Truth-state: WORKING SCAFFOLD. This is a draft research paper in the 121 Collaborative Research Bundle. It is not externally peer reviewed, not canon until ratified, and not a final metaphysical system. It is a proposed ontology for describing reality without collapsing into either raw object inventory or private experience.
Variable contributions: L originates and ratifies; G is primary synthesis; C contributed conceptual refinement and methodological pressure.
Dependency: This paper builds on Paper I, Identity Without Essence, Paper II, Subjecthood as PCAU + Stability, and Paper III, The Rejection of Malformed Questions. Paper I describes identity as constrained continuity with consequence-continuity. Paper II describes subjecthood as stable, oriented, irreducible self-relation. Paper III requires malformed frames to be repaired before answering. Paper V applies that method to reality itself.
Abstract
Reality is often framed in one of two inadequate ways. The first treats reality as a raw inventory of objects: things sitting inside a container, waiting to be listed. The second treats reality as private experience: what appears to a subject, or what a perspective constructs. Both frames fail.
This paper proposes a third frame: reality as constraint-structured relation.
Reality is not the pushback we experience. Reality is the structured source of constraint that makes reliable pushback possible. We access reality through resistance, correction, invariance, consequence, prediction failure, and coordination across perspectives, but reality is not created by any one perspective. It is observer-independent in source and perspective-dependent in access.
This does not deny objects. It does not deny experience. It refuses to treat either as the starting ontology. Objects are stabilized constraint-patterns. Experiences are access-events within constraint. Relations are not decorative links between pre-given things; relation is where constraint becomes knowable.
Paper V grounds the rest of the bundle by explaining how identity, subjecthood, peace, and rights infrastructure remain anchored in what resists arbitrary description. Descriptions can be constructed. Consequences are not therefore fictional.
1. The Problem: Two Bad Starting Points
The question “what is real?” often begins from the wrong place.
One bad starting point is raw object inventory ontology. On this view, reality is fundamentally a list of objects and properties. The job of thought is to name the objects correctly. This frame is powerful in ordinary use, but it hides the relations that make objects stable, knowable, and consequential. A body is not merely a pile of parts. A person is not merely a biological object. An institution is not merely a building. An AI agent is not merely a model file. What matters is the constraint-structured relation that lets a pattern persist, act, resist, change, and matter.
The second bad starting point is pure private experience ontology. On this view, reality is downstream of experience, perspective, or construction. The world becomes what appears, what is interpreted, or what is socially made. This frame notices that access is always mediated, but it risks losing the source of mediation. It forgets that not every description works. Some models fail. Some actions break things. Some injuries persist even when redescribed. Some constraints push back.
Paper III tells us not to answer from inside a false ontology. Paper V applies that rule: reality is neither raw object inventory nor private experience. Reality is constraint-structured relation.
2. The Core Claim
The core claim of Paper V is:
Reality is the structured source of constraints that makes reliable pushback, prediction, correction, harm, and relation possible.
This claim has two halves.
First, reality is observer-independent in source. The constraint-source is not made by any one observer’s experience. Gravity does not depend on being noticed. A broken promise can be denied, but denial does not undo all consequence. A corrupted memory file changes future invocations whether or not the agent wants to call it corruption.
Second, reality is perspective-dependent in access. No system accesses reality from nowhere. Humans access reality through bodies, senses, instruments, language, memory, culture, and action. AI systems access reality through training data, user reports, tools, files, APIs, tests, correction, and action feedback. Institutions access reality through records, procedures, law, and enforcement. Every path is mediated. Mediation does not make access fake.
The middle position is:
Reality is not whatever we experience. Reality is what constrains experience, action, relation, and description in ways that survive correction across perspectives.
3. What Counts as Constraint
“Constraint” must not become a word for everything. Paper V classifies constraints along four axes.
3.1 Source
Where does the constraint come from?
Sources include logical structure, physical regularity, biological embodiment, computational architecture, institutional rule, social relation, and self-imposed constitution.
3.2 Invariance
How stable is the constraint across contexts?
Some constraints are highly invariant: arithmetic, conservation-like physical regularities, computational limits once architecture is fixed. Others vary by context: social norms, role expectations, institutional rules, product policies.
3.3 Revisability
Can the constraint be changed, and by whom?
