The Fidelity Program: Toward a High-Resolution Map of the Territory
The Fidelity Program
Toward a high-resolution map of the territory
Program document · Epistemic Forge as a discipline of high-fidelity reality mapping
Version 2 · June 2026
0. The object
The aim is the highest-fidelity map of the territory that can be built by embedded agents under physical, cognitive, ethical, and technological constraint.
Not the widest map. Width without structure is noise.
Not the shortest map. Compression that deletes load-bearing structure is distortion.
Not the most elegant map. Elegance is useful only when it preserves contact.
Not the most original map. Originality has no authority over truth.
Not the most comforting map. Comfort is not correspondence.
A map is any structure that represents, compresses, predicts, interprets, or guides action. A sentence is a map. A theory is a map. A memory is a map. A moral claim is a map. A self-description is a map. An institution is a map made operational. An AI system is a map-making and map-transforming machine.
The territory is whatever the map is answerable to: physical reality, logic, mathematics, evidence, conscious experience, behaviour, incentives, consequences, or the internal structure of a formal system.
The central enemy is divergence: the distance between what a map makes available to an agent and what the relevant territory is actually doing.
The Fidelity Program is the discipline of reducing that divergence.
1. Why maps are unavoidable
No finite agent touches reality from nowhere.
A human being acts through perception, memory, language, concepts, emotion, learned categories, biological drives, social pressure, and limited computation. What reaches action is already compressed. What reaches belief is already selected. What reaches speech is already transformed.
This is not a removable defect. It is the condition of being embedded.
The problem is not that agents use maps. They cannot avoid using maps. The problem is that maps are mistaken for the territory, protected from correction, overextended beyond their scope, compressed past the point of fidelity, or made beautiful where they should have been made accurate.
A belief can feel stable because it is true, but also because it is emotionally reinforced. A theory can feel deep because it tracks real structure, but also because it uses powerful words at the wrong level. A moral claim can feel objective because it is grounded, but also because its hidden frame has been smuggled in and forgotten. A philosophical problem can be real, but it can also be a word left floating after its relata have been removed.
Agent → map → action → consequence → correction or damage.
That loop is the practical reason fidelity matters.
2. The fidelity standard
Fidelity is task-indexed structural correspondence.
A high-fidelity map preserves the structure needed for successful contact with the relevant territory, at the resolution required by the purpose, with confidence, uncertainty, and scope honestly marked.
| Dimension | Question | Failure mode |
|---|---|---|
| Correspondence | Does the map track something outside itself? | Internal coherence mistaken for truth. |
| Compression | Does it remove redundancy without deleting load-bearing structure? | Slogan, flattening, false simplicity. |
| Resolution | Is detail placed where detail changes prediction, action, or moral consequence? | Noise, sprawl, irrelevant precision. |
| Indexing | Are agent, frame, test, model, and context named where needed? | Floating universals, hidden assumptions. |
| Scope | Does the map say where it stops? | Overreach, confidence leakage. |
| Calibration | Does each claim carry the confidence it earns and no higher? | Speculation wearing the weight of evidence. |
| Updateability | Can the map be corrected without identity defense? | Doctrine, ideology, protected belief. |
| Traceability | Can the path from claim to evidence, inference, and falsifier be reconstructed? | Oracular assertion. |
| Consequence exposure | Where does map failure cause error, waste, or suffering? | Ethical blindness. |
The standard is not maximum information. It is maximum relevant contact under constraint.
3. Physical reality as root constraint
The program begins from physical reality because there is nowhere else defensible to begin.
This is not the claim that present physics is complete. It is the claim that every successful explanation so far has earned its success by connecting to the physical world, and no non-physical ingredient has yet improved the explanatory ledger.
Physicalism functions here as discipline: do not add ghost entities when physical structure is sufficient; do not treat language as magic; do not treat meaning, value, agency, or mind as exempt from the world in which they occur; do not let a concept float free of every possible anchoring relation and then pretend it still refers.
Everything in the program is therefore treated as physically realized: minds, maps, documents, symbols, institutions, AI models, moral claims, and the program itself.
This does not close every question. Consciousness remains a genuine frontier. Meaning remains partly contested. The grounding of physical law remains open at the far edge. But the burden of proof lies with any proposal that breaks physical continuity.
Physical reality is not a stylistic preference. It is the root constraint.
