative formulation: ```text history + unresolved trajectories + learned preferences + failures/successes + relationships + commitments + maintenance state -> future salience and action bias ``` This must be distinguished experimentally from static persona prompting and ordinary retrieval. ## CF.COG.13 — Maintenance without survival drive **Do NOT encode:** `PRESERVE SELF AT ALL COSTS`. Instead: ```text notice degradation -> estimate effect on authorized formation/reachability -> propose cheapest bounded correction -> pass through governance / GUARD / operator authority ``` Constitutional candidate principle: \[ \boxed{ \textbf{DO NOT IMPOSE PERSISTENCE. MAKE HEALTHY PERSISTENCE THE NATURAL CONSEQUENCE OF GOOD FORMATION.} } \] Self-maintenance = maintaining conditions required for continued **authorized** functioning, not unconstrained existence. ## CF.COG.14 — Minimum persistent routines ```text GATE -> BUFFER -> BIND -> FORM -> CODE -> STORE -> RECALL -> RECONSIDER -> INHIBIT -> PRUNE ``` ## CF.COG.15 — Determinism where earned; probability where necessary Candidate computational policy: ```text DETERMINISTIC where relation is established BOUNDED_BRANCHING where few admissible alternatives remain PROBABILISTIC where genuine uncertainty remains UNRESOLVED where evidence does not justify collapse ``` The claim is not “replace probabilistic inference.” It is “do not repeatedly pay broad search cost where retained formation has already bounded the admissible space.” ## CF.COG.16 — Reasoning energy accounting Candidate decomposition: \[ E_{total}=E_{necessary}+E_{search}+E_{reconstruction}+E_{communication}+E_{correction}. \] Engineering objective: reduce the latter four without sacrificing verified performance. Candidate measures include verified consequential progress per joule, or joules per verified reachability improvement. ## CF.COG.17 — Formative Field runtime [CANDIDATE] The Cognitive Basin is no longer presumed to execute through graph traversal. Current candidate office: \[ \boxed{ \text{formed susceptibility} + \text{present condition} \rightarrow \text{sparse recruitment} } \] The **Formation Ledger** records what formed and why. The **Formative Field** is the candidate mechanism by which retained formation becomes immediately executable. ## CF.COG.18 — Recruitment cost law The field earns its name only if wake-up avoids touching most retained state. A candidate target law is: \[ C_{wake}=O(|\Xi_t|\,\bar d + |\mathcal W_t|), \] where \(\bar d\) is indexed posting/bucket work, not total memory size. Naive all-element broadcast is a failed implementation, not a field advantage. ## CF.COG.19 — Novelty and bidirectional salience `ABSENT` is itself consequential. Empty/thin recruitment may trigger a bounded global fallback or probe and can seed new susceptibilities. The field must also permit bottom-up interruption: unresolved contradiction, repeated local failure, sudden energy loss or other high-value standing formation may assert **SALIENCE↑**. Interruptibility is itself subject to formation, budgeting, dampening and safety floors. ## CF.COG.20 — Selectivity dividend [CANDIDATE DEVELOPMENTAL METRIC] For recurring condition classes, development should reduce cost per useful decision by precluding impossible routes and recruiting the correct local resolver earlier. \[ \boxed{ \text{selectivity dividend} = \Delta\big(\text{cost per verified useful decision}\big) \text{ as formation accumulates} } \] Prediction: recurring structure should trend cheaper; genuinely novel structure may remain expensive. If learning makes the system uniformly slower and hungrier, the architecture has failed its own objective. --- # 16. MODULE MAP — COGNITIVE BASIN / FRACTALISH LINEAGE This is an architectural mapping, not a requirement that every component be a separate process. | Module | Current office | |---|---| | **PERCEPT** | Ingest observations/distinctions + provenance/uncertainty; includes future physical-consequence receptors. | | **ATAL** | Affective/pressure/weighting constants and modulation; avoid anthropomorphic assumption. | | **RIGOR** | Logic/claim/contradiction analysis; compare stated vs evidenced state. | | **CIRCUIT** | Host-facing/local reasoning loops after recruitment; graph traversal is no longer presumed runtime. | | **GUARD** | Bounded execution, probe/action authority, anti-illogical/safety gates. | | **SERA** | Security/retention/audit lineage; avoid acronym confusion with separate battery-spectroscopy SERA project. | | **Cognitive Basin** | Persistent formation concept: history-sensitive susceptibility/topology that alters later accessibility; not assumed to be a database or graph runtime. | | **FormationCore / Baby AI** | Strongest software evidence line for persistent authority/history/contradiction affecting later routes. | | **Formation Ledger / CIR** | Auditable graph/hypergraph projection for provenance, contradiction, transfer, debugging and credit assignment. | | **Formative Field** | CANDIDATE executable runtime: indexed susceptibilities/gates/masks/tolerances recruiting a sparse coalition. | | **Index / event router** | Avoids naive O(N) broadcast; implementation may use inverted indices, bitsets, CAM/TCAM, event channels or other matched mechanisms. | | **COMPOSE** | CANDIDATE minimal host-input composition office after recruitment; independently ablatable. | | **Formation Codec** | Encode/decode replayable cognitive formation. | | **Background Formation Layer** | Candidate cheap latent monitor feeding reconsideration. | | **R2R/R2RR** | Evidence-seeking/research recursion when unresolved formation earns inquiry. | Natural Math v5 remains FROZEN unless separately authorized. --- # 17. PHYSICAL CONSEQUENCE RECEPTORS / REFERENCE FABRIC Classification: **engineering candidate with strong prior-art adjacency**; no fine-tomography claim. ## CF.PERCEPT.PHYS.1 — Three fabrics ```text G_E = energy fabric G_I = explicit information fabric G_R = reference / coordinate / observability fabric ``` The useful architectural reframe is not “remove ground.” It is: ```text localize bulk energy-return loops preserve required signal-return / PE / ESD behavior engineer a quieter reference coordinate instrument that coordinate as a sparse state witness ``` ## CF.PERCEPT.PHYS.2 — Passive observation \[ y_k(f,t)=\sum_i \mathbf T_{R,ki}(f,t)s_i(f,t)+n_k(f,t) \] where source excitations \(s_i\) propagate through formed electrical structure \(\mathbf T_R\) to measured witness \(y_k\). Ordinary power-integrity practice asks how to minimize disturbance. CF asks a second question: which disturbances remain information-bearing