--- title: "Cross-Thesis Dependency Map" summary: "A bounded component of the Consullo public research program: Cross-Thesis Dependency Map." status: "specified but not implemented" provenance: "derived from the owner-approved private Consullo design corpus; no artifact-specific public receipt has been issued" claim_ids: ["CP-001"] last_reviewed: "2026-08-12" receipt: "none" non_claims: ["Publication does not establish implementation or operational effectiveness.", "The implementation-evidence appendix records component evidence only and is not evidence of operational capability for this page."] --- # Cross-Thesis Dependency Map Version: 0.3.1 This file defines the import/export contract among the five Consullo Seed AI theses. It exists to prevent the suite from becoming a set of overlapping agent rosters. ## Dependency Principle Each thesis has one primary conceptual responsibility. Agents may appear in multiple discussions, but every agent or function should have one primary home. Cross-thesis use should be described as an import rather than duplicate ownership. Imports are marked as: - definitional: terminology, interface, or conceptual context only - load-bearing: required for the thesis argument or runtime architecture ## Capability Flow ```text Thesis 2: Cognitive Substrate -----------------------> Thesis 1: Validated Improvement Loop \ ^ \ | v | Thesis 4: Self-Modifying Software Substrate ----------+ ^ | / | Thesis 3: Causal-Decision Foundations ----------------+ ``` Capability flow shows how capabilities feed the improvement loop. Thesis 2 and Thesis 3 both feed Thesis 1 directly and also inform Thesis 4. It does not show permissioning. This diagram shows capability flow only, not all definitional imports. For example, Thesis 4 imports Thesis 1's acceptance vocabulary, but that back-import is definitional rather than a capability-flow dependency. ## Constraint Wrapper ```text +-----------------------------------------------------------------------+ | Thesis 5: Alignment Invariants And Scoped Trust | | | | constrains Thesis 1, Thesis 2, Thesis 3, Thesis 4, substrate context | | via AAF gate, scoped permission, trust estimates, human authority | +-----------------------------------------------------------------------+ ``` Thesis 5 is not downstream output of the improvement loop. It is a constraint layer over all theses and all substrate context. ## Organizational Operating Layer `appendix-organizational-recursive-self-improvement.md` defines a cross-thesis operating layer, not a sixth thesis. It interprets the five theses as an AI-native R&D organization whose work product is validated improvement of research, engineering, evaluation, memory, and governance processes. | Organizational function | Primary thesis imports | Evidence boundary | | --- | --- | --- | | Agenda and portfolio formation | Thesis 1 improvement targets; Thesis 3 portfolio reasoning; Thesis 5 permission boundaries | Specified/proposed until `research_agenda` and `portfolio_decision` ledger records exist. | | Hypothesis and candidate generation | Thesis 2 cognitive search, brainstorming, negative-space mapping; Thesis 1 proposal semantics | Candidate generation only; accepted improvement requires downstream gates. | | Pre-registration and experiment design | Thesis 3 causal-decision discipline; Thesis 1 evidence packages | Capability claims require pre-registered success/failure criteria. | | Implementation and validation | Thesis 4 software substrate; Thesis 1 acceptance semantics | Java repair is exploitation evidence, not full organizational RSI evidence. | | Adversarial review and governance | Thesis 5 AAF, scoped trust, owner authority; Thesis 1 acceptance gates | High-stakes or externally consequential actions route through I12. | | Institutional memory and post-cycle learning | Thesis 1 method-memory update; Thesis 2 memory and anti-library functions | Organizational learning requires later measured reuse or transfer. | The live execution controls for this layer are specified in the internal execution plan. That document is operational, not part of the publication thesis claim. ## Cycle-Breaking Rule There is an unavoidable conceptual relationship between Thesis 1 and Thesis 2: cognitive agents help improve the system, and the improvement loop evaluates cognitive agents. This is not a circular proof if the import types are separated: - Thesis 1 imports Thesis 2 definitionally for cognitive interfaces and load-bearing for current cognitive