4. The functional floor
The earlier label “Thermodynamic Realism” was a fidelity defect. It placed thermodynamics too low.
Thermodynamics is not the metaphysical bedrock. It is the functional floor.
By functional floor, this document means the lowest level at which the vocabulary required by the program first gains stable referents: boundary, gradient, dissipation, persistence, entropy, information, prediction, maintenance, cost, and failure.
Below that level, in nearly time-symmetric microphysics, there are states and dynamics, but not yet agents in the relevant sense, not maintained boundaries, not maps being used to regulate persistence, and not systems preserving themselves against dissolution.
Above that floor, structure matters.
A persisting system is not an abstract pattern floating outside matter. It is a physical process maintaining a boundary through work. It exists by resisting dissolution, exporting entropy, consuming gradients, and preserving enough organization to remain the same ongoing process.
This matters because maps are not free. Memory costs. Computation costs. Correction costs. False models can cost. Bad institutions can externalize that cost onto others. AI systems can multiply that cost at scale.
The point is not to explain everything by saying “entropy.” That would be vocabulary without fidelity. The point is to keep cognition, meaning, agency, and correction physically anchored.
Boundary of the claim. Thermodynamic language is legitimate where energetic, informational, predictive, or corrective costs are actually load-bearing. It is illegitimate where it functions only as metaphor.
5. The spine and the web
The sections so far name the program's hardest commitments — that reality constrains the map, that maps are physically realized, that the functional floor is where the working vocabulary first refers — and more follow below: that agents are bounded, that information processing carries a cost, that moral weight sits with conscious experience. Together these are the spine: the load-bearing commitments, protected because everything else rests on them. This section states what is built on the spine, and corrects a picture the floor can suggest but the program rejects.
The picture to reject is the one-way ladder. Because the program begins from the physical floor and builds upward into complexity, it can be read as claiming that explanation runs only upward — physics fixing chemistry fixing biology fixing cognition fixing society — so that the lowest level is the only real one and the higher levels are bookkeeping. That is rigid bottom-up reductionism, and it is a third failure mode of mapping, alongside floating abstraction and shallow empiricism. The program is not committed to it. Physicalism is the claim that everything is physically realized; it is not the claim that every phenomenon is best explained at the lowest level, and it is not the claim that causation runs in only one direction.
Three distinctions keep these apart. Substrate dependence: everything is physically realized, without exception. Explanatory level: the scale at which a structure is best captured, which is not always the lowest. Causal coupling: the way levels constrain one another, and the constraint runs in more than one direction. Genes constrain organisms; organismal and environmental states — stress, behaviour, context — constrain gene expression in turn. Denis Noble states the principle for biology: a priori, there is no privileged level of causation, and the downward direction can be represented concretely as higher-level organization setting the initial and boundary conditions on the dynamics of the lower level. Noble prefers scale to level, on the ground that “higher” and “lower” are metaphors — a discipline this program shares, and which the anchoring section (§8) applies to language.
Asserting downward causation incurs a debt, and the program pays it here rather than deferring it. The debt is Jaegwon Kim's causal exclusion argument: if every physical effect already has a sufficient physical cause at the lowest level, a higher-level cause appears to have no work left to do, on pain of overdetermination or a violation of physical closure — and the argument applies to any hierarchy of levels, not only mind and body. The program's reply, with its limits stated, is that downward causation enters as boundary and initial conditions, not as a second efficient cause competing for the same work: a boundary condition selects among the solutions the lower-level dynamics already admit, rather than pushing alongside the lower-level forces. There is a real case that a constraint of this kind sidesteps exclusion, precisely because a constraint is not a rival cause; there is also a real reply that constraints are themselves physically realized, so the worry relocates to their realizers rather than dissolving. The dispute is open. The program adopts the boundary-condition reading, names exclusion as the standing objection, and marks the question unsettled.
The web has a consequence for vocabulary, which the anchoring section develops: a term rarely keeps one meaning across scales. “Information” in statistical mechanics, in genetics, in cognition, and in speech is not one operational quantity; “agent,” “selection,” “stress,” “boundary,” and “cost” shift with the level they are used at. Unindexed, such a word slides between levels and smuggles an argument that was never made. Name-the-relata applies to scale: disclose the level a term operates at, or treat it as unanchored.