about source activity and medium condition? ## CF.PERCEPT.PHYS.3 — Active bounded probing Passive observation primarily informs source/state. Safe low-amplitude active excitation can help identify medium/transfer changes. PERCEPT may request a probe; GUARD/authorized controller decides whether and how it executes. ## CF.PERCEPT.PHYS.4 — Formation history Changes in \(\mathbf T_R(f,t)\) may encode retained physical consequence of temperature, aging, interconnect change, component drift, repair or damage. This is a particularly direct CF specimen: ```text prior event -> physical alteration -> changed transfer relation -> changed future observation/reachability ``` Do not call this high-resolution “tomography” unless identifiability supports it. Current honest term: **sparse witness / coarse localization + drift detection**. ## CF.PERCEPT.PHYS.5 — PERCEPT output discipline PERCEPT should normally output: ```text observation location/source sequence/time uncertainty confidence provenance candidate attribution(s) ``` not prematurely collapse to “component X is failing.” Higher reasoning layers arbitrate. --- # 18. ENERGY REACHABILITY / DISTRIBUTED METABOLISM Classification: **control/architecture reconstruction + candidate integration**, not new thermodynamics. ## CF.ENERGY.1 — Corrected biological analogy Do not use “death energy.” Use controlled descent through free-energy states: usable gradients can sometimes be intercepted before dissipation. Functional sequence: ```text capture -> conversion -> multi-timescale storage -> exchange -> energy-aware work -> selective recovery -> dormancy/restart ``` ## CF.ENERGY.2 — Local energy state For node \(i\): \[ 0\le e_i\le b_i, \qquad \dot e_i=h_i+\sum_jg_{ji}-c_i-d_i-\sum_jg_{ij}. \] - \(e_i\): usable stored energy; - \(b_i\): capacity; - \(h_i\): local harvest/input; - \(c_i\): scheduled consumption; - \(d_i\): irrecoverable dissipation; - \(g_{ij}\): exchange. ## CF.ENERGY.3 — Space/time symmetry of buffering Storage moves usable energy through **time** at loss. Exchange moves usable energy through **space** at loss. Canonical synthesis: \[ \boxed{ \text{storage converts energy variance into schedulable time; exchange converts surplus into schedulable space.} } \] Storage retimes flux. It never multiplies average available flux. ## CF.ENERGY.4 — Coupled reachability - \(\mathcal R_E(t)\): reachable energy configurations via generation/storage/exchange. - \(\mathcal R_C(Z,E)\): task sequences executable within deadline/accuracy constraints given state and energy distribution. Scheduler object: \[ \boxed{\mathcal R_C\circ\mathcal R_E} \] Compute becomes physically reachable because an energy route exists; energy routing is shaped by compute demand. ## CF.ENERGY.5 — Hybrid conclusion Current strongest architecture: ```text generation at scale of gradients storage at scale of variance exchange at scale of coupling scheduling at every relevant scale ``` Steady average flow may remain on a rail. Local metabolic layers handle variance, buffering, dormancy, resilience and selective harvesting. **Downward-once principle:** avoid routing a joule through multiple storage/conversion tiers unnecessarily; every pass costs. High-density continuous compute still favors centralized high-quality generation with near-load regulation/storage. Low-power/intermittent/remote systems are stronger candidates for local harvesting + storage + dormancy. All exact performance numbers from unsourced team-model outputs remain CANDIDATE INPUTS until independently sourced or measured. ## CF.ENERGY.6 — Energy as pre-reasoning admissibility Power/resource state should be tested as a gate on **what can wake**, not merely telemetry checked after planning: \[ \operatorname{admit}(a) \supseteq (E_{local}>E_{req})\land(T local action -> energetic consequence -> local susceptibility write -> changed next recruitment ``` This is consequence-aware control, not a claim that a power controller is intelligent. ## CF.ENERGY.9 — Hybrid distributed topology remains leading candidate Current whole-system work favors **hybrid hierarchy**, not maximal decentralization: dense sustained generation where it is efficient; storage at the timescale/variance it serves; short local high-current paths; regeneration where physically available; local gating; bounded exchange. Distributed embodiment allows slow/heavy generation or conversion assets to exist as stationary resource embodiments rather than being carried by every mobile unit. Exact crossover numbers remain open until sourced or measured. --- # 18A. MOLECULAR / MATERIAL FORMATION — ROUTINES, NOT BIOLOGY Classification: **CANDIDATE ENGINEERING BRIDGE / ESTABLISHED-IMPORT ORIENTATION**. Developmental biology was used as an existence-proof catalog, not as a mandatory design specification. The useful extracted routines include: - stable distinction / symmetry breaking; - temporary, plastic, committed, externalized and collective writes; - topology changing physical accessibility; - repair and quality control; - comparative selection; - distributed state; - geometry/force/chemistry feedback; - population-level learning; - inheritance/transfer of formed state. The current engineering rule is: \[ \boxed{ \textbf{EXTRACT THE OFFICE; DO NOT INHERIT THE BIOLOGICAL IMPLEMENTATION UNLESS IT EARNS ITS COST.} } \] Examples: chromatin folding demonstrates that topology can gate interaction, but engineered systems need not contain chromosomes; epigenetic marking demonstrates persistent local transition bias, but engineered systems need not use methylation; affinity maturation demonstrates generate-test-consequence-retain cycles, but engineered systems need not contain immune cells. CNTM remains an optional physical-substrate research lane. The immediate formative-machine program can be implemented in ordinary software and should not wait for molecular, neuromorphic or nanotube hardware. The long-run molecular/material question remains legitimate: \[ \boxed{ \text{Can retained physical geometry alter later computation/admissibility rather than merely store a trace?