Reference-Indexed Realism
got-v1-2a-1b | Canonical paper | Issue 1
https://goto-go.com/machine/documents/got-v1-2a-1b/

GOT-V1-2a-1b <https://goto-go.com/machine/documents/got-v1-2a-1b/>

Reference-Indexed Realism

RRP, Ontological Targets, and the Discipline of Closure

Abstract

Realism begins from an apparently simple commitment: that something exists, or is true, independently of our thought, perception, language, desire, or social agreement. It is the philosophical refusal to reduce the world to appearance, interpretation, construction, belief, fantasy, or wish. Against illusion, relativism, idealism, and social agreement treated as reality-making, realism insists that reality has authority. The real is what exceeds us, resists us, corrects us, and remains answerable to conditions beyond our preference.

Yet realism depends on a concept whose boundary is unstable. The real gains much of its force by contrast with the unreal: the fictional, illusory, merely possible, artificial, constructed, mistaken, imagined, or apparent. To call something real is usually to distinguish it from something that fails to be real in the relevant sense. The real therefore depends on the not-real as its contrast.

This creates a structural tension. Realism must exclude the unreal in order to preserve its force. But the unreal is not simply nothing. Illusions are real as experiences. Fictions are real as cultural objects and vehicles of disclosure. Possibilities are real as branch-space, risk, and counterfactual constraint. Absences are real in their consequences. Social constructions are real in their institutional force. Mathematical structures may be real without being physical objects. Moral claims may be real as normative constraints without being empirically observable things. Political myths may be unreal as description while remaining real as organising force.

The problem is not that realism is false. The problem is that unindexed realism is unstable. It asks whether something is real without specifying real as what, for what locus, through what access relation, at what scale, over what horizon, under what comparator, and with what residual. It then risks mistaking a finite model of an ontological target for complete ontology.

This paper proposes reference-indexed realism as the answer to that instability.

Reference-indexed realism holds that realism claims are finite models of ontological targets formed through the RRP scaffold: the physically anchored access architecture by which physically supported difference becomes operationally available as distinction for a locus. The index is not merely a semantic qualification placed upon a claim after the fact. It is the access condition under which difference can become available, classed, registered, retained, and possibly redeployed by a finite observer.

Here, substrate difference is a deliberately minimal term. It identifies target difference whose operational informational status for the evaluated locus has not yet been established. It does not deny that the target may already carry thermodynamic, statistical, physical, or other informational relations beyond the current access scaffold.

The target constrains what can be preserved. The observer constricts what can be accessed, classed, registered, retained, and redeployed. The RRP scaffold couples them. The model approximates the target without exhausting it. The residual prevents premature closure.

Reference-indexed realism does not dissolve the distinction between real and unreal. It disciplines it. It replaces the fantasy of complete ontology with corrigible modelling under constraint.

1. The instability of the real

1.1. Realism and the authority of reality

Realism begins from an apparently simple commitment.

Something exists, or is true, independently of our thought, perception, language, desire, preference, convention, or agreement. The world is not exhausted by what appears to us. The true is not identical with what we believe. The real is not produced by wanting, naming, imagining, or collectively endorsing it.

This is the force of realism.

Realism is the philosophical refusal to reduce reality to appearance, interpretation, construction, discourse, fantasy, belief, or wish. It insists that there is something to which our claims must answer. Reality resists us. It exceeds us. It corrects us. It can surprise us, injure us, outlast us, contradict us, and make our models fail.

This gives realism its critical power. Against illusion, it says that appearance may mislead. Against wishful thinking, it says that desire does not determine truth. Against relativism, it says that not all claims are equally answerable to the world. Against idealism, it says that existence is not reducible to mind. Against social construction taken too far, it says that collective agreement cannot make any state of affairs true merely by agreement. Against political mythology, it says that power can enforce a description without making it adequate.

In this sense, realism is a discipline of resistance.

But realism is not only a doctrine of independence. It is also a discipline of distinction. It must distinguish the real from what is not real, less real, falsely taken as real, real only in some derivative way, or not real in the relevant respect.

This distinction is where the difficulty begins.

For realism to have force, the real must not include everything without qualification. If every appearance, fantasy, mistake, fiction, projection, and possibility is real in exactly the same way as a body, wound, planet, organism, event, or physical constraint, then realism loses its contrast. It no longer tells us what must be resisted, corrected, or distinguished. It becomes a universal permission: everything is real somehow.

But if realism contracts too sharply, it excludes too much. It cannot explain the reality of illusions as experiences, fictions as cultural objects, absences as causes of action, possibilities as structuring constraints, institutions as social powers, values as obligations, models as knowledge-bearing abstractions, or myths as political forces.

The real must be distinguished from the unreal. Yet the unreal continually returns as something that matters.

Realism therefore begins with authority, but immediately confronts a boundary problem. The question is not simply whether reality exists independently of us. The question is how realism can preserve the authority of reality without collapsing into either naïve possession or indiscriminate inclusion.

That is the problem this paper addresses.

1.2. The real as contrast term

The word real rarely works alone.

It usually gains meaning by contrast. A real coin is not counterfeit. A real danger is not imagined. A real illness is not feigned. A real memory is not confabulated. A real friend is not merely performative. A real object is not hallucinated. A real event is not fictional. A real apology is not strategic. A real cause is not accidental. A real obligation is not decorative. A real pattern is not noise.

To call something real is often to deny that it is unreal in some relevant sense.

This does not make the real dependent on the unreal for its existence. But it does mean that realism depends on a boundary of contrast. The real must be separated from illusion, fiction, mistake, possibility, appearance, artificiality, construction, imagination, abstraction, or mere seeming.

The difficulty is that these excluded terms are not simple negations.

An illusion is not nothing. A mirage is not water, but it is a real appearance. A hallucinated voice is not an external speaker, but it is a real experience. A false memory is not a faithful record of an event, but it may be real as a retained mental state that shapes action. A fiction is not a historical occurrence, but it is a real text, object, performance, social reference, affective trigger, and sometimes a powerful disclosure of human relation. A legal fiction may not describe a natural object, but it can alter inheritance, liability, personhood, punishment, and power. A border may not be a natural feature of the earth, but it can determine whether a body is admitted, detained, expelled, protected, or killed.

