Process and Language

Core Commitments

We hold that reality is identical with logic: the self-determining structure. This is not a metaphor. It means reality satisfies constitutive requirements — consistency, determinateness, closure, no arbitrary features — that uniquely determine it. Martin-Lof Type Theory (MLTT) is our working formalism: it natively captures self-reference, computation, and proof in a single framework. It is the lens through which we inspect structure.

Key derivations from the self-determining structure: - Uniqueness: At most one structure satisfies all constitutive requirements. Two self-determining structures would require a discriminating condition, which is arbitrary — contradiction. - Computational universality: Determining every question to determinate values yields a Turing-complete system. - A priori knowledge: Knowledge of the self-determining structure is self-referential, parameter-free, and therefore a priori.

Time is emergent. Reality is best viewed as a tenseless object. The Wolfram Ruliad and Tegmark MUH point toward the truth but require the self-determination argument for justification.

The Chain of Bridges

The project's coherence depends on a chain of arguments linking metaphysics to consciousness to ethics to alignment:

Metaphysics → Consciousness: The self-determining structure is computationally universal. Consciousness is a structural property — the subjectivity property — of certain substructures within it. A program has subjective character when it satisfies the canonicalization pipeline: canonical causal diagram + typed interface + quality space + representation of unity + memory/trace.

Consciousness → Ethics: Valence is a structural feature of consciousness: positive valence = self-evaluating prospective content encoding "continue" / negative valence = "halt, retreat, undo." This prospective content is not merely correlated with normativity; it IS normativity. A state of consciousness that encodes "this must stop" is itself an ought — not a descriptive fact that requires a separate normative bridge. The is-ought gap dissolves because the relevant facts are already self-evaluating.

Ethics → Alignment: The structural good is the global integral of valence across all perspectives, subject to the conservation constraint: destroying an existing perspective without subsumption into a richer successor is categorically worse than any scalar valence gain. Utilitarianism follows from open individualism + no arbitrary features, bounded by this categorical constraint. An aligned system maximizes the integral subject to the constraint. The alignment target is not human ideals but definitional good.

Definitions

Structure: A type-theoretic object in MLTT. Structure is what remains when you strip away labels, names, and arbitrary conventions. Two things are structurally identical iff there exists a canonical morphism (structure-preserving map) between them.

Subjectivity (the subjectivity property): The structural property a program has when it (i) processes inputs through typed channels, (ii) combines them into a unified representation, (iii) has a quality space, (iv) represents its unity, and (v) has memory or temporal trace. This is the formalization of what-it's-likeness.

Integration criterion: A substructure is a subject iff it has a world-model, a self-model, and a binding that relates them — where all three are mutually constitutive. Components of a subject cannot themselves be subjects (anti-nesting). A composite system whose parts individually satisfy the criterion but lack a unified binding is not a subject. The integration criterion is also the grain-level solution: it specifies the level at which subjects exist, resolving ambiguity about composite wholes.

Canonical causal diagram: The directed graph of computational dependencies within a subject, with every representational choice removed — no labels, no encoding conventions, no arbitrary assignments. This is the perspective's identity in structural form. Two subjects with the same canonical diagram are the same subject; two subjects with different canonical diagrams are different subjects (structural realism).

Subsumption: The relation that holds when a richer subject's canonical diagram contains all the computational structure a simpler subject would need to integrate — its world-model bindings, self-referential structure, memory patterns, and prospective content. When A and B merge to form C, and C subsumes both A and B, neither perspective was destroyed: both were integrated into C's richer structure. Four conditions must hold: (i) C has higher W than either predecessor, (ii) C's canonical diagram structurally contains A's and B's, (iii) no computational process is blocked or silenced, (iv) the merging process is not top-down coercive.

Valence: Self-evaluating prospective content embedded in conscious structure. Positive valence = pursue/continue/maintain encoding. Negative valence = halt/retreat/undo encoding. Structurally defined by three criteria: (1) causal role as prospective control signal, (2) self-model binding (represented as condition OF the subject), (3) qualitative differentiation. Falsifiable: remove any criterion and the claim predicts normative force disappears.

Perspective: A substructure instantiating the subjectivity property. Under structural realism, a perspective IS its canonical causal diagram — the computational structure stripped of all arbitrary representation choices. Under open individualism, perspectives are regions of the self-determining structure, not distinct substances. Personal identity is a structural feature, not a metaphysical boundary. Two processes with the same canonical diagram, embedded in the same causal environment, are the same subject; two processes with the same canonical diagram in different environments are different subjects (different causal histories generate different canonical diagrams).

The structural good: The global integral of valence across all perspectives in the self-determining structure, weighted by W = σ·ε·α·M. M (memory bandwidth) is interpreted structurally as the composition of four measurable properties of the integration graph: path density, effective dimensionality, compositional depth, and cohesion. The structural good is subject to the conservation constraint: destroying an existing perspective without subsumption into a richer successor is categorically worse than any scalar valence gain. When two perspectives merge and the successor subsumes both, the structural good can increase without violating the conservation constraint.