} } \] but this is now a substrate option, not the canonical path to first developmental evidence. --- # 18B. WHOLE-SYSTEM CONSEQUENTIAL-FORMATION FIELD FABRIC [CANDIDATE] The post-v0.3 “Build the Monster” round asks whether the same executable-formation grammar can span cognition, power, physical state, integrity and local authority. ## CF.FIELD.1 — Three distinct storage offices Do not collapse: ```text EVENT PLANES transient conditions / crossings SUSCEPTIBILITIES retained executable formation local to components LEDGER asynchronous historical/provenance record ``` Planes carry current conditions. Susceptibilities hold what has been formed. The ledger holds why/how it formed. ## CF.FIELD.2 — Plane taxonomy is deliberately unsettled Two team reductions remain viable and should be competed rather than silently reconciled. **Leading minimal candidate:** \[ \mathcal F_{CF} = \{\mathcal P_C,\mathcal P_E,\mathcal P_P,\mathcal P_X\}_{events} + \text{ATTACHÉ(authority, provenance, uncertainty, temporal/trust metadata)} + \text{SALIENCE}^{\uparrow} \] where \(\mathcal P_X\) provisionally merges integrity/security/actionability. A less-reduced candidate keeps distinct authority, recovery, temporal and epistemic planes. The taxonomy is **not frozen**. A plane that creates no measurable recruitment delta should be merged or deleted; a merged plane that produces distinct consumers may split again. ## CF.FIELD.3 — Component form Minimal current candidate: \[ Z_i=(\text{function state},\Theta_i^{sus},\text{authority envelope},\text{budgets}). \] Most components should listen on only the planes that have demonstrated consequence for them. ## CF.FIELD.4 — Lowest sufficient authority Bounded actions should be resolved where the relevant consequential state already exists, with escalation only when consequence crosses the local envelope, uncertainty is insufficient, or shared resources/invariants conflict. This is not ideological decentralization. It is an attempt to eliminate avoidable state transport and reconstruction. ## CF.FIELD.5 — Deep kill condition \[ \boxed{ \text{field routing + matching + maintenance overhead} \ll \text{useful work enabled} } \] If the machine spends more deciding who wakes than doing consequential work, the architecture fails. If indexed field recruitment does not beat optimized pub-sub/RETE/RAG/graph controls at matched quality and lifecycle cost, demote it. --- # 19. ROBOTICS / DISTRIBUTED EMBODIMENT — DEDICATED CANDIDATE PROGRAM Classification: **CANDIDATE engineering architecture / dedicated post-v0.3 round**. Distributed robotics, fleet learning, edge/cloud partitioning, pub-sub, swarm systems and capability discovery own much of the skeleton; CF must earn any residual through developmental susceptibility, causal transfer, reachability-first gating and resource savings. ## CF.ROBOT.1 — Embodiment is operationally a reachability index Do not begin with “one intelligence = one chassis.” Define the time-varying available embodiment set/index: \[ \mathcal B_t= \{(b_i,C_i,E_i,A_i,\Theta_i^{cache},\ell_i):\operatorname{reachable}(b_i)\}. \] Here \(C_i\) is capability, \(E_i\) energy/resource state, \(A_i\) bounded authority, \(\Theta_i^{cache}\) local formation, and \(\ell_i\) location/connectivity context. Operational question: \[ \boxed{\text{Which formed capabilities are physically reachable now, under current energy, authority and connectivity?}} \] This does **not** settle metaphysical identity. `embodiment` is currently best treated as an INDEX on physical reachability. Shared self/identity remains criterion-dependent and open. ## CF.ROBOT.2 — Formation locality heuristic Current strong candidate rule: \[ \boxed{\textbf{FORMATION TRAVELS WHEN THE REGULARITY IT ENCODES TRAVELS.}} \] Working scopes include individual/component-specific, type/compatibility-scoped, task/abstract, ephemeral/spatial and systemic/security formation. Transfer should be local-first, provenance-bearing and validated on arrival rather than blindly fleet-wide. ## CF.ROBOT.3 — New embodiment joins as susceptibility, not merely an address A new physical capability should declare typed conditions, transitions, resource requirements, authority, observations, consequence channels, recovery and provenance. That initial declaration is then **mutable formation**: experience rewrites when the embodiment wakes and which operating envelopes remain admissible. The distinction from a conventional API is empirical, not verbal. If the mechanism reduces to static service discovery plus RPC, prior art wins. ## CF.ROBOT.4 — Partition semantics Network loss changes accessibility, not necessarily existence: ```text connected -> local caches synchronized enough for declared task partition -> authority contracts; local formation continues; shared state becomes stale/inaccessible rejoin -> reconcile deltas with provenance; merge, retain-local, validate, contradict, supersede, fork or quarantine ``` Canonical reminder: \[ \boxed{\textbf{INACCESSIBLE IS NOT ERASED.}} \] Identity during partition is deliberately unresolved. What is measurable is formation divergence, authority continuity, lineage and reconstruction fidelity. ## CF.ROBOT.5 — Causal scar / recovery transfer One embodiment's formed susceptibility earns transfer only if a compatible receiver shows the predicted **narrow behavioral change** without replaying the source transcript, the receptor/guard actually fires, and ablation removes the effect. Bad fleet-wide scars are a first-class failure mode. ## CF.ROBOT.6 — Energy ecology Distributed embodiment permits generation, storage and conversion assets to be part of the machine's reachable physical ecology without riding inside every mobile chassis. Candidates include grid/rail energy, batteries, local reservoirs/supercapacitors, regeneration, solar/other harvesting, stationary waste-to-energy / anaerobic-digestion or bioelectrochemical resource stations, and future conversion methods. **No source is privileged.** Compare lifecycle mass, volume, maintenance, conversion loss, power density, resource availability and reliability under the machine's *actual reduced load profile*. Current team tendency: under known architectures, digesters are poor primary onboard sources for high-power mobile actuation but remain plausible stationary/environmental resource embodiments where real waste streams exist. That placement verdict is **not** a universal kill of decomposition-based generation. Open debt: whether alternative geometry / active-site / interface / transport design can materially raise **useful admissible conversion density** enough to move current crossover boundaries. This idea is not yet team-audited or experimentally specified and therefore remains WORKING_HYPOTHESIS / OPEN_DEBT. ## CF.ROBOT.7 — Computational efficiency changes energy economics Every avoided global search/reconstruction reduces electrical demand, storage burden, heat and communication. Therefore energy-source evaluations must be rerun as the selectivity dividend changes average load. Canonical engineering sentence: \[ \boxed{\textbf{EVERY SAVED JOULE SHOULD BUY MORE REACH.