The same applies to possibility. A possible danger is not an actual injury, but it may rationally alter behaviour. A possible future is not a present fact, but it can organise planning, investment, fear, hope, law, and collective action. An absence is not a present object, but a missing key, absent witness, lost parent, unpaid debt, unfilled role, or unrecorded harm may structure reality with great force. A mathematical object is not a physical body, but mathematical structure can govern proof, measurement, engineering, and theory. A moral wrong may not be a material object, but it may impose obligation, guilt, demand, repair, punishment, and reform.

The unreal therefore does not remain outside realism as nothing.

It returns as remainder.

The fiction returns as disclosure.

The illusion returns as experience.

The possibility returns as branch-space.

The absence returns as consequence.

The construction returns as power.

The abstraction returns as model.

The myth returns as political force.

The mistake returns as error condition.

Realism cannot ignore these returns. But each return pressures the boundary that realism needs.

If realism admits every such case as real without distinction, the concept of the real becomes too broad. If realism excludes them all, it fails to explain much of what structures human life, knowledge, perception, science, politics, law, mathematics, morality, and art.

The real is therefore not merely a settled domain. It is a boundary under pressure.

1.3. The return of the unreal

The return of the unreal appears wherever realism tries to secure its boundary.

The simplest version of realism might say: the real is what exists independently of us. Rocks, stars, bodies, cells, rivers, planets, forces, fields, organisms, and physical processes are real. They do not require our recognition in order to exist.

This is an important commitment. It blocks the collapse of reality into thought. But it does not settle the status of everything that matters.

Are laws real? Are patterns real? Are memories real? Are social roles real? Are obligations real? Are absences real? Are risks real? Are possibilities real? Are mathematical structures real? Are political myths real? Are fictional characters real as characters? Are institutional categories real where they govern life?

The realist cannot simply say no. Too much would disappear from explanation.

But the realist also cannot simply say yes without qualification. Too much would enter reality under the same name.

The instability lies in the pressure between exclusion and absorption.

If realism excludes the unreal too aggressively, it protects the boundary but impoverishes reality. It becomes unable to account for the very things that organise experience and action. If realism absorbs the unreal too generously, it explains those things but dilutes the boundary. Everything becomes real in some register, and the critical force of realism weakens.

This is not an accidental difficulty. It is structural. The real needs the unreal as contrast, but the unreal continually appears as mode, effect, condition, distortion, mediation, abstraction, potential, norm, institution, fiction, or residual of the real.

The problem, then, is not that realism fails because the unreal exists. The problem is that realism often lacks a disciplined way to say in what respect something is real, in what respect it is not, under what conditions the claim holds, at what scale the claim operates, and what remains outside the claim.

The question “is it real?” is too coarse.

A hallucinated voice is not real as external speaker, but it is real as auditory experience. A fictional revolution is not real as historical event, but it may be real as narrative form, political imagination, affective orientation, and social critique. A debt is not real as a natural object, but it is real as institutional obligation. A future danger is not real as present occurrence, but it may be real as risk. A mathematical theorem is not real as material object, but it may be real as formal necessity. A moral injury is not real as measurable mass, but it may be real as violation, demand, and relation.

The failure occurs when these distinctions are not indexed.

Realism becomes unstable when it demands a single answer where the phenomenon requires a disciplined plurality of respects.

1.4. The survey of realisms

The same tension appears across the major forms of realism.

Metaphysical realism gives the real its strongest form. It claims that there is a world independent of human minds. The world does not depend on our perception in order to exist. Stars formed before human observers. Rocks weather whether or not they are named. Bodies decay whether or not a theory describes them. This position preserves the authority of the real with maximum force.

But metaphysical realism immediately faces difficult cases. If reality is restricted to actual material existence, then laws, absences, possibilities, values, institutions, mathematical structures, and fictional objects become difficult to place. If they are admitted too easily, the boundary of the real expands until nearly everything becomes real in some way. The metaphysical realist therefore needs a way to distinguish modes of reality without collapsing them into one flat domain.1

Direct realism intensifies the issue at the level of perception. It claims that perception gives access to the world itself. When I see a table, I see the table, not a private mental representation detached from it. This preserves ordinary confidence in experience. But illusion and hallucination interrupt the claim. A mirage is not real as water, but it is real as appearance. A hallucinated voice is not real as external speaker, but it is real as experience. The direct realist must deny one reality while affirming another. The unreal enters perception not as nothing, but as a real misrelation to the world.

Critical realist traditions respond by separating an independently existing reality from historically produced and revisable knowledge of it. In Bhaskar's formulation, this distinction appears as the intransitive objects of knowledge and the transitive dimension of scientific knowledge. This makes critical realism more flexible than a simple identity between reality and present theory. Yet the same tension remains. The mediations through which reality is known - concepts, instruments, models, languages, institutions, calibration practices, and interpretive regimes - are themselves real features of inquiry. Critical realism must therefore preserve the independence of what is encountered while accounting for the reality and fallibility of the structures through which it is encountered.2

Scientific realism applies the problem to theory. It claims that successful scientific theories refer, at least approximately, to real entities, structures, or relations, including unobservable ones. Electrons, genes, black holes, gravitational waves, and fields may be real even when they are known only through instruments, models, and inference. The success of science would be difficult to explain if its theories did not track reality at all.

But scientific realism must also account for idealisations, approximations, discarded theories, useful fictions, simplified models, statistical constructs, and entities once treated as real but later abandoned. Science often reaches reality through models that are not literally complete. A frictionless plane, point mass, ideal gas, rational agent, isolated system, or perfectly random sample may be unreal as actual object while useful as model. The unreal model may disclose a real relation. Again, realism depends on what it cannot fully admit as literal and cannot simply exclude as false.