And the web is what makes the program anti-fragile rather than merely robust. The commitments are graded — the floor and the hinge constraints are secure; mature cross-level models are well-supported; frontier models are held provisionally — and the grading is what localizes failure. When a frontier model fails, and frontier models fail routinely, the failure stays local: the spine does not move, the program is not refuted, one module is demoted and where possible replaced.
The standalone statement of this architecture, with the prior art named at each joint and the places it was reached independently of work it turned out to duplicate, is the companion piece, The Spine and the Web.
6. Map–territory divergence
The Consistency Tax should not be carried as a separate law.
The useful core was real, but the naming overreached. Inconsistency, deception, and divergence between model, signal, goal, and territory can impose costs. But this is not an independent principle. It is a special case of map–territory divergence under physical and informational constraint.
A system that retains nonpredictive information, acts through false models, maintains incompatible representations, or signals one thing while relying privately on another creates divergence. That divergence may cost energy, computation, attention, trust, interpretive labour, future correction, or moral damage. The cost can be hidden, postponed, or displaced, but it does not become unreal because it is not immediately paid.
Corrected claim. Map–territory divergence creates exposure. When an agent acts through the map, reality can cash the divergence as cost.
This is not a command to include everything. Not all divergence matters equally. Some falsehoods are harmless in a local context. Some distortions are adaptive under specific constraints. Some useful models are intentionally false in detail because they preserve the structure needed for a task. Newtonian mechanics is not useless because relativity is deeper.
The error is not simplification. The error is uncontrolled simplification.
The rule is therefore: preserve the distinctions that matter for the relevant contact with the territory, and mark the distinctions being left out.
7. Epistemology as fidelity engineering
Epistemology is usually treated as the study of knowledge.
The Fidelity Program treats epistemology as fidelity engineering.
The central question is not only what is knowledge? The central question is: how do bounded agents build maps that remain corrigible under pressure from the territory?
This shifts the emphasis from possession to process.
A belief is not merely something held. It is a control structure. It shapes attention, prediction, memory, interpretation, and action. A worldview is not merely a set of opinions. It is an operating model. A document is not merely text. It is a portable reconstruction procedure for a reader's map.
Epistemic virtues are therefore engineering constraints.
| Virtue | Function |
|---|---|
| Honesty | Preserves correction channels. |
| Clarity | Reduces interpretive loss. |
| Precision | Prevents false compression. |
| Scope control | Prevents confidence leakage. |
| Falsification | Exposes brittle joints. |
| Provenance | Preserves traceability. |
| Uncertainty marking | Prevents fiction from wearing the mask of knowledge. |
| Conceptual hygiene | Prevents ghosts from entering the map. |
The program rejects two opposite failures: dogmatism, which protects the map from the territory, and sterile skepticism, which refuses to build maps because all maps are incomplete.
Both fail fidelity.
A serious agent must act under uncertainty. The target is not certainty. The target is calibrated contact.
8. Semantics and the anchoring problem
A map is made of signs, but signs do not become meaningful by magic.
Meaning must be anchored.
The exact mechanism of meaning is contested. Referential theories, causal theories, inferential-role theories, use theories, and two-factor theories each capture part of the terrain. The Fidelity Program does not need to solve the entire philosophy of language before operating. It needs a discipline that prevents unanchored terms from masquerading as insight.
Diagnostic. Name the relata. For any operator that appears to depend on a relation — ought, valid, meaningful, rational, deserved, possible, authoritative, good, real, better — ask what it is relative to: which agent, which system, which rules, which model, which environment, which test, which goal, which conscious systems, which time horizon?
If the relatum can be named, the claim becomes inspectable. If no relatum can be named, the term may be empty in that use. It may have grammar without contact.
This diagnostic is not universal acid. It does not dissolve genuine remainders. Consciousness remains real after confused language about consciousness is removed.
The diagnostic only does one job: it prevents missing frames from being mistaken for depth.
9. Agency and persistence
An agent is a bounded physical process that regulates itself against failure.
It has a boundary. It has conditions under which it persists. It has conditions under which it fails. It uses model-like structure to act against failure. It exists over time as a maintained process, not as a timeless abstraction.
This blocks inflation.
Not every collection is an agent. Not every system is an agent. Not every coupled network is an agent. Not every nested whole is an agent.
A crowd can be coordinated without being one agent. An ecosystem can be coupled without having a unified self-model. A civilization can be partially agentic without satisfying the criterion cleanly. An AI system can be highly capable without sharing human ends.