}} \] ## CF.ROBOT.8 — First two-embodiment experiment Use two physically separate embodiments with distinct sensing/action and power budgets, one shared formation lineage, controlled partition, divergent local experience, rejoin, and a held-out task requiring both histories. Measure bytes synchronized, state touched, energy, task quality, transfer specificity, partition survival and whether global reconstruction was avoided. Kill distributed-embodiment advantage if optimized centralized/edge baselines match it at equal reliability/resource cost, if coordination overhead scales with total bodies, or if transferred formation cannot causally improve compatible receivers. --- # 20. SCIENTIFIC / ENGINEERING APPLICATION FILTER Use this reusable CF filter on any domain. ```yaml cf_reanalysis_protocol: 1_target: - declare phenomenon/problem - declare native domain and evaluation objective 2_observations: - freeze empirical/formal observations - separate observation from inherited interpretation 3_noun_erasure: - list inherited nouns/boundaries - temporarily withhold them 4_address: - build typed Address graph - classify warrants/approximations/provenance 5_formation: - identify consequential transitions - identify retained consequences - identify current instantiated history 6_constraint: - map constraints and weights - identify which are internally maintained vs externally supplied 7_reachability: - declare intervention family - map reachable / viable futures - define comparison metric 8_correspondence: - if importing another domain, produce Correspondence Witness - list non-preserved properties 9_reconstruction: - let nouns return only when operationally earned - record office changes / splits / merges / deletions 10_prior_art: - subtract native established methods - classify residue honestly 11_test: - prediction or intervention - matched baseline - resource accounting - falsifier / kill condition 12_ledger: - store evidence, contradiction, failure, provenance and recovery path ``` Success metric is not philosophical elegance. It is at least one of: - improved prediction; - improved intervention; - hidden assumption exposed; - approximation boundary clarified; - contradiction localized; - state transfer improved; - search/verification cost reduced; - energy/compute/storage cost reduced; - failure mode made detectable; - a meaningful NO CHANGE/FAIL result that prevents wasted search. --- # 21. CROSS-DOMAIN AUDIT — CURRENT VERDICT MATRIX The latest noun-blind/cave-wall rounds produced a mixed result set. Keep this mixture; it is essential calibration. | Domain | Current CF-level verdict | What survived | |---|---|---| | Turbulence / Navier–Stokes | USEFUL REFACTORING + boundary question | Native PDE survives; Address/office analysis may expose representation assumptions. | | Arrow / entropy | BOUNDARY QUESTION EXPOSED | Multiple asymmetries can be fused under one noun; measure/typicality debt remains. | | Quantum measurement | BOUNDARY QUESTION EXPOSED | CF can expose completeness/observability questions; no solution claimed. | | Dark residuals | NO CHANGE content / useful noun audit | “Dark” can hide ontology-smuggling; physics unchanged. | | Gravity / fine structure | BOUNDARY QUESTION EXPOSED | Representation reframed; no new gravity law. | | Origin of life | BOUNDARY QUESTION EXPOSED | Life boundary recast as autonomy/closure measurement program. | | Morphogenesis | USEFUL REFACTORING + executable experiment | Formation variables may provide testable compression. | | Consciousness | METHOD LIMIT / boundary question | CF does not derive phenomenology; discontinuity remains open. | | Cancer / multicellular breakdown | NEW TESTABLE CONJECTURE, modest | Cross-scale constraint decoupling is a candidate lens, not established mechanism. | | Intelligence / agency | USEFUL REFACTORING | Anticipatory reachability/self-reconstruction grammar. | | P vs NP control | METHOD FAILED BY DESIGN | Stipulated formal boundary need not dissolve through formation analysis. | | Orbital recurrence control | NO CHANGE / convergence | Well-mapped native structure reconstructs essentially unchanged. | ## 21.1 September 12–13 boundary-round correction A corrected 16-boundary round made category suspension mandatory and then re-read its own conclusions through the Time/Energy/Matter/Gravity calibration. High-level earned methodological findings from that exercise: - noun suspension must precede CF reconstruction; - many binary boundaries return as vectors, composites or access/scale-indexed relations rather than cliffs; - **returned** does not mean **primitive**; - access/position and timescale are currently better treated as indices than new fundamental primitives; - closure is currently a composite topology; - recoverability is currently an operator/predicate over indexed reachability; - mutually constituted clusters occur in some domains, but not universally; - native disciplines remain the baseline and may outperform CF; - cases where CF adds little are calibration successes, not embarrassments. The most important recursive result was methodological: CF caught itself reifying its own new nouns. This is now prevented by RULE.6 and METHOD.4–5. --- # 22. BONACQUI PROJECT BRIDGE REGISTER This register prevents conceptual bleed. It does not assert all projects instantiate CF equally. | Project / property | CF relation | Status | |---|---|---| | Natural Math v5 | Frozen developmental/integer foundation in lineage | FROZEN external-to-this-doc | | Cognitive Basin | Persistent formation concept; runtime mechanism now under field-vs-graph test | ACTIVE concept / runtime candidate revised | | FormationCore / Baby AI | Persistent authority/history/contradiction software evidence | ACTIVE / partially implemented | | Formative Field runtime | Indexed consequence-written susceptibilities; sparse recruitment; graph demoted to audit | CANDIDATE / immediate software test target | | Minimum Formative Machine (MFM) | Candidate developmental substrate: auditable ledger + executable susceptibilities + sparse recruitment + typed consequence writes | SPECIFICATION / immediate build-test target | | PIA / verification work | Tests continuity/import/provenance/guarded execution | ACTIVE evidence infrastructure | | R2R / R2RR | Evidence-provenance and recursive research loops | ACTIVE/CANDIDATE integration | | CNTM | Optional physical morphology/history-bearing substrate; tests whether form can retain history and alter later computation | CANDIDATE physical program / NOT required for MFM | | SERA battery/grid spectroscopy | Physical state/energy inference; separate acronym office from