Moral realism relocates the issue from fact to value. It claims that moral truths are not merely personal preferences or social conventions. Cruelty, injustice, domination, betrayal, exploitation, or unnecessary suffering may be wrong whether or not a culture approves of them. This extends reality beyond physical description. But it raises a question: what kind of reality do moral facts possess? They are not material objects. They are not measured like distances. They do not appear as ordinary empirical entities. If moral truths are real, then reality includes normative force. If realism includes normative force, then reality already exceeds simple material actuality.

Mathematical realism pushes the problem toward abstraction. It claims that numbers, sets, structures, or mathematical truths exist independently of human invention. Mathematics is discovered rather than merely created. But mathematical objects do not occupy space and time in the ordinary sense. They do not act causally like stones or bodies. To call them real detaches reality from physical objecthood. Yet to deny their reality entirely seems to ignore the objectivity, necessity, and applicability of mathematical structure. Mathematical realism therefore needs a disciplined account of abstract reality.

Modal realism gives the tension a radical form. In Lewis's modal realism, possible worlds are treated as real worlds rather than merely as abstract descriptions of possibility. Other forms of realism about modality need not carry that full ontological commitment. The strong Lewisian version makes the point sharply: the merely possible is no longer simple non-being. Possibility becomes a domain of reality. This supports a robust treatment of counterfactuals and necessity, but it greatly expands what counts as real. Actuality becomes one mode among a plurality of worlds.3

Realism about social reality introduces dependence as a central problem. Money, borders, citizenship, gender categories, corporations, laws, offices, marriages, debts, property, and academic degrees depend on human practice and recognition. They are constructed, but they are not imaginary. They organise life, distribute power, create obligations, permit violence, enable exclusion, and shape identity. A debt is not a natural object, but it can govern a life. A border is historically produced, but it can decide whether someone moves freely or is detained. Realism about social reality therefore breaks the simple opposition between real and constructed.4

Realist fiction complicates the relation through invented representation. A realist novel may describe events that never happened and persons who never existed. It is unreal as historical record. Yet it may disclose real social conditions, psychological structures, historical pressures, moral patterns, and relations of power. Fiction can be false as event and still disclose a real relation. The invented character may expose a real pattern more sharply than a factual report. Realist fiction therefore shows that unreality may become a vehicle of reality.

Political realism uses realism in a strategic and practical sense. It insists that politics must be understood through power, interest, coercion, conflict, scarcity, and institutional constraint rather than moral idealisation alone. It tells us to get real about power. But politics is also sustained by myths, symbols, imagined communities, legal fictions, ideological narratives, legitimacy claims, and projected futures. A state is materially powerful, but it is also representationally sustained. Political reality is made of force and fiction together. The realist critique of illusion must therefore acknowledge that illusion itself can be politically real.

Across these cases, the same pattern recurs.

Realism secures the real by excluding the unreal. But the excluded term returns because it has consequence. Once realism accounts for that consequence, the real expands. If it expands without discipline, realism becomes inflated. If it contracts without discipline, realism becomes blind.

The instability is not intrinsic to realism as such. It is intrinsic to unindexed realism.

2. The missing ground: RRP as access scaffold

2.1. Why context is not enough

The answer to realism’s instability cannot be merely contextual.

It is tempting to say that reality-claims must always be qualified by context: real in this context, real from this perspective, real for this purpose, real under this description. That move is useful, but insufficient. If reference-indexed realism were only contextual qualification, it would remain a procedural caution applied after the fact. It would tell us how to speak carefully, but not why finite access takes this form in the first place.

Reference-indexed realism is stronger than that.

The index is not merely how a claim is described. It is how difference becomes available to a finite locus at all.

A finite observer does not receive the real as total state and then decide how to interpret it. It encounters reality through boundary, interface, channel, threshold, scale, abstraction, registration, retention, and possible redeployment. These are not optional additions to observation. They are the mechanical conditions under which anything can become available as an object, value, signal, measurement, perception, record, model, or claim.

This means that reference-indexed realism is not just a theory of epistemic humility. It is a physically grounded account of finite access.5

The real may exceed any observer. Substrate difference may exist before observation. But for a difference to become available to a locus as distinction, there must be an access scaffold. There must be a reference node from which the difference can matter. There must be an interface through which contrast can arrive. There must be an abstraction regime that can class it. There must be a registration event. There must be retention if a record is claimed. There must be redeployment if the record is to become operative in later inference, action, control, or correction.

Without this scaffold, the philosophical language of realism remains too coarse. It speaks of world and observer, object and subject, real and unreal, but leaves ungrounded the pathway through which world becomes available to observer as anything at all.

Reference-indexed realism therefore begins below the level of belief, judgement, theory, or interpretation. It begins at the access condition of operational distinction.

The index is not a footnote to realism.

The index is the bridge between substrate ontology and finite reality-claim.

2.2. Substrate difference before Observation

Realism must begin by preserving the independence of substrate difference.

There may be physical difference before observation, without observation, and beyond observation. A wavelength may exist outside a creature’s sensory range. A molecular relation may hold before any laboratory detects it. A pressure gradient may dissipate without being registered. A gravitational relation may structure motion without being conceptualised. A geological process may unfold before any witness. A harm may occur before any institution records it. A pattern may exist before any model captures it.

This is the realist ground.

Substrate difference names the physically real basis of the relation without closing its wider informational status. A state, distribution, correlation, constraint, or transformation may already be informational under a thermodynamic, statistical, physical, or other legitimate description. The narrower question here is whether any such difference has become operationally available as distinction for the finite locus under evaluation.

Where that relation has not been established, the paper withholds attribution to the locus. It does not infer that the target contains no information beyond the current access path.

Reality is not created by registration. Difference does not begin when observed. Magnitude, duration, energy, structure, constraint, relation, and lawful process may exist whether or not any finite locus can access them.

But substrate difference is not yet observed reality for a locus.

A physical difference may exist and never enter an access relation. It may never reach an interface. It may fall below threshold. It may be masked by noise. It may occur at a scale unavailable to the observer. It may dissipate before registration. It may remain unclassed. It may leave no retained record. It may be real and unavailable.

This distinction is decisive.

Substrate difference is not the same as established informational availability, observation, knowledge, perception, model, or phenomenological availability for a locus. The difference may be real and may stand in other informational relations. Its reality-claim for this finite locus becomes evaluable only where it enters a relation that can make it operationally available as distinction.