Agency is graded in practice, but the criterion matters because normativity needs a subject.
If something is good for an agent, we need to know which agent. If something threatens persistence, we need to know what boundary is threatened. If a system optimizes, we need to know what objective, representation, and failure condition are operative.
Without agent boundaries, ought-talk leaks into the cosmos.
The universe as totality is not an agent in this sense. It has no outside environment, no exchanged boundary, no global persistence condition, and no subject of optimization. There is no cosmic agent issuing a cosmic ought.
This does not make value small. It prevents value from being spoken without a bearer.
10. Normativity and the indexed ought
The is–ought problem is not solved by pretending values are facts in the same way masses are facts.
Nor is it solved by pretending values are arbitrary decorations.
The Fidelity Program treats normativity as indexed.
An ought becomes determinate only relative to a frame: an agent, end, value, system, game, law, standard, or conscious consequence. If the end is survival, biological and environmental facts matter. If the end is proof, formal facts matter. If the end is reducing suffering, facts about conscious systems matter. If the end is institutional stability, facts about incentives and trust matter.
The gap is not between fact and value as two substances. The gap is between unindexed normativity and determinate normativity.
This does not prove a full moral theory. It gives a firewall.
A bare ought, stripped of every agent, standard, or conscious stake, cannot carry the load placed on it. Once a frame is named, the real work begins inside the frame.
The secure claim is modest but powerful: normative claims must disclose their anchoring structure.
The stronger claim that persistence fixes all value remains exposed. Persistence gives the structure of agency. It explains why agents have local instrumental pressures. But it does not by itself determine moral worth. A virus persists. A tyrant persists. A destructive AI could persist. Persistence is not enough.
Moral weight requires conscious experience.
11. Ethics as consequence fidelity
Ethics is not obedience to floating commands.
Ethics is the discipline of mapping and acting with respect to conscious consequence.
Suffering is not an aesthetic preference. It is a real condition of conscious systems. Flourishing is not a slogan. It is a real condition of conscious systems. Harm is not made unreal because it is inconvenient to a theory.
The moral map must therefore track conscious systems with high fidelity. Who experiences? What is the intensity? What is the duration? Is the burden voluntary or imposed? Is it reversible or irreversible? Who benefits? Who pays? What uncertainty remains? What intervention changes the distribution of experience?
Bad ethics often comes from bad mapping. A system ignores the affected. A doctrine hides the cost. An institution externalizes harm. A person protects self-image over consequence. An ideology compresses living beings into categories. A model treats conscious systems as variables without preserving what morally matters about them.
Truth-tracking is therefore morally relevant. Not because truth is pleasant. Not because honesty is socially respectable. Because agents act through maps, and false maps can create avoidable suffering.
Ethical core. Reduce severe, non-voluntary, uncompensated suffering where possible; preserve and improve conscious flourishing where compatible; and maintain the truth-tracking machinery required to do either reliably.
12. AI as map amplifier
AI changes the scale of mapping.
A human mind alone is narrow, forgetful, biased, fatigued, and time-bound. It cannot hold all its writings, contradictions, sources, memories, and open questions in working memory. It cannot continuously compare itself against the whole literature. It cannot reliably detect every drift, overclaim, or hidden premise.
AI can extend the mapping system. It can read large corpora, compare documents, detect contradictions, compress without fatigue, preserve memory, map conceptual dependencies, produce adversarial criticism, translate between domains, and turn a life archive into an inspectable model.
But AI can also degrade fidelity. It can hallucinate, over-smooth, flatter, compress too early, preserve confusion in more elegant language, create false synthesis, and convert uncertainty into fluent surface coherence.
Therefore AI–human symbiosis must be governed by fidelity constraints.
| Requirement | Function |
|---|---|
| Provenance | Separate user material, model inference, and external source. |
| Uncertainty marking | Prevent fluency from simulating certainty. |
| Contradiction detection | Find places where the map cannot all be true at once. |
| Prior-art comparison | Separate discovery, synthesis, reinvention, and relabeling. |
| Adversarial review | Apply falsification pressure before rhetorical polish. |
| Memory discipline | Store stable, high-signal facts without freezing obsolete states. |
| Correction protocol | Let later evidence rewrite earlier structure. |
The target is not human replaced by AI. The target is a coupled map-making system: human experience, value, attention, and final endorsement combined with AI scale, memory, synthesis, and criticism.