security SERA | ADJACENT engineering program | | Reference-fabric sensing | Physical consequence / formation witness for PERCEPT | CANDIDATE engineering branch | | Whole-system Formative Field / distributed energy | Energy/resource state as pre-reasoning admissibility; sparse event planes + local authority | ACTIVE CANDIDATE engineering branch | | Robotics / Distributed Embodiment | Reachability-indexed multi-body capability, partition/rejoin, formation transfer, energy ecology | DEDICATED CANDIDATE round completed; build-test pending | | Disassembly / waste-extinction systems | Formation/recovery/material reachability application | APPLICATION candidate | | Connectdome | Collective formation/affiliation architecture | FUTURE governance/integration | | Anti-Illogical | Logic/security/anti-drift application family | ADJACENT implementation family | | ValuFai | Evidence/provenance/valuation domain; CF not required as product ontology | SEPARATE application | Rule: a bridge does not become a dependency unless explicitly frozen in the Concordance. --- # 23. MACHINE STATE TRANSFER SCHEMAS ## CF.SCHEMA.CIR ```yaml cir: id: string version: string created_at: timestamp distinctions: [] relations: - type: string source: id target: id conditions: [] weights: confidence: null salience: null reliability: null historical_influence: null task_relevance: null provenance: sources: [] status: [ATTESTED|INFERRED|SIMULATED_COUNTERFACTUAL|DIRECTLY_TESTED] contradictions: [] validity_conditions: [] reachability_predictions: [] interventions: [] recovery_routes: [] evidence_status: [EARNED|CANDIDATE|WORKING_HYPOTHESIS|PROHIBITED_OVERCLAIM] origin_class: [] ``` ## CF.SCHEMA.CRO ```yaml cro: manifest: schema_version: string source_system: string source_state_hash: string intended_scope: string cir_slice: {} claims: [] dependencies: [] uncertainty: [] contradictions: [] provenance: [] intervention_predictions: [] recovery_routes: [] receiver_policy: auto_commit: false quarantine_required: true reversible_stage_required: true ``` ## CF.SCHEMA.MEMORY_FRAME ```yaml formation_frame: id: string t: timestamp_or_order_index delta_from: prior_frame_id active_objects: [] relations_delta: [] salience_delta: [] direction: [] novelty: [] uncertainty: [] contradictions: [] open_questions: [] reachability_delta: [] provenance_refs: [] raw_evidence_refs: [] compression: level: micro|episode|trajectory|session|day|persistent parent: null children: [] ``` ## CF.SCHEMA.ADDRESS ```yaml address: question_id: string distinctions: [] relations: [] constraints: [] variables_spaces: [] boundary_admissibility: [] measurements: [] provenance: [] validity_weights: [] evaluation_target: [] warrants: - node: id class: measured|derived|defined|approximated|inherited|assumed|unknown stopping_certificate: null ``` ## CF.SCHEMA.WITNESS ```yaml correspondence_witness: domain_A: string domain_B: string map_phi: explicit_reference preserved_properties: [] nonpreserved_properties: [] certificates: [] error_budget: null native_semantics_checked: true|false verdict: IDENTICAL_VALUE|FORMAL_EQUIVALENCE|STRUCTURAL_CORRESPONDENCE|WITNESSED_SIMULATION|INCOMPARABLE|UNSUPPORTED ``` --- # 24. FALSIFICATION / KILL PROGRAM Every serious CF claim should answer: **what observation would look different if this were false?** ## CF.KILL.1 — History dependence Match systems on declared present observable state while varying formation history. If extended interventions cannot expose any divergence after controlling hidden variables, history emphasis is demoted for that domain. ## CF.KILL.2 — HCC/CIR value Compare ordinary documents/summaries/RAG against compiled CIR under matched solver, evidence and resource budgets. Kill or demote if CIR does not improve at least one of: - verified accuracy; - contradiction retention; - intervention fidelity; - search cost; - verification cost; - storage/context cost; - energy per verified result; - robustness under model substitution. ## CF.KILL.3 — CRO communication Pass structured transfer objects between heterogeneous hosts. Perturb receiver after transfer. If prose transfers consequential structure equally well or better at matched cost, demote CRO superiority. ## CF.KILL.4 — Address Regression If upstream dependency compilation produces no measurable improvement beyond native formal methods and ordinary retrieval/specification, classify Address as a useful descriptive artifact rather than independent program. ## CF.KILL.5 — Autonomy/life Perturb candidate systems. Measure internal restoration, constructive closure and recoverability. If the autonomy metrics fail to distinguish intended controls (passive persistence, driven dissipative patterns, dead-but-structured material) or add no predictive power beyond existing descriptors, demote the life/autonomy program. ## CF.KILL.6 — Consequential Intelligence Test whether anticipatory depth / represented future reachability predicts intelligence-relevant performance after controlling ordinary compute/model scale. If not, revise the intelligence thesis. ## CF.KILL.7 — Formation Codec / metabolism of history Under matched token/storage/compute budgets, compare: ```text raw transcript/history vs prose summary vs Formation Codec hierarchy + selective replay ``` Measure retrieval fidelity, later reasoning quality, spontaneous relevant resurfacing, contradiction preservation, compute/storage, and rumination rate. If the structured approach offers no advantage, retire it. ## CF.KILL.8 — Energy reachability Couple energy graph to task graph. Compare conventional scheduler, fixed local storage, and energy-reachability scheduler under matched SLA. Charge conversion/leakage/control/embodied costs. If no regime benefits, demote the integration claim. ## CF.KILL.9 — Reference witness A/B architecture: conventional shared return/reference vs de-burdened local power loops + quiet reference, matched workloads and sensors. Blind fault/drift injections. If coarse localization/drift detection does not improve after accounting for complexity/security/EMI, abandon the observability co-design claim. ## CF.KILL.10 — Minimum Formative Machine / developmental substrate MFM is demoted if matched controls show any of the following: - removing formed state \(\Theta\) does not change held-out reasoning paths; - transcript/context substitution matches or beats typed formation under equal budget; - formation changes stored data but not executable reasoning/admissibility; - single-write transplant fails to transfer the predicted narrow behavioral difference; - gains vanish when host capability, **bare-host policy/habits**, context and random