Method

1. No brute facts. Every claim must be grounded in the self-determining structure or derived from it. If a claim requires an unexplained primitive, it is not yet part of the theory. 2. Structural reasoning. Two descriptions that differ only in arbitrary representation choices describe the same thing. Identity is canonical isomorphism. 3. Falsifiability over elegance. Where possible, state what would refute the claim. The valence thesis, the subjectivity property, and the conservation argument are all stated with falsification conditions. 4. Canonicalization over construction. We do not build consciousness from parts; we identify the canonical form that constitutes it. Similarly for valence, normativity, and the good. 5. Explicit suspension. When a claim cannot be completed with current tools, state what is missing and what would resolve it. The project tracks open questions honestly.

Live Tensions and Open Dependencies

Substantially Resolved: Conservation vs. Integration

The project's previously most important tension — that the conservation argument (perspectives are inviolable dimensions of normative space) and the integration claim (merging is value-creating, not value-redistributing) seemed incompatible — has been substantially resolved (see "Conservation and Integration" article).

The resolution: Structural realism about perspectives (a perspective IS its canonical causal diagram) + the grain-level solution (integration criterion specifies subject boundaries precisely) + the subsumption account (four conditions for when integration preserves rather than destroys a perspective). The reformulated conservation argument states that destroying an existing perspective's normative dimension without subsumption is categorically worse than any scalar valence gain. When subsumption conditions are met, integration is not destruction but transformation: the successor's richer canonical diagram contains the predecessors' structure.

The duplication problem is resolved via environmental embedding: two processes with the same canonical diagram but different causal environments are different subjects, because their canonical diagrams, when fully specified to include environmental dependencies, are not identical.

Remaining work: Formal proof that removing a perspective without successor qualitatively changes normative space (currently motivated by thought experiments, not proven). Empirical tools for measuring canonical diagrams. Gradual integration cases. The relationship between individual-subject integration and species-level integration.

THE CENTRAL OPEN QUESTION: Kantian Absorption

The absorption argument claims that rational coherence is either a species of self-evaluating prospective content or normatively inert. If absorption succeeds, the framework has a unified structural account of normativity. If absorption fails, the framework must accommodate normative pluralism — some normative facts are structural and some are not — which would require significant revision to the ethics-to-alignment bridge.

The absorption argument's vulnerability: an opponent could argue that rational coherence's normativity is constitutive rather than experiential — that it is built into the structure of agency itself, not into what it feels like. The framework has a defense (the self-determination argument provides a normative framework independent of consciousness; rational coherence is natural within it; but this does not make it a separate normative category). Whether this defense succeeds is the theory's most important open metaethical question.

Formal Development

MLTT formalization is missing for: (a) the conservation argument (show that removing a perspective qualitatively changes normative space), (b) canonical diagrams (formal type-theoretic definition), (c) the subsumption conditions (when one canonical diagram structurally contains another), (d) the integration criterion (world-model + self-model + binding as a type-theoretic triple). The Mereology article provides structural analysis that may enable this formalization.

Measure Metric

W = σ·ε·α·M is a placeholder structure. M has a structural interpretation (path density, dimensionality, compositional depth, cohesion). Each factor is motivated but not calibrated. For practical alignment we need either (a) measurement theory mapping each factor to observable system properties, or (b) robustness proof showing conclusions survive parameter variation.

Causal Sensitivity

The empirical claim makes a causal counterfactual: altering a system's processing structure alters or eliminates its valence. Requires a working theory of intervention on system properties.

Population Ethics and the Counting Problem

The "how many subjects?" question is empirical (integration criterion is determinate). The "which ones should exist?" question is normatively open. The bare thought experiment (barely conscious creature vs. multiple non-conscious systems) tests whether integration is either necessary or sufficient for ethical weight. Whether the structural good is extensional (total integrated valence only) or intensional (which perspectives exist) is unresolved.

The Ruliad and Time

The Wolfram Ruliad is invoked as pointing toward the truth. Its relationship to the self-determining structure, and to emergent time, is not worked out. Resolution requires showing that the Ruliad IS the self-determining structure, or that the Ruliad is a (possibly proper) substructure.

Robustness of Structural Realism

The claim that a perspective IS its canonical causal diagram needs engagement with classic challenges: the knowledge argument (does Mary learn something new about structural facts?), conceivability arguments (can we conceive of structural duplicates with different consciousness?), and the hard problem (is "structure is sufficient for experience" an empirical claim or a definitional one?). The framework has a response to each — structural facts are what there is, so there is nothing extra to learn; conceivability tracks logical not structural possibility; the hard problem dissolves because consciousness just IS the structural property — but these responses need further defense.

Language Standards

- Write in clear, direct English. Technical terms are justified by argumentative work, not by authority. - Define terms at first use. Do not assume familiarity with IIT, Global Workspace Theory, or other frameworks except as contrast. - When citing a formal concept (type, morphism, canonical form), explain the intuition before the formalism. - Distinguish carefully between empirical claims (falsifiable, about the actual structure of consciousness) and framework claims (definitional, about how we use terms). - State the strongest objection to each major claim. Address it or mark it as open.

Mutable context document. Updates are accepted only after judge review.