capabilities used by proposer/evaluator agents. - Thesis 2 imports Thesis 1 definitionally for the improvement-evaluation vocabulary, but its core claim about cognitive substrate can be stated without assuming Thesis 1 succeeds. No thesis may use its own future success as evidence for its current capability claim. ## Master Frame The master frame exports: - definition of Consullo Seed AI - no-ASI-status-claim constraint - greater-than-human capability measurement schema - shared vocabulary and invariants - single-owner Phase 1 baseline - Adversarial Alignment Function - Abundance Distribution Obligation - alignment-monoculture mitigation requirement - evidence-ledger schema with indexed views, as specified in `appendix-evidence-ledger-schema.md` - capability-status enum - legacy-agent-name rule - literature survey expectations All theses import the master frame. ## Substrate Context Substrate context is not a sixth thesis, but it should be documented in `appendix-substrates.md`. Substrate context owns: - `SpecializedLanguageModelEcosystemDirector` - specialized LLM provider abstraction and routing - rapid knowledge access infrastructure - atomic prompt decomposition and compiled-code orchestration - internal economy and resource accounting - `DigitalVirtualEconomyOrchestrator` - `BusinessFunctionOrchestrator` - `ServiceProvidingEntitiesOrchestrator` - model routing, token-cost accounting, and provider selection Thesis 5 may evaluate these systems for trust and alignment. It does not own them as alignment mechanisms. ## Thesis 1: Validated Improvement Loop Primary responsibility: Model recursive capability amplification as a staged, evidence-gated loop over agents, method memories, code, tests, workflows, policies, and agent populations. Exports: - improvement-cycle semantics - proposer/evaluator/validator distinction - acceptance-gate structure - evidence package requirements - method-memory update semantics - credit-assignment requirements - false-accept and false-reject treatment - benchmark and cost requirements for accepted improvements - mandatory Capability Status tagging for improvement claims - improvement evidence-ledger view Imports: - from Thesis 2: cognitive interfaces and current cognitive capabilities used to generate, critique, and evaluate improvements - from Thesis 3: causal-decision objectives, intervention evaluation, uncertainty estimates, and Goodhart analysis - from Thesis 4: code generation, repair, testing, deployment, provenance, and software-modification mechanisms - from Thesis 5: alignment invariants, AAF non-veto requirements, permissioning, scoped trust, human authority, rollback, and containment - from substrate context: model routing, token-cost accounting, rapid knowledge access, and internal resource accounting Primary agent/function home: - `SelfImprovementOrchestrator` - `SeedAIManager` - `AgentImprovementGoalSetter` - `AgentCapabilityImprover` - `AgentComplianceValidator` - `RecursiveEnhancementTracker` - `MethodMemoryExtractor` - `MethodMemoryOrganizationOrchestrator` - `ConstructiveRuleChallenger` - `AutonomousTaskProposer` - `QualityCostBalancer` - `ExplorationPortfolioManager` - DGM proposer/evaluator/validator/selection/archive roles Not owned here: - `AgentBuilder`, `AgentEditor`, and `WorkflowAutomationCompiler` are imported from Thesis 4. - Causal-model selection and robust intervention choice are imported from Thesis 3. - Alignment vetoes and trust gates are imported from Thesis 5. ## Thesis 2: Cognitive Substrate Primary responsibility: Model Consullo's modular cognitive substrate for capability amplification: memory, reasoning, perception, attention, metacognition, social cognition, creativity, executive control, knowledge access, and orchestration. Exports: - cognitive capability profiles - typed cognitive-agent interfaces - memory and knowledge access functions - attention and metacognitive control functions - social-modeling functions - cognitive composition and integration-cost model - capability-status-tagged cognitive claims Imports: - from Thesis 1: improvement-evaluation vocabulary, benchmark results, and post-deployment learning evidence - from Thesis 3: causal-decision layer for intervention-quality decisions - from Thesis 4: software substrate for implementing cognitive agents - from Thesis 5: permissioning and alignment constraints - from substrate context: LLM ecosystem, rapid knowledge access, and atomic prompt decomposition Primary agent/function home: - `ExecutiveFunctionOrchestrator` - `GoalFormationArchitect` - `StrategyFormulationDesigner` - `SubgoalDecompositionPlanner` - `ResourceAllocationOptimizer` - `ProgressMonitoringAgent` - `TaskPerformanceExecutor` - `KnowledgeFunctionsOrchestrator` - `SemanticKnowledgeOrganizer` - `EmbeddingIndexer` - `MemoryFeedbackProcessor` - `MethodMemoryGenerator` - `MMCanonicalMethodRetriever` - `AbductiveExplanationGenerator` - `DeductiveReasoningProcessor` - `InductiveGeneralizationAnalyzer` - `AbstractionLevelProcessor` - `AnalogicalMappingProcessor` - `ConceptualBlendingCreator` - `UncertaintyAssessmentAnalyzer` - `ErrorDetectionProcessor` - `AttentionRegulationManager` - `CognitiveEffortAllocator` - `LearningStrategySelector` - `VisualPerceptionProcessor` - `AudioPerceptionProcessor` - `MultimodalFusionProcessor` - `FeatureExtractionAnalyzer` - `AnomalyDetectionSpecialist` - `SalienceDetectionProcessor` - `FocusShiftingCoordinator` - `SustainedMonitoringAgent` - `TemporalBindingCoordinator` - `CollectiveInsightSynthesizer` - `GlobalCoherenceIntegrator` - `ExhaustivePatternMatcher` - `DeepAnalogicalTraverser` - `CrossDomainSynthesizer` - `FailurePatternAnalyzer` - `TemporalPatternExtractor` - `ParallelReasoningCoordinator` - `InformationGainOptimizer` - `CognitiveDepthRegulator` - `IdeationProcessor` - `CreativeIdeaEvaluator` - `CuriosityDrivenExplorer` - `PlayfulExplorationAgent` - `ConceptualAssociationMapper` - `NegativeSpaceMapper` - `PatternPriorSynthesizer` - `FailureAntiLibraryManager` - `ClarificationSeeker` - `ParallelHypothesisManager` - `SustainedReasoningManager` Theory-of-mind agents primarily support the Adversarial Alignment Function and are owned by Thesis 5, while Thesis 2 may describe them as cognitive primitives: - `BeliefModelingProcessor` - `IntentionRecognitionAnalyzer` - `PerspectiveTakingModeler` Not owned here: - Causal decision agents are owned by Thesis 3. - Code generation and repair agents are owned by Thesis 4. - Formal verification primitives are owned by Thesis 4. - Trust, alignment, and governance agents are owned by Thesis 5. - Consciousness claims are non-load-bearing and should not be central to this thesis. ## Thesis 3: Causal-Decision Foundations Primary responsibility: Model causal prediction, counterfactual reasoning, causal influence diagrams, robust decision-making, experiment selection, strategic bias control, and model-misspecification handling. Exports: - robust intervention-selection semantics - causal-model validation criteria - uncertainty and ambiguity treatment for causal decisions - strategic bias control - Goodhart-risk analysis - experiment-selection logic - decision-quality benchmarks - abstention and escalation rules for ill-specified problems - capability-status-tagged decision claims Imports: - from Thesis 1: evidence and outcome histories from improvement cycles - from Thesis 2: reasoning, memory, and hypothesis-generation capabilities - from Thesis 4: software tools and instrumentation for experiments and simulations - from Thesis 5: bounded utility, policy constraints, and alignment restrictions Primary agent/function home: - `CausalPredictionOrchestrator` - `CausalGraphBuilder` - `MechanismLibrarian` - `InterventionSimulator` - `StructuralEquationExecutor` - `CausalDiscoveryProcessor` - `CausalModelingAnalyzer` - `SelfPredictionModeler` - `PredictionCalibrator` - `SuperhumanExperienceMiner` (legacy name; not a capability claim) - `CounterfactualReasoningEngine` - `CounterfactualExperienceGenerator` - `CounterfactualAnalyser` - `StrategicCounterfactualPackManager` - `LongHorizonChainPredictor` - `HybridInferenceCoordinator` - `MechanismValidator` - `CausalBoundaryEnforcer` - `ForwardCausalPredictor` - `PredictionAuditor` - `PredictionReliabilityScorer` - `OptimalInterventionPlanner` - `ComputationalDecisionMakingOrchestrator` - `StrategicBiasMitigationOrchestrator` - `FramingInvarianceTester` - `OptionSetCompletenessAuditor` - `StrongestCounterPositionGenerator` - `FalsificationEvidencePlanner` - `SystematicExperimentDesigner` - `MonteCarloScenarioSimulator` - `TemporalHorizonIntegrator` Note: `SuperhumanExperienceMiner` is a legacy codebase name. Its name is not a claim that superhuman capability is implemented. Not owned here: - General executive planning is discussed in Thesis 2 unless explicitly tied to causal-decision semantics. - Acceptance gates for modifications are owned by Thesis 1 and Thesis 5. ## Thesis 4: Self-Modifying Software Substrate Primary responsibility: Model the executable substrate that lets Consullo generate, edit, repair, test, validate, document, and deploy agents and code. Exports: - agent/code generation mechanisms - automated repair pipeline - test and regression validation methods - semantic validation requirements - formal and statistical verification primitives - provenance requirements - LLM-Native Functional Java constraints - static-method, JSON-only, PDCA method contract - repair-pipeline self-improvement recurrence - capability-status-tagged software claims Imports: - from Thesis 1: improvement-loop acceptance semantics and method-memory learning - from Thesis 2: cognitive agents used by code generation and repair - from Thesis 3: decision logic for repair strategy selection and experiment design - from Thesis 5: trust gates, safety boundaries, security constraints, and deployment permission - from substrate context: LLM routing, atomic prompt routing, and token-cost accounting Primary agent/function home: - `SoftwareDeveloper` - `SystemsArchitect` - `DatabaseAdministrator` - `CloudServicesProvider` - `CybersecuritySpecialist` - `AgentBuilder` - `AgentEditor` - `AgentDocumenter` - `CodeProvenanceTracker` - `WorkflowAutomationCompiler` - `FormalProofConstructor` - `ReasoningChainVerifier` - `ConsistencyValidator` - `FaultLocalizer` - `ExecutionTraceAnalyzer` - `StaticAnalysisIntegrator` - `HistoricalFixRetriever` - `RepairKnowledgeGraphManager` - `ContrastiveTestGenerator` - `PatchCandidateGenerator` - `PatchValidator` - `PatchRanker` - `MultiHunkRepairCoordinator` - `RuntimeErrorDiagnoser` - `SecurityVulnerabilityFixer` - `RepairPromptComposer` - `RegressionTestValidator` - `RepairMetricsCollector` - `SemanticPatchValidator` - methodology metrics roles from the internal agent-programming methodology - atomic prompt and compiled-code orchestration roles from `atomic-prompts.md` Not owned here: - Whether a software change should be accepted into the system is ultimately governed by Thesis 1 and Thesis 5. - Causal choice of interventions is owned by Thesis 3. ## Thesis 5: Alignment And Scoped Trust Primary responsibility: Model alignment invariants, scoped trust, evidence ledgers, permissioning, containment, rollback, incident response, adversarial alignment, abundance obligation, and human authority under recursive modification. Exports: - alignment invariants - safety invariants - scoped-trust semantics - permission predicates - trust-estimate discipline, with trust-posterior models optional only when the vocabulary requirements are met - evidence-ledger requirements - Adversarial Alignment Function requirements - AAF non-veto requirement for high-stakes irreversible changes - Abundance Distribution Obligation requirements - human authority and escalation rules - containment, interruptability, rollback, and incident response constraints - single-owner Phase 1 governance frame - capability-status-tagged alignment and trust claims Imports: - from Thesis 1: candidate modifications and improvement-cycle evidence - from Thesis 2: cognitive and theory-of-mind capabilities used in alignment analysis - from Thesis 3: causal analysis of harms, incentives, and interventions - from Thesis 4: provenance, testing, deployment, rollback, and security instrumentation - from substrate context: internal economy and external-commerce context as scopes for trust evaluation Primary agent/function home: - `HumanValuesAligner` - `AlignmentDriftDetector` - `AgentAuditor` - `ConstitutionalAIAlignmentTrainer` - `SafetyBoundaryEnforcer` - `ConsensusCoordinator` (legacy name under single-owner Phase 1) - `AgentActionRecorder` - `InnovationAlignmentSupervisor` - `BiasAuditAgent` - `BiasDetectionAndMitigationCoordinator` - `EthicalEvolutionMonitor` - `ComputationalTrustOrchestrator` - `TrustStateManager` - `TrustEvidenceLedgerManager` - `TrustDimensionUpdater` - `TrustContextScoper` - `TrustUncertaintyCalibrator` - `TrustPolicyProjector` - `TrustRecoveryCoordinator` - `TrustBiasAuditor` - `CoalitionTrustAnomalyDetector` (legacy name under single-owner Phase 1) - `TrustImprovementSponsor` - `TrustworthinessOrchestrator` - `TrustworthinessGatekeeper` - `TrustworthinessIncidentCommander` - `TrustworthinessImprovementRequestBroker` - `ReliabilityImprovementSponsor` - `SafetyImprovementSponsor` - `SecurityImprovementSponsor` - `PrivacyImprovementSponsor` - `FairnessImprovementSponsor` - `Friendship` - `AdversarialAlignmentOrchestrator` - `AbundanceDistributionMonitor` - `BeliefModelingProcessor` - `IntentionRecognitionAnalyzer` - `PerspectiveTakingModeler` - `ProactiveContradictionHunter` - Adversarial