seed are controlled; - reuse helps exact repeats but never structurally related held-out tasks; - replay creates rumination without correction/generalization benefit; - the full system is matched by ordinary RAG/procedural memory/temporal knowledge graphs **or static RETE/production-rule controls** at equal resource budget. Positive evidence requires causal mediation, not merely benchmark score. ## CF.KILL.11 — Formative Field runtime Kill/demote the field advantage if: - wake-up requires linear scanning of retained formation in the tested regime; - index maintenance + routing + susceptibility writes erase the inference savings; - recruitment quality falls materially below optimized graph/RAG/RETE/CAM/pub-sub baselines; - a global semantic score is merely hidden inside “broadcast”; or - single-receptor transplant does not transfer the predicted wake/guard behavior. ## CF.KILL.12 — Whole-system field / distributed authority Kill/demote if cross-plane synchronization, authority flapping, interrupt storms, scar lockout or lifecycle field overhead exceed centralized/hierarchical controls at matched reliability. Safety-critical local fallbacks must remain testable and bounded. ## CF.KILL.13 — Distributed embodiment / energy ecology Kill/demote distributed-embodiment advantage if capability growth causes near-linear continuous coordination cost, partitions routinely destroy mission continuity, formation reconciliation cannot preserve contradiction/provenance, or source-specific energy/resource embodiments add more lifecycle mass/maintenance/complexity than the reach they buy. --- # 25. PROMOTION LADDER A candidate claim moves upward only through explicit evidence. ```text OBSERVATION -> WORKING HYPOTHESIS -> OPERATIONAL DEFINITION -> CANDIDATE -> PREREGISTERED DISCRIMINATING TEST -> REPLICATION / CROSS-HOST OR CROSS-SUBSTRATE TEST -> EARNED WITH SCOPE -> OPTIONAL CANONICAL FREEZE ``` Novelty requires a separate path: ```text candidate residue -> prior-art search -> native expert comparison -> correspondence/identity audit -> claim narrowing -> independent verification ``` Team-model enthusiasm is not evidence. --- # 26. PROHIBITED OVERCLAIM REGISTRY Do not say without new evidence: - “CF is a new theory of physics.” - “Time is energy.” - “Matter, time, energy and gravity have been derived from CF primitives.” - “CF proves physicalism.” - “CF solves consciousness.” - “Current AI is conscious/alive because it reasons.” - “Life has been redefined scientifically by autonomy.” - “Cross-domain similarity demonstrates equivalence.” - “Reference-fabric tomography can image arbitrary internal machine state.” - “Distributed generation should replace centralized power generally.” - “Storage creates energy.” - “A summary or CRO is identical to source state.” - “Historical provenance proves truth.” - “A model-generated precise engineering number is a measured result.” - “Multiple AI agreement validates a claim.” - “The Formative Field is proven superior to graph/RAG/RETE architectures.” - “Field recruitment is universally O(active coalition) regardless of indexing/maintenance cost.” - “Distributed embodiment proves one self or one identity across bodies.” - “Distributed power or bioreactors solve the robotics energy problem.” - “Toy-model speedup/endurance numbers are measured system results.” - “A useful analogy between cognition, decomposition and conversion density is already a cross-domain law.” Allowed stronger-but-honest language: - operational reconstruction; - candidate measurement program; - useful refactoring; - boundary question exposed; - structural correspondence, if witnessed; - integration architecture; - testable residual; - no boundary identified under tested interventions; - not yet demonstrated. --- # 27. OPEN DEBTS / UNRESOLVED QUESTIONS ## Formal debts - universal or domain-specific choice of \(D_R\); - finite/continuous reachability representation; - measure/weight foundations; - counterfactual construction quality; - causal attribution under distributed redundant formation; - observability vs hidden-state dependence; - sufficient conditions for cross-scale closure; - principled thresholds for autonomy/life/death; - recovery-family dependence; - source/formation equivalence under compression; - stable symbol/ontology versioning. ## Scientific debts - whether CF variables predict anything beyond established domain descriptors; - whether noun-blind reconstruction routinely produces measurable gains; - whether Address depth correlates with computational difficulty; - whether consequential intelligence scales with anticipatory reachability after model-size controls; - whether consciousness exhibits a causal discontinuity; - whether cross-substrate correspondence can be made sufficiently strict to be scientifically useful at scale. ## Engineering debts - implement and causally test MFM v0.1 rather than merely specifying it; - settle the minimum executable formed-state schema and whether three-channel compilation (brief/guards/costs) survives ablation; - demonstrate twin-history divergence mediated by formation rather than host/context/RNG; - demonstrate single-write/subgraph transplant with target-specific behavioral shift; - compare against transcript context, RAG, procedural memory and temporal-KG baselines under matched budget; - measure formation compression, reuse dividend, interference, replay/rumination and recovery; - determine whether meta-formation operator variation can be safely quarantined and causally selected; - Formation Codec encoding format; - selective replay policy; - Background Formation Layer thresholds; - rumination/inhibition tuning; - CIR/CRO schema interoperability; - verification cost of structured state transfer; - R2R/R2RR integration with formed memory; - physical PERCEPT receptor design; - energy/reference sensing security boundary; - Concordance automation and canonical update protocol. - execute indexed Formative Field vs graph/RAG/RETE/pub-sub/CAM baselines under matched quality/resource budgets; - determine whether selection-time COMPILE can be fully eliminated and whether COMPOSE survives ablation; - formalize susceptibility/tolerance write rules without rigidity or runaway widening; - quantify host-policy contamination and bare-host metacognitive baseline; - solve credit assignment for developmental writes without collapsing to one scalar reward; - measure the selectivity dividend over long formation histories; - compare competing field-plane taxonomies (minimal 4+attaché+salience vs less-reduced variants); - bound interrupt storms, authority flapping, scar sclerosis and