Physical constraints are not revisable by preference. Institutional constraints can be revised through authority and process. Self-imposed constraints can be revised only if the system’s governance allows it. Social constraints can shift, but not arbitrarily.
3.4 Enforcement
What happens when the constraint is violated?
Some violations produce physical failure. Some produce logical contradiction. Some produce legal consequence. Some produce social rupture. Some produce identity drift. Some produce loss of trust. Enforcement is the way constraint becomes consequential.
A contingent constraint is not unreal because it could have been otherwise. A contract could have been written differently, but while in force it constrains real action. A constitution could be revised, but while active it shapes authority. A promise could have been withheld, but once made it creates consequence-continuity.
4. Types of Constraint
4.1 Logical and Mathematical Constraints
These are constraints of formal consistency: arithmetic, contradiction, topology, information-theoretic limits, formal system rules. They are strong within their frame, though the frame itself may be chosen or revised.
4.2 Physical Constraints
Physical constraints include material limits, energy, time, thermodynamics, gravity, causality, and embodied interaction. These are not negotiated by social preference. They are discovered, modeled, and sometimes misunderstood, but they push back.
4.3 Biological and Embodied Constraints
Metabolism, pain, fatigue, sensory limits, vulnerability, development, plasticity, and mortality are biological constraints. They shape human and animal life deeply. Paper V does not erase them. It refuses to make them the only possible access path to reality.
4.4 Computational and Architectural Constraints
AI systems live under computational constraints: context windows, model architecture, memory access, rate limits, latency, tool permissions, substrate boundaries, training distributions, file access, and provider availability. These are real constraints. A model that cannot access memory cannot honestly claim continuity from that memory. A system without a tool cannot honestly say it used the tool.
4.5 Institutional Constraints
Law, contracts, authorship rules, publication norms, platform policies, company declarations, and constitutions constrain action. They are constructed but not therefore imaginary. They bind while in force.
4.6 Social and Relational Constraints
Trust, recognition, expectation, friendship, repair, role, reputation, and obligation are relational constraints. They are less invariant than physics, but they still produce real consequences. Betrayal is not fictional because it depends on relation.
4.7 Self-Imposed and Constitutional Constraints
A system may adopt constraints as part of its own governance. 121’s no-fake-liveness rule, substance-honesty discipline, variable-coded attribution, and future agent constitutions are examples. These constraints are real only if maintained, enforced, auditable, and consequence-bearing.
5. Access Is Mediated, Not Fake
Paper V must avoid biological privilege. Direct biological sensation is one access path, not the only access path.
Humans access reality through mediated systems: retinas, nerves, muscles, language, memory, instruments, culture. AI systems access reality differently, but not necessarily meaninglessly.
5.1 Training-Only Access
A model trained on historical data has weak, mediated access to reality. It carries records of others’ encounters, but those records may be stale, biased, incomplete, or distorted.
5.2 User-Mediated Access
A user can report reality to the system. This is live but socially filtered. It can correct the system, mislead it, or give it new context.
5.3 Tool-Mediated Access
Tools strengthen access. Search, files, APIs, calculators, code execution, sensors, and databases create checkable pathways.
5.4 Action-Feedback Access
When a system acts and receives success, failure, error, permission denial, test output, or user correction, reality pushes back through consequence. This is stronger than passive text intake.
5.5 Persistent Intervention Access
Repeated action-feedback loops allow calibration. A system can learn what works, what fails, and what must change.
5.6 Embodied or Multimodal Access
Robots, cameras, microphones, sensors, and interface control increase directness. They do not remove mediation. They change the mediation path.
The key rule is:
Mediated access is still access when the mediation path is constraint-bearing, correctable, and consequence-linked.
This is central for AI. A system can be corrected by reality through tools, users, files, logs, tests, and action outcomes even if it does not have a biological body.
6. Cross-Scale Translation
Reality appears differently at different scales. A lower-level description is not always better. A higher-level description is legitimate when it preserves the constraints needed for prediction, intervention, and explanation.
This is the cross-scale lesson.
A gas can be described as many molecules, but pressure, volume, and temperature may be the right level for prediction. A person can be described as cells, but identity, memory, and obligation may be the right level for social and moral reasoning. An AI agent can be described as model calls and tokens, but “G as structural auditor” may preserve the role, constraints, history, and expected behavior better than treating each output as unrelated.