This avoids two errors.

The first error is constructivist collapse. If all available reality is mediated, one may conclude that reality is constructed by the observer. That does not follow. Mediation does not create the target. It supplies the access relation through which the target can become available.

The second error is naïve realism. If substrate difference exists independently, one may conclude that the observer can possess it as it is, without mediation. That also does not follow. Existence independent of observation is not the same as total availability to observation.

Reference-indexed realism preserves both claims.

Substrate difference may be real before observation.

Observation requires an access scaffold.

2.3. The RRP scaffold

The Relative Reference Point names the situated reference node from which distinction evaluation becomes possible for a locus.

It is not a merely human viewpoint. It is not subjective perspective. It is not arbitrary interpretation. It is not an analyst’s preference. It is the locus-interface-regime position at which some differences become available, some remain unavailable, some can be classed, some can be registered, some can be retained, and some can be redeployed.

The RRP scaffold can be compressed as a sequence.

RRP fixes the reference node.

OP supplies available contrast.

OA classes available contrast.

OR registers one classed alternative into state.

RAS retains the registered distinction.

RSD redeploys retained distinction into later transition, inference, action, control, or branch-realisation.

Each element is necessary.

Without RRP, there is no reference node for the informational or observational claim. A difference may exist, but it has not yet been specified for what locus, through what interface, at what scale, over what horizon, under what discrimination regime, or through what abstraction it can count as distinction.

Without OP, there is no available contrast field. The target may contain difference, but the difference has not reached the locus as available alternative.

Without OA, there is no classing. Available variation remains unpartitioned. It has not become this rather than that, signal rather than noise, within tolerance rather than outside tolerance, present rather than absent, relevant rather than inert.

Without OR, there is no registration event. One classed alternative has not become selected into state.

Without RAS, there is no retained observation record. The registration may have occurred transiently, but it has not persisted as an identifiable state over the relevant horizon.

Without RSD, the retained state has not become operative in later registration, inference, action, correction, or downstream branch-realisation.

This scaffold is the missing mechanical bridge between substrate ontology and phenomenological or model availability.

It explains how a world that exceeds the observer can nonetheless become available to the observer without being reduced to the observer. It also explains why that availability is always bounded, partial, scale-bound, abstraction-bound, and corrigible.

The target does not become real, structured, or informational merely because it enters RRP. It may already be real and may already stand in legitimate thermodynamic, statistical, physical, or other informational relations. What the RRP-structured relation establishes is that some target difference has become operationally available as distinction for a finite locus.

The bridge can be stated simply.

Substrate difference becomes available through OP.

Available contrast becomes classed through OA.

Classed difference becomes registered through OR.

Registered difference becomes retained as RAS.

Retained distinction becomes operative through RSD.

This is not merely epistemology. It is the mechanical grounding of finite access.

2.4. Observation as mechanical grounding

Observation is not a certificate.

Ordinary language often treats observation as if it automatically guarantees truth. Something has been observed, so it is treated as secured by reality itself. But observation does not eliminate mediation. It specifies a pathway of mediation.

Observation is bounded registration.

An observation event occurs where an OP-supported alternative is classed through OA and selected into state through OR. An observation record exists where that registration persists as RAS. Strong observation requires RSD, where the retained state is redeployed into later registration, inference, action, control, or downstream transition.

This means that observation is not raw reception of the world. It is not unmediated access. It is not possession of the target as total state. It is not a mental certificate. It is not simply the fact that something affected something else.

A perturbation may occur without observation. A signal may arrive without retained record. A persistent physical alteration may exist without being an observation record for the system under evaluation. A record may exist without grounded redeployability. A retained state may be available without becoming operative. A model may use a record while misclassifying its target relation.

Observation therefore grounds realism, but not by giving direct possession. It grounds realism by showing how substrate difference can become registered through bounded access.

A thermometer does not possess the full thermal state of a system. It registers a relation between thermal conditions and instrument state under material design, calibration, contact, scale, threshold, and readout conditions. A photograph does not possess the event it records. It registers light-mediated relation under lens, exposure, angle, sensor, surface, and retention conditions. A biological receptor does not possess the molecular environment. It registers only those molecular distinctions made available by binding architecture, concentration, boundary conditions, and pathway thresholds. A legal record does not contain the full social event. It registers a procedurally classed relation under evidence, admissibility, authority, category, and record conditions.

In each case, the observation is real enough to matter and bounded enough to require discipline.

The real is not first given whole and then interpreted. The real becomes observable through physically constrained registration pathways.

This is the mechanical grounding of reference-indexed realism.

Reality exceeds RRP.

Reality constrains RRP.

Reality becomes observable only through RRP-structured access.

2.5. The phenomenological as operationally available substrate relation

The RRP scaffold also reframes the phenomenological.

The phenomenological should not be treated as a free-floating interior realm detached from substrate ontology. Nor should it be dismissed as merely subjective appearance opposed to reality. What appears to a finite locus appears through an access relation. It is not the target whole, but it is also not nothing.

The phenomenological is substrate relation as operationally available to a finite locus.

A perceived object is not substrate ontology whole. It is target relation made available through sensory, motor, affective, attentional, and memory conditions. A measurement is not the target whole. It is target relation made available through instrument, scale, calibration, threshold, and record. A scientific datum is not the world itself. It is registered distinction produced through a disciplined observation program. A legal fact is not the full social event. It is a procedurally classed record under institutional conditions. A memory is not the past itself. It is retained relation in present accessible state. A model is not the ontological target. It is an abstraction that preserves selected relations under finite conditions.

This means that the phenomenological is neither pure invention nor complete reception.

It is the world as available under RRP constraints.

This is why reference-indexed realism is able to bridge substrate ontology and lived, measured, institutional, modelled, or conceptual reality. It does not place an unbridgeable gulf between the physical and the phenomenological. Nor does it collapse the phenomenological into total reality. It identifies the scaffold through which one becomes available as the other.

The phenomenological is not the opposite of the real.

It is the real as available through an RRP-structured access relation.