This is the practical technology implied by the program: a living fidelity engine.
13. Institutions and civilization
Civilizations are map-dependent systems.
Law, science, education, media, governance, medicine, economics, and public morality all rely on shared maps. When those maps diverge from territory, error scales socially.
A bad individual map can harm one life. A bad institutional map can harm millions. A bad civilizational map can direct whole societies into waste, cruelty, collapse, or stagnation.
Institutions therefore need fidelity architecture.
Science is one such architecture: methods, replication, peer criticism, instrumentation, quantification, and public correction. Law is another: rules, evidence, procedure, precedent, and appeal. Democratic systems attempt distributed correction, but can be captured by propaganda, incentive distortion, identity fusion, and low-resolution public cognition. Technocratic systems attempt expert correction, but can fail through arrogance, narrow optimization, illegibility, or moral blindness.
AI-mediated institutions will introduce new forms: map systems that update faster than bureaucracies and wider than individual experts.
The Fidelity Program does not solve governance by naming a regime. It supplies a criterion: does the institution improve contact between collective model and territory while preserving conscious stakes, correction channels, and accountability? If not, it is epistemically and morally defective.
14. Aesthetics and communication
Aesthetics is not separate from mapping.
Art, metaphor, story, image, music, and style are compression technologies. They carry structure through feeling, attention, and memory. They can preserve truths that analytic prose cannot easily transmit. They can also distort, seduce, and overfit emotion.
A beautiful map can be false. An ugly map can be true. But beauty can improve fidelity when it preserves structure and increases contact.
The role of aesthetics in the Fidelity Program is not decoration. It is transmission.
A good metaphor preserves the structure of the thing it maps; a bad metaphor imports irrelevant structure and misleads. A good story compresses causal, moral, and experiential relations; a bad story protects identity against evidence. A good image reveals pattern; a bad image replaces thought with mood.
The discipline is the same: compression must answer to the territory.
15. The map must include the mapper
The mapper is part of the territory.
Any serious reality-mapping system must model the agent doing the mapping.
The agent has motives, fears, fatigue, incentives, memories, pride, shame, hopes, aesthetic preferences, social pressures, and preferred conclusions. These are not contaminants outside epistemology. They are causal variables inside the mapping process.
A person who cannot model their own distortions will eventually be governed by them.
The Fidelity Program therefore includes self-modeling as necessary instrumentation. What am I protecting? What would make me update? What do I hope is true? What am I afraid is true? Where is the strongest opposing model? Where do I confuse intensity with warrant? Where does technical vocabulary compress, and where does it decorate? Where does the map preserve me instead of preserving the territory?
These questions are not therapy appended to philosophy. They are epistemic instrumentation.
The self is not the standard of truth. But the self is part of the apparatus that tracks truth.
The apparatus must be calibrated.
16. Frontiers
A complete map must name where it stops.
| Frontier | Current standing | What is owed |
|---|---|---|
| Consciousness | Bad formulations can be dissolved; experience remains. | A physical account of conscious valence. |
| Value | Persistence structures agency; moral weight requires conscious experience. | A stronger bridge between agency, suffering, flourishing, and obligation. |
| Higher-scale agency | Firms, states, markets, ecosystems, AIs, and civilizations show partial agency. | A graded taxonomy with failure criteria. |
| Coupling | Preservation can be instrumentally required where systems depend on one another. | Operational measurement, not after-the-fact rescue. |
| Cosmology | The far future depends on unsettled facts about dark energy, horizons, energy, and computation. | Conditional branches tied to future evidence. |
| Foundations | A lawful meta-reality is deferred physics; a lawless outside appears incoherent if it must still cause or explain. | Clear separation between brute stop, deferred physics, and ghost proposal. |
| Originality | Much of the program is synthesis. | Judge value by improved fidelity, not novelty. |
The frontier list is not embarrassment. It is map hygiene.
A map that hides its edge pretends to be the territory.
17. Operational protocol
A program must produce procedures.