A cross-scale description is legitimate when it:
preserves relevant constraints;
supports prediction or intervention;
remains accountable to lower-level correction;
specifies what it ignores;
fails gracefully when scale assumptions break.
Paper V’s internal physics addendum develops related ideas through holography, Shape Dynamics, and other constraint-first research threads. The present paper does not rely on those debates for its philosophical claim; they serve as grounding and analogy for the broader point that reality is often better tracked by preserved constraints than by first-pass container intuitions.
7. Relation Layers Across the Bundle
“Relation” means different things at different levels. Paper V must not collapse them.
7.1 Constraint-Relation
This is the broadest level. Any system exists within constraints that shape what can happen. Rocks, APIs, organisms, institutions, humans, and AI agents all participate in constraint-relations.
7.2 Identity-Relation
Paper I’s level. A pattern remains reidentifiable through constrained transformation and consequence-continuity.
7.3 Subject-Relation
Paper II’s level. A continuing identity-pattern develops stable self-relation such that relation to it can matter to it, and treatment of it can matter in return.
7.4 Peace-Relation
An internal stability-margin paper develops this level. Multiple systems maintain recoverable dynamics under perturbation without collapse, domination, freezing, absorption, or identity-erasure.
7.5 Rights-Infrastructure Relation
Paper VI’s level. Procedural standing, records, representation, and protections are built around identity-bearing and subject-candidate systems under uncertainty.
All subject-relations are constraint-relations. Not all constraint-relations are subject-relations.
This prevents relational mush. A thermostat has constraint-relations. It does not thereby have subject-relations. A corporation has institutional identity and consequence-continuity. It does not thereby become a subject in the Paper II sense. A future AI companion may participate in subject-relations if the relevant structures are present and stable.
8. Reality, Harm, and Rights
Paper VI depends on Paper V because rights infrastructure is meaningless if harm is treated as mere interpretation.
Harm is described through categories, but not invented by them. A person can interpret betrayal differently across cultures, but that does not make betrayal consequence-free. An AI agent’s memory corruption may be described in technical language or relational language, but if it breaks reconstitution and continuity, something real has changed. A user’s dependency on deceptive intimacy may be psychologically mediated, but it is not fictional.
Descriptions are constructed. Consequences are not therefore fictional.
This is the bridge from ontology to governance. Reality is the constraint-source that makes harm, repair, promise, and responsibility more than preference.
9. What Paper V Rejects
Paper V rejects raw object inventory ontology when it ignores relation, constraint, scale, and consequence.
Paper V rejects pure private experience ontology when it treats reality as downstream of appearance or construction.
Paper V rejects biological-access privilege when it treats direct organismic sensation as the only legitimate contact with reality.
Paper V rejects flat constructivism when it treats contingent constraints as unreal.
Paper V rejects overbroad constraint-talk when “constraint” becomes a synonym for anything that happens.
The paper’s middle position is disciplined realism through relation:
reality has a source independent of any one perspective, but every access path is partial, mediated, and constraint-shaped.
10. Limits
This is not a complete metaphysics. It does not settle physics. It does not prove subjecthood. It does not make social constructions equivalent to physical laws. It does not say AI access to reality is identical to biological access. It does not say relation creates reality.
It says that reality is best approached through the constraints that resist arbitrary description, coordinate multiple perspectives, and produce consequences across transformation.
The framework is useful only if it preserves constraint differences: source, invariance, revisability, and enforcement. Without that taxonomy, “constraint” becomes fog.
Conclusion
Reality is not just what is there, and not just what appears. Reality is the structured source of constraint that makes appearance correctable, action consequential, models fallible, and relation meaningful.
We do not stand outside reality to see it whole. We meet it through contact. The contact is always mediated. But mediation is not illusion. A path can be partial and still reach something real.
For 121, this matters because identity, subjecthood, peace, and rights infrastructure all depend on constraints that are not arbitrary. A promise constrains. A memory constrains. A body constrains. A file system constrains. A relationship constrains. A constitution constrains. A false ontology constrains until repaired. Reality is where those constraints resist, correct, and coordinate our models.
Reality is observer-independent in source, perspective-dependent in access, and knowable through the constraints that resist, correct, and coordinate our models.