3. Ontological target, finite Observer, and model

3.1. Ontological target and model

Reference-indexed realism distinguishes ontological target from model.

The ontological target is whatever the realism claim concerns. It may be an object, process, relation, structure, value, event, institution, absence, possibility, law, pattern, system, body, norm, mathematical invariant, political order, or causal field.

The model is the finite abstraction through which that target becomes available, intelligible, comparable, usable, contestable, or correctable.

The model may be perceptual, scientific, mathematical, legal, moral, political, aesthetic, social, practical, computational, or linguistic. It may be simple or complex. It may be crude or powerful. It may be embodied in a sensory system, formal theory, measuring instrument, legal category, social practice, diagram, narrative, equation, dataset, memory, map, or institution.

But however strong the model becomes, it is still a model. It is not the target whole.

This is the decisive asymmetry.

The target constrains the model.

The model never exhausts the target.

The target constrains the model because the model must answer to something beyond itself. A relation either holds under intervention or it does not. A pattern survives scale translation or it does not. A measurement preserves a target distinction or it does not. A category tracks the relevant structure or it fails. A theory predicts, explains, and survives correction, or it breaks. A moral claim preserves a relation of harm, obligation, dignity, reciprocity, or justice, or it distorts it. A legal category captures a relevant social relation, or it misclassifies it. The target is not made true by the model.

But the model never exhausts the target because every model selects. It admits some differences and excludes others. It stabilises some relations and backgrounds others. It treats some variation as signal and some as noise. It operates at a scale. It holds over a horizon. It depends on an abstraction regime. It contains a residual.

This means that realism must give up the fantasy of complete ontology without giving up the authority of the real.

A finite model can be powerful without being total. It can be objective without being viewless. It can be constrained without being complete. It can preserve a real relation without containing the target whole.

This is why reference-indexed realism does not say that all models are equal. Some models preserve target relations better than others. Some have greater fidelity, broader range, stronger comparator support, lower distortion, clearer scale discipline, better correction pathways, and more explicit residuals. Others collapse under pressure.

The difference is not between model and no model. Finite access is always model-mediated. The difference is between disciplined and undisciplined models.

A realism claim is therefore not complete ontology.

It is an indexed model of an ontological target under constraint.

3.2. The finite Observer as modelling constraint

The distinction between target and model must not be read as if the finite observer were absent from the modelling relation.

In this paper, observer is being used in the broad realism sense: a finite modelling locus through which an ontological target becomes available as model, claim, record, category, judgement, measurement, perception, or practice. This is not yet the capitalised technical status introduced in the observation paper, where an Observer is realised NTO: retained record actually redeployed as a controlling discriminator under comparator discipline. The present claim is more general. Every finite observer constricts access because every finite observer encounters the target only through boundary, interface, scale, horizon, abstraction, registration, retention, comparator pathway, and residual.

The observer is not a transparent aperture through which the target appears. Nor is the observer an immaterial point outside the world. The observer is part of the access relation by which the target becomes available under finite conditions.

This means that the model is constrained from two sides.

The target constrains the model by what it is. A contrast either reaches an interface or it does not. A structure either survives intervention or it does not. A pattern either remains stable across scale or it collapses. A claim either preserves the relevant relation or fails under comparator pressure. The target is not invented by the observer, and the model cannot make any target relation true merely by imposing a frame upon it.

But the observer also constricts the model by what it is.

Its substrate determines what can count as available, salient, registerable, retainable, comparable, actionable, and correctable. A biological observer is constrained by sensory range, metabolism, attention, affect, fatigue, memory, pain, habit, and viability. A scientific observer is constrained by instruments, calibration, units, sampling practices, statistical thresholds, formal variables, and disciplinary inheritance. A legal observer is constrained by categories of personhood, evidence, admissibility, harm, responsibility, remedy, and authority. A computational observer is constrained by tokenisation, architecture, training distribution, context limits, memory, optimisation history, interface, and deployment conditions. A political observer is constrained by institutional position, ideological inheritance, threat perception, available records, strategic incentives, and power relations.

The observer is not merely looking at the target. The observer is part of the access condition through which the target becomes modelled.

This is why the observer cannot be totally split from the observed.

Once a target is observed, measured, modelled, classified, remembered, represented, narrated, or acted upon, the resulting model bears the marks of both sides of the relation. It carries constraint from the target and constriction from the observer. The observed is not reducible to the observer, but neither is it available to the observer apart from the observer’s conditions of access.

The RRP scaffold names the coupling between them.

The target supplies the differences that may become available. The observer supplies the locus, interface, threshold, scale, abstraction, registration, retention, and redeployment capacities through which some differences can become operative. The model arises where target constraint and observer constriction meet.

Reference-indexed realism therefore rejects the spectator model of knowledge. The observer does not stand outside ontology and look in. The observer is an ontological participant in the access relation. Its substrate is part of the reason the model has the form it has.

This does not make realism subjective. It makes realism more disciplined.

If the observer contributes constriction, then objectivity cannot mean removal of the observer. It must mean audit of the observer’s contribution: the boundary it imposes, the scale it fixes, the abstraction it uses, the salience it inherits, the channels it permits, the categories it stabilises, and the residuals it produces.

The target constrains the model.

The observer constricts the model.

The RRP scaffold couples them.

The model never exhausts the target.

The observer never stands outside the relation.

A realism claim becomes stronger when both sides are explicit. It must ask not only what target is being modelled, but also what kind of observer could model it in this way, through what substrate, at what scale, with what omissions, and under what correction discipline.

3.3. The model as approximation

Total knowledge of the ontological target is unavailable to finite observers.

This is not a sceptical claim. It is a structural claim. A finite observer cannot receive the target as total state. It must encounter selected relations through boundary, interface, channel, threshold, scale, abstraction, registration, retention, and possible redeployment. Every model is therefore approximate.

Approximation is not failure. It is the normal form of finite realism.

A model may approximate a target weakly or strongly. It may preserve only a rough relation, or it may preserve a relation with extraordinary precision. It may operate only locally, or it may travel across wide domains. It may be fragile under intervention, or it may remain stable under severe comparator pressure. It may hide its residuals, or it may track them explicitly. It may be corrected and strengthened over time.