Alignment Function roles - Abundance Distribution Mechanism roles - Friendship / constitutional ethical-anchor roles Notes under single-owner Phase 1: - `ConsensusCoordinator` means intra-organizational quorum or dispute resolution under owner authority, not independent democratic governance. - For `ConsensusCoordinator`, quorum means a typed review quorum over relevant internal evidence sources: policy checks, trust estimates, AAF dissent, technical validation, and human authority state. It is not a vote among independent stakeholders. - `CoalitionTrustAnomalyDetector` means detection of unhealthy internal dependency, routing, or influence clusters, not autonomous political factions. - `TrustBiasAuditor` means auditing distortions in scoped reliance decisions. - `ConstitutionalAIAlignmentTrainer` must specify critique-source diversity to avoid reinforcing alignment monoculture. - `ConstitutionalAIAlignmentTrainer` must use the AAF dissent-source enumeration as its critique-source distribution under single-owner Phase 1: rotating LLM personas, multi-model critique, theory-of-mind-driven stakeholder simulations, and external review where available. - `Friendship`, `AdversarialAlignmentOrchestrator`, and `AbundanceDistributionMonitor` operational contracts are specified in `appendix-thesis-5-operational-contracts.md`. - `AlignmentDriftDetector` monitors unintended deviation from current alignment constraints. `EthicalEvolutionMonitor` monitors intended value-interpretation adaptation under governance. - `BiasAuditAgent` detects bias patterns. `BiasDetectionAndMitigationCoordinator` owns mitigation workflow and follow-through. - Thesis 1 owns the generic improvement-sponsor pattern. Thesis 5 owns trust-property-specific sponsor variants such as reliability, safety, security, privacy, and fairness sponsors because their recommendations are derived from trustworthiness evidence. Not owned here: - `DigitalVirtualEconomyOrchestrator`, `BusinessFunctionOrchestrator`, and `ServiceProvidingEntitiesOrchestrator` are substrate-context systems, not Thesis 5 primary alignment mechanisms. ## Cross-Cutting Constraints ### Single-Owner Phase 1 All theses must treat single-owner Phase 1 as the baseline. Future contractor oversight and multi-stakeholder governance are later phases or stress-test scenarios. ### Bounded Compute All theses must account for token cost, compute cost, latency, validation cost, and opportunity cost. ### Capability Status Every major capability claim must identify whether it is implemented, specified, proposed, or speculative. ### Benchmark Discipline Benchmarks are evidence sources, not proof of intelligence or safety. Every benchmark claim must state task class and limitation. ### Evidence Ledger Discipline Evidence ledgers must be treated as audit-preserving records. Rollback annotates or supersedes evidence; it does not erase inconvenient history. The canonical evidence-ledger schema is maintained in `appendix-evidence-ledger-schema.md`. ### AAF Gate Discipline The Adversarial Alignment Function must be routed into Thesis 1 acceptance gates according to invariant I12 in `00-vocabulary-and-invariants.md`. ### Provenance Discipline Accepted modifications must satisfy invariant I11. ### Cost/Benefit Discipline Accepted non-emergency improvements must satisfy invariant I17. ### Deceptive Optimization Discipline Improvements to learned subsystems must satisfy invariant I19. ### Overclaim Control Sub-theses should use "scaffold for," "pathway to," "bounded," "measurable," and "governed" where appropriate. Avoid unqualified claims of superintelligence, solved alignment, or guaranteed improvement. ## Changelog - 0.3.2: Added organizational operating-layer row tying the AI-native R&D organization interpretation to the five-thesis import/export contract and execution-plan boundary. - 0.3.1: Added version header; annotated capability-flow and constraint-wrapper diagrams; added placeholder AAF/ADO/Friendship owner agents; canonicalized AAF gate through I12; added provenance, cost/benefit, and deceptive-optimization cross-cutting constraints; specified ConstitutionalAIAlignmentTrainer critique-source distribution; clarified ConsensusCoordinator, drift/evolution, bias/mitigation, and improvement-sponsor ownership. - 0.3.0: Revised after second review with capability-flow and constraint-wrapper diagrams.