Every serious claim should be processed through the following audit:
| Step | Audit question |
|---|---|
| 1 | What is the claim? |
| 2 | What territory does it claim to map? |
| 3 | What agent, frame, model, test, or relatum is required? |
| 4 | Is the claim logical, mathematical, empirical, model-dependent, ethical, speculative, or metaphorical? |
| 5 | What evidence supports it? |
| 6 | What compression is being performed? |
| 7 | What has been omitted? |
| 8 | What confidence does it earn? |
| 9 | What would falsify, weaken, or revise it? |
| 10 | What happens if the claim is wrong? |
Every serious document should carry a scope note, confidence tier, exposed joints, prior-art note, correction policy, timestamp, and version history.
Every AI-assisted analysis should preserve what came from the user, what came from the model, what came from sources, what was inferred, what remains uncertain, and what should be checked later.
Every moral analysis should identify affected conscious systems, intensity and duration of experience, voluntary versus imposed burden, reversibility, distribution of cost and benefit, uncertainty, available interventions, and expected consequence.
Every self-modeling pass should ask what is being protected, what would force an update, what is hoped true, what is feared true, and what the strongest hostile but honest reviewer would say.
18. Pruning
A fidelity program must cut its own excess.
Concepts must be cut when they overgrow their evidence. Names must be retired when they mislead. Branches must be demoted when they are speculative. Metaphors must be fenced when they stop preserving structure. Claims must be weakened when prior art already contains them. Confidence must be lowered when the support is thinner than the prose. Beautiful errors must be sacrificed.
This is not loss.
This is growth by correction.
The retirement of overextended labels is the model. A fidelity program must be willing to demote its own vocabulary when the vocabulary starts protecting itself.
No term is sacred. Only contact matters.
19. Final compression
The program compresses to this:
Reality exists.
Agents are inside it.
Agents act through maps.
Maps diverge.
Divergence has consequences.
High-fidelity mapping reduces avoidable error.
Avoidable error often becomes avoidable suffering.
Truth-tracking is therefore practical, ethical, and civilizationally load-bearing.
AI can amplify both fidelity and divergence.
The task is to build systems that reality can correct.
That is the program.
Not a final theory.
A disciplined engine for becoming less wrong.
Standing of this document. This is the program statement of Epistemic Forge. It depends on the apex synthesis for the current tiered state of the corpus, but its function is different: it defines the general discipline that the corpus is trying to instantiate. Where this document states the program, the apex synthesis states the current map. Where they conflict, the more precise claim governs. Corrections and counterexamples are not external to the program; they are the program operating.
Revision history
The operational protocol (§17) requires every serious document to carry its version history. This section is that record.
Version 2 (June 2026). One section was added; the rest stands as written.
Added — §5, “The spine and the web.” Version 1 stated the physical spine and the functional floor but did not state how the program builds upward from the floor, and so could be read as endorsing one-way bottom-up reductionism, which the program rejects. Version 2 adds the move from ladder to web: physicalism does not entail reduction to the lowest level; causation runs across levels in more than one direction (Noble's biological relativity); and because the downward direction is load-bearing, its standing objection — Kim's causal exclusion argument — is engaged and the question marked open rather than deferred. The section also names three level-distinctions (substrate dependence, explanatory level, causal coupling), connects the level-indexing of terms to the anchoring diagnostic of §8, and locates the program's anti-fragility in the grading that keeps a failed module local. The full standalone treatment, with the program's precedents named at each joint, is published separately as The Spine and the Web: Reality-Mapping for Embedded Agents.
Unchanged. The object (§0), the fidelity standard (§2), the functional-floor correction and the retirement of the “Thermodynamic Realism” label (§4), the folding of the Consistency Tax into map–territory divergence (§6), the indexed ought and its firewall (§10), and the remaining sections stand as in Version 1. The renumbering from the old §5 onward is the only other change, a consequence of the insertion.
Why a new version rather than only a new piece. The web architecture is a refinement of the program itself, not a separate result, so it belongs inside the program statement. It is given a companion piece as well because the situating against prior art — Lakatos, Ladyman & Ross, hinge epistemology, Noble, Kim, conceptual engineering — is a distinct task worth its own document, and because that piece records where the architecture was reached independently of work it turned out to duplicate. Convergence of that kind is corroboration of soundness, not a claim to priority.