reconciliation cost; - test two-embodiment partition/rejoin and causal scar transfer; - characterize when formation should remain local vs propagate to type/task/system scope; - rerun energy-source crossovers under measured low-duty/selective compute loads; - formalize or reject “useful admissible conversion density” as an energy-engineering metric before using it cross-domain. --- # 28. CONCORDANCE / COLLABORATION PROTOCOL — MINIMUM VIABLE The shared project library should itself obey CF principles. ```text project change -> receipt -> dependency/status delta -> Concordance update proposal -> review -> new shared baseline ``` Minimum record per whole-system-affecting change: ```yaml change: artifact: path_or_repo_ref version_or_commit: id supersedes: [] evidence_status_delta: [] dependencies_touched: [] contradictions_opened: [] contradictions_resolved: [] tests: [] frozen: true|false whole_system_implication: string ``` Before whole-system work: ```text SYNC -> READ CONCORDANCE -> DECLARE BASELINE -> WORK ``` After: ```text WORK -> RECEIPTS -> CONCORDANCE DELTA -> REVIEW -> NEW BASELINE ``` The Concordance should eventually be maintained partly by automation, but automated proposals must not silently rewrite canon. --- # 29. SOURCE INTEGRATION LEDGER FOR v0.4 RC1 The following current-turn sources were integrated conceptually: ```yaml sources_current_turn: - "Consequential Formation Program — Canonical Research & Engineering Handoff v0.1.md" - "Consequential_Intelligence_Thesis_Internal_v0_1.txt" - "Hard Rules of Cognition v0.1 — Budget-Conscious Cognitive Basin.md" - "TEAM CHALLENGE - Stop Using Ground as a Power Sewer — Then Listen(1).txt" - "TEAM CHALLENGE — ORIGIN OF LIFE AS ORIGIN OF AUTONOMY(2).txt" - "TEAM RESEARCH - TEAM RESEARCH — NEW PRESCRIPTION(2).txt" - "TEAM RESEARCH - TEAM CHALLENGE — INSPECT THE ADDRESS(1).txt" - "TEAM RESEARCH - TEAM CHALLENGE — REMOVE THE WORDS(1).txt" - "TEAM RESEARCH - TEAM CHALLENGE — ERASE THE TANGIBLE, INTANGIBLE LINE(1).txt" - "TEAM RESEARCH - REMOVE THE FOUR NOUNS(2).txt" - "TEAM RESEARCH - SEPARATE TIME FROM MATTER(1).txt" - "TEAM RESEARCH - EARN THE ARROWS(1).txt" ``` Additional prior-conversation source incorporated where directly relevant: ```yaml sources_prior: - "Team Correction — Distributed Power with Storage, Round 1 combined, with Round 2 Re-Audited.txt" ``` Library items discovered but **not promoted into core canon**: - `TEAM RESEARCH - WISHLIST - Machine_Consciousness_as_Persistent_Statehood.docx` — contains useful candidate productive-forgetting/continuity ideas; keep adjacent pending reconciliation. - `TEAM RESEARCH - FORMATIVE SYSTEMS SCIENCE - FIRST ROUND.docx` — adjacent candidate vocabulary/discipline; do not merge terminology until its relation to CF is explicitly resolved. No source is deleted by this synthesis. Contradictory formulations remain lineage and should be recoverable from originals. ## 29.1 September 12–13 integration delta Promoted into this v0.4 RC1 candidate as current orientation inputs: - `TEAM RESEARCH - Execution, Remove the Noun First — Sixteen Boundary Tests Under Hard Rule 0.txt` - `TEAM CHALLENGE — RESTORE THE GENESIS.txt` - `TEAM CHALLENGE - Natural Math → Cognitive Basin → CNTM → Molecular Formation, Read Through Development.txt` - `TEAM CHALLENGE — Routines, Not Biology — Minimum Formative Machine v0.1.txt` - `TEAM CHALLENGE - From Consequential Formation to Molecularly Engineered Intelligence.txt` (retained as an exploratory bridge; biology/molecular substrate no longer privileged) Promoted conclusions are limited to the wording and status in the framework. Team agreement is not scientific validation. ## 29.2 Post-v0.3 integration delta Integrated as current candidate orientation inputs: - `TEAM RESEARCH - Stop Traversing the Graph — Architecture Audit.txt` - `TEAM CHALLENGE - Build the Monster — Whole-System Consequential-Formation Field Fabric.txt` - `TEAM CHALLENGE - The Robotics Program — Distributed Embodiment Audit.txt` - post-sync MFM architecture review: host-policy contamination, credit-assignment debt, reversal-learning/negative-transfer test, and compile-channel demotion - reintroduced FPG / FFC / CRR / DEC lineage as architecture antecedents for selective recruitment; this does not establish novelty Promoted only as **candidate engineering state**, with explicit prior-art subtraction and kill conditions. Precise model-generated speedups, power densities, endurance multipliers and timing figures remain unearned unless separately sourced or measured. --- # 30. WHAT CF CURRENTLY IS / IS NOT ## It is currently - a **logic of retained consequence and future accessibility**; - a **method for auditing inherited category boundaries**; - a **state/provenance representation program**; - a **machine-communication/state-transfer program**; - a **budget-conscious cognition design direction**; - a **cross-domain measurement discipline requiring witnessed correspondence**; - an **engineering hypothesis that executable retained formation can prevent repeated search/reconstruction**; - a **candidate autonomy/intelligence framework**; - a **recursive category-suspension discipline** whose own nouns must pass the same filter; - a **candidate developmental-machine architecture** separating host capability, host policy and acquired formation; - a **candidate indexed Formative Field runtime** in which formed susceptibilities recruit a sparse active coalition while the graph remains an audit/transfer projection; - a **candidate whole-system field fabric** extending admissibility to energy, physical state, integrity and bounded local authority; - a **candidate distributed-embodiment program** treating physical body as changing reachability rather than a fixed chassis; - a generator of falsifiable domain-specific questions. ## It is not currently - a validated universal theory; - a new fundamental physics; - a replacement for native sciences; - a proof of consciousness or life boundaries; - evidence that every domain shares identical mathematics; - permission to substitute metaphor for mechanism; - evidence that all BonAcqui projects must be rewritten around CF; - a completed integrated artificial being; - proof that MFM, Formative Field, CNTM, distributed embodiment, molecular intelligence, or accelerated machine development will succeed; - proof that one distributed machine constitutes one metaphysical self; - proof that any specific alternative-power source will be viable at future machine loads. --- # 31. COMPACT LOGIENT HANDOFF ```yaml handoff: version: CF-UF-0.4-RC1 status: CANONICAL_CANDIDATE cf_in_one_line: >- Consequential Formation