Approximation therefore admits degrees of strength.

A model is stronger when it preserves target relations with high fidelity. It is stronger when it survives intervention. It is stronger when its scale is explicit. It is stronger when its horizon is known. It is stronger when its abstraction regime is clear. It is stronger when it admits the observer’s constriction. It is stronger when it declares residuals. It is stronger when it can be revised without collapsing into dogma.

A model is weaker when it hides the access relation. It is weaker when it mistakes its scale for all scales. It is weaker when it treats the observer’s limitations as features of the target. It is weaker when it converts residual into irrelevance. It is weaker when it cannot specify what would defeat it. It is weaker when it acts as complete ontology.

Approximation is therefore compatible with objectivity. Indeed, objectivity is possible only when approximation is made explicit enough to be corrected.

The alternative to approximation is not total access. The alternative is unacknowledged approximation pretending to be total.

Reference-indexed realism makes approximation explicit.

3.4. The error of complete ontology

The central failure mode of unindexed realism is complete ontology.

Complete ontology occurs when a finite model of a target is treated as if it were identical with the target itself. The record is mistaken for the thing. The measurement is mistaken for the full relation. The category is mistaken for the natural structure. The theory is mistaken for the world. The institution is mistaken for justice. The legal verdict is mistaken for the whole event. The modelled risk is mistaken for the future itself. The fictional form is dismissed as mere unreality. The abstraction is taken as reality without remainder.

This error is not merely theoretical. It has practical consequences.

When a model is treated as complete ontology, its omissions become invisible. What it cannot register appears not to exist. At the informational boundary, what the locus cannot access may likewise be mistaken for what the target does not contain. What it misclassifies appears deviant or exceptional. What lies outside its scale appears irrelevant. What exceeds its categories appears irrational, fictional, subjective, pathological, impossible, or confused. What it cannot measure appears unreal. What it cannot legally admit appears uninjured. What it cannot economically price appears valueless. What it cannot formally express appears meaningless.

Error is then protected by closure.

A model may begin as a useful approximation. It may preserve a real relation within a valid domain. But once it is treated as complete ontology, its domain limits disappear. The model is exported beyond its scale. Its abstraction becomes authoritative after the relation it preserved has changed. Its exclusions harden into ontology. Its residuals are reclassified as nothing.

This is one reason the unreal returns.

It returns wherever a closure claim has exceeded its warrant. The excluded remainder reappears as anomaly, resistance, suffering, misfit, exception, breakdown, contradiction, revolt, pathology, uncertainty, or new evidence. The realist then faces a choice: revise the model or defend the closure.

Reference-indexed realism chooses revision where correction is required. It treats closure as necessary but never absolute. Models must close enough to function, but not so completely that residual disappears from view.

The target exceeds the model. The observer constricts the model. The residual marks the excess and constriction that closure has not resolved.

Complete ontology denies this structure.

It mistakes finite access for total possession.

3.5. Closure, residual, and error space

Finite systems must close.

They must classify, measure, decide, act, judge, model, predict, remember, narrate, legislate, diagnose, and intervene. No finite observer can hold reality open in total complexity before every action. A scientist must define variables. A court must reach verdicts. A doctor must diagnose. A person must act under uncertainty. A model must simplify. A measurement must choose a scale. A theory must state what counts as evidence. A political order must classify persons, rights, obligations, and harms.

Closure is therefore not a defect by itself.

The problem is unindexed closure.

A legitimate closure claim says: under these specified conditions, this model preserves this target relation strongly enough for this purpose. It states the locus, interface, scale, horizon, abstraction, registration mode, comparator, and residual. It remains open to correction if the relation fails, if the observer’s constriction becomes consequential, if the target exceeds the model, or if the residual becomes operative.

An illegitimate closure claim says, explicitly or silently: nothing relevant remains outside the model.

This is premature closure.

Premature closure creates error space because it hides uncertainty inside verdict. It converts approximation into ontology. It converts residual into irrelevance. It converts scale-bound success into universal success. It converts observer-constriction into target structure. It converts a local model into a complete world.

Residual is the discipline that prevents this.

Residual names what remains unresolved, compressed, excluded, omitted, under-specified, unregistered, mis-scaled, untested, or outside the model’s validated range. It is not a failure to think. It is the mark of finite modelling. It keeps the model from pretending to be complete ontology.

In the present context, residual may therefore include target structure, correlation, or information that the current scaffold does not preserve or make operationally available. Its absence from the model is not evidence of absence from the target.

Residual is not the same as ignorance in general. It is localised remainder inside a specific model, claim, scale, horizon, access relation, comparator regime, or closure. It asks: what is not yet captured here? What would change the claim? What scale has not been tested? What access relation is missing? What comparator is unavailable? What observer-constriction may be shaping the result? What target relation may have been compressed beyond warrant?

This residual discipline is not yet the full discipline of the unknown. It does not provide the later procedure for uncertainty, deferral, update conditions, or unknown-handling. That broader burden belongs to Part 3. The present paper makes the narrower realism claim: no finite model should mistake its closure for complete ontology, and no realism claim should hide its unresolved remainder inside the authority of the real.

The discipline of residual allows realism to remain strong without becoming dogmatic. It permits closure while retaining corrigibility.

Realism does not need complete ontology. It needs disciplined closure.

4. Reference-indexed realism

4.1. The reference-indexed turn

Reference-indexed realism is the view that reality is not created by the observer, but every finite reality-claim is formed through an RRP-structured access relation in which physically supported difference becomes operationally available as distinction for a locus. The relation indexes what the claim can attribute; it does not determine the total informational or ontological content of the target.

This definition matters.

Reference-indexed realism is not merely the claim that reality depends on context. It is not the claim that observers make their own worlds without constraint. It is not the claim that every description is equally valid. It is not the claim that realism should be replaced by perspectival pluralism.

It is a physically grounded realism of finite access.