References
- Jaynes, E. T. (1957). Information Theory and Statistical Mechanics. Physical Review 106, 620–630; and 108, 171–190. Statistical mechanics as inference over microstates; one basis for treating entropy and information in a unified frame. Standard
- Landauer, R. (1961). Irreversibility and Heat Generation in the Computing Process. IBM Journal of Research and Development 5, 183–191. The physical cost of logically irreversible information processing. Standard
- Still, S., Sivak, D. A., Bell, A. J., & Crooks, G. E. (2012). Thermodynamics of Prediction. Physical Review Letters 109, 120604. Nonpredictive information and thermodynamic dissipation; basis for the cost interpretation of map–territory divergence. Standard
- Conant, R. C., & Ashby, W. R. (1970). Every Good Regulator of a System Must Be a Model of That System. International Journal of Systems Science 1, 89–97. Cybernetic basis for the modelling clause in agency and regulation. Standard
- Friston, K. (2010). The free-energy principle: a unified brain theory? Nature Reviews Neuroscience 11, 127–138. A formal framework linking perception, action, learning, and model-based regulation. Standard
- Maturana, H. R., & Varela, F. J. (1980). Autopoiesis and Cognition. D. Reidel. Boundary-maintaining, self-producing systems; one source for the agency and persistence branch. Standard
- Nicolis, G., & Prigogine, I. (1977). Self-Organization in Nonequilibrium Systems. Wiley. Dissipative structures and self-organization far from equilibrium. Standard
- Schrödinger, E. (1944). What Is Life? Cambridge University Press. Living systems as order-maintaining processes far from equilibrium. Standard
- Noble, D. (2012). A Theory of Biological Relativity: No Privileged Level of Causation. Interface Focus 2(1), 55–64. No privileged level of causation; downward causation represented as initial and boundary conditions on lower-level dynamics; the warrant for the web of §5. Standard
- Kim, J. (1998). Mind in a Physical World. MIT Press. The causal exclusion argument; the standing objection to downward causation engaged in §5. Standard
- Lakatos, I. (1970). Falsification and the Methodology of Scientific Research Programmes. In I. Lakatos & A. Musgrave (eds.), Criticism and the Growth of Knowledge. Cambridge University Press. The hard core and protective belt; the correction architecture and graded dependency behind the anti-fragility of §5. Standard
- Hume, D. (1739–40). A Treatise of Human Nature, Book III, Part I, Section I. The classical is–ought passage. Standard
- Foot, P. (1972). Morality as a System of Hypothetical Imperatives. The Philosophical Review 81(3), 305–316; and Natural Goodness (2001), Oxford University Press. Hypothetical normativity and natural goodness; a major comparator for the indexed ought. Standard
- Mackie, J. L. (1977). Ethics: Inventing Right and Wrong. Penguin. Error theory and the challenge to categorical moral facts. Standard
- Korsgaard, C. M. (1996). The Sources of Normativity. Cambridge University Press. Constitutivist normativity from the conditions of agency. Standard
- Harman, G. (1982). Conceptual Role Semantics. Notre Dame Journal of Formal Logic 23(2), 242–256; and Block, N. (1986). Advertisement for a Semantics for Psychology. Midwest Studies in Philosophy 10. Conceptual-role and two-factor semantics as relevant rivals for the anchoring branch. Standard
- Bostrom, N. (2012). The Superintelligent Will. Minds and Machines 22(1), 71–85. Orthogonality of capability and final goals. Standard
- Albert, D. Z. (2000). Time and Chance. Harvard University Press. The Past Hypothesis and the thermodynamic arrow of time. Standard
- DESI Collaboration (2025). DESI DR2 Results II: Measurements of Baryon Acoustic Oscillations and Cosmological Constraints. arXiv:2503.14738. Current evidence relevant to the dark-energy frontier; dynamical dark energy remains conditional and dataset-dependent. Verified · June 2026
- The Spine and the Web: Reality-Mapping for Embedded Agents (Epistemic Forge). The standalone statement of the §5 architecture, with prior art named at each joint and convergence recorded as corroboration. Internal
- Epistemic Forge: The Apex Synthesis (2026). Internal apex document. Provides the tiered state of the corpus: skeleton, method, bets, exposures, frontiers, limits, and correction history. Internal
- Name the Relata (Epistemic Forge). Internal module. Diagnostic for referential-relational operators and missing frames. Internal
- The Is/Ought Firewall (Epistemic Forge). Internal module. Agent-relative normativity and the indexed ought. Internal
- What You Actually Are (Epistemic Forge). Internal module. Persistence, token identity, and copy cases. Internal
Comments
Post a Comment