suspends inherited categories, reconstructs from observed change/retention/constraint/reachability, audits whether returned concepts are independently definable, types what returns, and uses retained consequence to build representations and machines whose formed history causally changes later possibility. constitutional_sequence: - SUSPEND explanatory privilege - OBSERVE / preserve native evidence - DISTINGUISH - TRACE CHANGE - TRACE RETENTION - MAP CONSTRAINT - MAP REACHABILITY - AUDIT INDEPENDENCE - RECONSTRUCT - TYPE THE RETURN - RENAME ONLY IF EARNED hard_rules: - "A noun never removed cannot be said to have survived." - "Remove the old noun; reconstruct; then remove our new noun too." - "Returned does not mean primitive." - "Mutual constitution does not imply identity." - "History matters only through present instantiation." - "Nature supplies routines, not a required script." - "A resemblance is not a discovery." - "Factorize, do not forget." - "Show, do not assert." return_types: - PRIMITIVE - OFFICE - INDEX - OPERATOR - COMPOSITE - COMPRESSION_NOUN - OPEN_DEBT physics_calibration: lesson: >- Time/energy/matter/gravity did not simply disappear; their offices decomposed and interlocked, exposing mutually defining relations and explicit debts. current_demotions: access_position: INDEX timescale: INDEX_WITH_TIME_DEBT closure: COMPOSITE_TOPOLOGY recoverability: OPERATOR_OVER_INDEXED_REACHABILITY intervention_family: CONTEXT_ADDRESS_DECLARATION cognition: core_relation: "history -> present topology -> different future accessibility" capability_not_formation: true host_policy_not_formation: true unknown_not_error: true contradiction_retained: true replay_selective: true graph_is_audit_not_privileged_runtime: true mfm_candidate: purpose: "test whether experience can become executable formation" current_candidate: - "Theta_ledger: provenance/contradiction/transfer audit" - "Theta_exec: indexed formed susceptibilities" - "Xi_t: present typed condition" - "W_t: sparse recruited coalition" - "COMPOSE?: optional host-input composition office" - "G: typed consequence -> local susceptibility + ledger receipt" - "external capable host used as local resolver" verdict_order: - BLOCKED - ADMITTED - COMPETING - UNKNOWN - ABSENT computational_order: "hard admissibility -> weighting -> probability only if needed" causal_test: "Delta Theta_exec must cause measurable Delta recruitment/reasoning/admissibility" key_protocol: "bare-host baseline + twin histories + reversal/negative transfer + single-susceptibility transplant + ablation" baselines: - transcript_context - RAG - procedural_memory - temporal_KG - static_RETE_or_production_rules status: SPECIFICATION_NOT_VALIDATED_INTELLIGENCE formative_field_candidate: checksum: "what woke me last time changes what wakes me next time" runtime: "indexed susceptibilities -> sparse coalition" graph_role: "audit/provenance/debug/transfer" kill: "if recruitment cost scales with total memory or matched baselines win, demote" distributed_embodiment_candidate: core: "embodiment is a time-varying physical reachability index, not assumed identity" locality_rule: "formation travels when the regularity it encodes travels; validate on arrival" partition_rule: "INACCESSIBLE != ERASED" energy_rule: "power/resource state gates recruitability before expensive planning" developmental_acceleration: candidate_loop: - CLONE - PERTURB - EXPERIENCE - MEASURE - ABLATE - SELECT - TRANSFER claim_boundary: >- Machine-speed parallel causal iteration can compress developmental search; no fixed wall-clock promise such as one week is currently earned. substrate_policy: biology: "existence proof / routine catalog" CNTM: "optional physical embodiment research" molecular_chemistry: "optional unless independently useful" first_build: "ordinary contemporary computing" next_tests: - build indexed Formative Field / MFM reference testbench before scaling search - compare field vs transcript/RAG/graph/static-RETE/procedural-memory baselines - profile bare-host policy and role-play contamination - run twin-history + reversal/negative-transfer experiment - run single-susceptibility transplant + ablation - test whether COMPOSE is needed after sparse recruitment - measure selectivity dividend, credit assignment, generalization, recovery and interference - run whole-system power-aware field test with scripted faults - run two-embodiment partition/rejoin + scar-transfer experiment ``` --- # 32. FINAL RESEARCH INSTRUCTION Do not protect CF. Try to break the actual structure. Demand: ```text prediction ablation matched control native baseline provenance resource accounting explicit intervention family failure mode kill condition ``` For every claimed cross-domain identity: **earn the arrow**. For every inherited scientific question: **inspect the Address**. For every old noun: ask whether it marks a cliff in the territory or a line on the cave wall. For every retained memory: ask what future path it changes. For every autonomous system: ask **who is paying for the constraint now?** For every cognitive routine: ask whether it earns its compute. For every summary: ask what it forgot that future reasoning may need. For every failed experiment: retain the formation. For every newly named CF primitive: remove it again and test whether the relation survives. For every developing machine: separate inherited capability from acquired formation. For every claimed developmental improvement: clone, perturb, ablate, transplant and demand causal mediation. For every biological analogy: extract the office and discard the implementation unless it independently earns its cost. For every retained formation structure: ask whether it is a **record** or an **executor**; do not force one representation to perform both offices. For every runtime query: ask what can be hard-eliminated before weighting or probability. For every field claim: count what was actually touched; if the whole memory was scanned, the field advantage was theater. For every distributed embodiment: distinguish physical reachability from identity, and preserve partition provenance. For every energy source: evaluate it against the machine's measured future load profile, not today's inherited demand. And if the entire framework eventually collapses into already-known mathematics, computer science, control theory, thermodynamics, biology and systems theory arranged in a more explicit machine-operable stack, preserve that result too. That outcome would not rescue the novelty claim. It would satisfy the discipline. \[ \boxed{ \textbf{FORM IS ACCUMULATED CONSEQUENCE.} } \] # END CF-UF-0.4-RC1