Substrate difference may be real before observation and may carry informational structure beyond the current access relation. A finite reality-claim becomes evaluable only where some target relation becomes available to a locus through an access scaffold. The RRP scaffold specifies this relation. It states how difference becomes available as contrast, classed as alternative, registered as state, retained as record, and redeployed into further action, inference, or correction.

The index does not make something real.

The index makes the realism claim evaluable.6

Without an index, the claim is under-specified. A difference may be real, but for what locus is it available? Through what interface? At what scale? Under what abstraction? Registered by what event? Retained as what state? Redeployable through what pathway? Correctable under what comparator? Leaving what residual?

Reference-indexed realism therefore preserves the realist commitment while rejecting unbounded possession.

The target constrains the model. The observer constricts the model. The RRP scaffold supplies access. The model approximates the target. The residual prevents premature closure.

This position avoids the two errors between which realism often oscillates.

Against naïve realism, it denies that finite observers possess reality as total state. No observation, measurement, theory, category, or model is the real whole.

Against empty constructivism, it denies that reality is simply made by the observer. Not every model works. Not every abstraction preserves the target. Not every scale is valid. Not every category survives correction. Not every perspective carries the relation it claims to carry.

The relation is relative.

The constraint is real.

4.2. The reference index

The reference index is the minimal discipline required to make a realism claim evaluable.

It asks a series of questions.

What is the locus?

For what observer, system, organism, institution, practice, theory, instrument, community, model, or regime is the claim being made? A distinction available to one locus may be unavailable to another. A record meaningful to one institution may be inert outside it. A signal available to one organism may not exist as signal for another. A mathematical structure may be operative within a formal regime and irrelevant in another context.

What is the interface?

Through what access relation does the target become available? The interface may be sensory, instrumental, linguistic, mathematical, institutional, computational, social, practical, or embodied. It determines what can reach the observer and how the target relation is transformed in becoming available.

What is the scale?

At what level of comparability is the claim being made? A molecular claim may not settle an organismic claim. A population claim may not settle an individual case. A legal category may not exhaust lived harm. A mathematical abstraction may preserve structure while omitting material implementation. A political claim may hold at institutional scale and fail at personal scale.

What is the horizon?

Over what temporal, operational, historical, developmental, modal, or practical range does the claim hold? A model may be strong over a short horizon and weak over a long one. A classification may hold under ordinary conditions and fail under stress. A political arrangement may appear stable until the horizon expands to include suppressed residuals.

What is OP?

What contrast becomes available? What part of the substrate target enters the access relation as available difference? What remains unavailable to this locus—below threshold, masked by noise, too diffuse, too fine-grained, too distant, too degraded, or outside the operative interface? Unavailability within this relation limits attribution to the locus; it does not establish that the target lacks structure or information beyond the access path.

What is OA?

What abstraction or classing regime makes the available contrast registerable? What counts as same or different, signal or noise, present or absent, relevant or irrelevant, admissible or inadmissible, valid or invalid, within tolerance or outside tolerance?

What is OR?

What registration event selects one classed alternative into state? Where does the model’s claim pass from possible contrast to actual registered difference?

What is RAS?

What is retained? Does the registration persist as a re-identifiable state over the relevant horizon? Is there a record, memory, inscription, measurement, legal entry, perceptual state, data point, or other retained abstraction state?

What is RSD?

Can the retained distinction be redeployed? Does it enter later inference, action, control, correction, explanation, classification, or branch-realisation? If strong objectivity is claimed, does downstream dependence actually turn on the retained distinction?

What is the comparator?

What would show that the model fails, weakens, misclassifies, overextends, or requires revision? A claim without a comparator cannot easily be corrected. It may be asserted, but not disciplined.

What is the residual?

What does the model omit, compress, exclude, defer, mis-scale, under-specify, or leave unresolved? What remains outside closure? What would need to be tracked if the claim were extended?

These questions do not make realism cumbersome. They make it answerable.

Ordinary claims may not need all dimensions explicitly stated. But serious philosophical, scientific, moral, legal, political, and ontological claims do. The stronger the claim, the stronger the index required.

The index is not an ornament added to realism. It is the condition under which finite realism becomes responsible.

4.3. Objectivity as RRP-disciplined correction

Objectivity is often imagined as the removal of the observer.

The objective claim is supposed to be what remains when perspective disappears, when mediation is stripped away, when the world is seen as it is from nowhere.7

Reference-indexed realism rejects this image.

The observer cannot be removed from finite access. The observer’s substrate constricts the model. Its body, instrument, language, scale, category, memory, institution, method, and history all shape what can be registered. The model bears these marks.

Objectivity cannot therefore mean unmediated possession.

Nor can it mean consensus alone. A community may agree because it shares the same constricting abstraction. An institution may stabilise a category that misclassifies. A discipline may inherit a model beyond its valid scale. A political order may enforce an official reality. Agreement can strengthen objectivity only where correction under constraint remains possible.

Objectivity is disciplined correction of the model under pressure from the target.8

A claim becomes more objective when its reference index is explicit enough to be tested, compared, corrected, narrowed, strengthened, or rejected. The locus is named. The interface is understood. The scale is specified. The horizon is bounded. The abstraction is visible. The registration pathway is inspectable. The provenance of the retained state is explicit. The redeployment condition is specified. The comparator is strong. The residual is declared. The correction pathway remains open.

Objectivity is not the disappearance of the observer. It is audit of the observer’s contribution.

What did the observer’s substrate make visible?

What did it make invisible?

What did it treat as signal?

What did it treat as noise?

What scale did it fix?

What abstraction did it inherit?

What did its categories admit?

What did they exclude?

What target relation survived comparator pressure?

What residual remains?

A model becomes more objective not by pretending to be model-free, but by becoming answerable to the target and to its own conditions of formation.

This is why approximation is not a failure of realism.

Approximation is the normal form of finite realism. A model may approximate a target with great strength. It may preserve relations across many contexts. It may predict, explain, intervene, correct, and survive challenge. But it remains approximate because it is finite, indexed, and constrained by both target and observer.

The alternative to approximation is not complete ontology. It is unacknowledged approximation pretending to be complete.

Reference-indexed realism makes approximation explicit.

4.4. Scale, abstraction, and approximation

Every finite observation is abstraction-bound.

A locus does not receive total reality. It selects, partitions, compresses, thresholds, stabilises, and retains. It admits some differences and excludes others. It treats some variation as signal and some as noise. It establishes equivalence relations. It makes selected relation usable by sacrificing totality.

This elementary abstraction operation can be called a cut.

A cut makes difference registerable. It is not an accidental defect in observation. It is the condition under which finite observation becomes possible. Without a cut, there is no stable this rather than that, no same rather than different, no signal rather than noise, no object rather than background, no evidence rather than irrelevance, no record rather than undifferentiated state.

But every cut opens an abstraction gap.

The abstraction gap is the discrepancy between the model, registration, word, measurement, category, or perspective and the fuller relation it compresses. The gap is not identical with error. It is the structural space in which error can arise. Error occurs when what has been omitted, compressed, misclassified, misweighted, over-stabilised, under-specified, or placed outside the frame becomes consequential for the claim being made.

Every finite observation is also scale-bound.

A claim must operate at some level of comparability. Scale is not merely physical size. It is the operative relational level at which a perspective stabilises differences into comparison. A molecular relation, cellular process, organismic behaviour, personal identity, legal category, institutional pattern, market price, political order, or mathematical model each becomes available at a scale where differences can be individuated, compared, registered, retained, and corrected.

Scale opens scale gap.

Scale gap occurs where a relation is registered, interpreted, measured, modelled, explained, or acted upon at a scale different from the one at which its relevant constraints are stable. A population relation may be forced into an individual case. A molecular fact may be treated as a complete organismic explanation. A legal category may be treated as if it exhausted lived harm. A market value may be treated as if it exhausted human value. A model variable may be exported beyond the scale at which it preserves the target relation.

Abstraction and scale make approximation unavoidable.

No finite model receives total reality. It receives a target relation through cut and scale. The question is not whether abstraction and scale are present. They always are. The question is whether they preserve the relation being claimed.

Reference-indexed realism therefore treats objectivity as correction of abstraction and scale under target constraint.

A model is not objective because it escapes cuts. It is objective to the degree that its cuts are explicit, its scale is disciplined, its comparator is strong, its residual is visible, and its correction pathway remains open.

5. Reclassifying the unreal

5.1. The unreal as misindexed, derivative, or residual reality

Reference-indexed realism resolves the return of the unreal by refusing the flat binary between real and unreal.

It does not say that everything is real.

It asks in what respect the excluded term failed, and in what respect it remains operative.

An illusion is not real as the object it presents. The mirage is not water. The hallucinated voice is not an external speaker. The distorted memory is not a faithful copy of the event. But the illusion may be real as experience, real as neural event, real as perceptual registration, real as behavioural cause, real as evidence of misrelation, and real as a failure of perceptual modelling.

A fiction is not real as historical occurrence. The invented character did not walk through the documented street at the documented hour. But the fiction is real as text, artefact, performance, social reference, emotional trigger, symbolic structure, cultural inheritance, and possible disclosure of social or psychological relation.

A possibility is not real as actuality. The future danger has not yet arrived. The alternative path has not been taken. The possible state is not the present state. But possibility may be real as branch-space, risk, affordance, counterfactual constraint, planning horizon, strategic orientation, or modal structure.

An absence is not real as present object. The missing witness is not in the room. The unpaid debt is not a material object. The unspoken word is not an utterance. But absence may be real as gap, loss, omission, silence, exclusion, lack, unfulfilled role, causal failure, or constraint on action.

A social construction is not real as mind-independent natural kind merely because it is socially enforced. But it may be real as institutional category, legal force, practical constraint, identity structure, power relation, obligation, permission, exclusion, or distributive mechanism.

A mathematical object is not real as physical body. It has no mass, location, or ordinary causal pressure. But it may be real as formal invariant, structural necessity, object within a formal regime, or constraint on valid inference.

A moral value is not real as empirical object. Wrongness cannot be weighed like stone. Obligation cannot be photographed like a tree. But moral reality may appear as normative constraint, reason, demand, harm, dignity, violation, accountability, or repair.

A political myth may be unreal as description. It may falsify history, conceal violence, exaggerate unity, or invent destiny. Yet it may be real as mobilising force, legitimating structure, identity formation, institutional glue, and justification for action.

In each case, reference-indexed realism asks: under what index was the thing denied, and under what index does it return?

This is the central answer to realism’s instability.

The unreal returns because many things rejected under one index remain operative under another. They are not real as first claimed. But they are not nothing.

The discipline is not to admit them all as equally real. The discipline is to reclassify the claim.

The fiction is not real as event, but real as disclosure.

The illusion is not real as object, but real as experience.

The possibility is not real as actuality, but real as branch-space.

The absence is not real as presence, but real as missing condition.

The construction is not real as nature, but real as institution.

The abstraction is not real as target, but real as model.

Reference-indexed realism therefore preserves distinction while preventing blindness.

5.2. Re-reading the modes of realism

Metaphysical realism remains indispensable.

There is reality beyond any particular observer. The target is not created by the model. Physical differences may exist whether or not any locus registers them. Stars, organisms, bodies, energies, processes, structures, and events do not depend on being named in order to exist.

Reference-indexed realism preserves this commitment.

But it denies that metaphysical independence grants complete model access. To say that something exists independently is not to say that any finite observer possesses it as total state. Independence belongs to the target. Access belongs to the relation. Modelling belongs to the indexed encounter between target and observer.

Mind-independence does not entail model-completeness.

Perceptual realism can be re-indexed in the same way.

Perception may disclose real relations. When I see a table, I am not necessarily trapped inside private appearances. My perceptual system may register a real relation between body, light, surface, spatial arrangement, motor possibility, and environmental stability. But this registration is bounded. It occurs through sensory range, embodiment, attention, scale, history, action capacity, and abstraction.

Illusion does not destroy perceptual realism. It indexes it. The illusion is a failed model of the target relation, not absence of all reality. The appearance is real. The perceptual process is real. The behavioural consequence may be real. The external object as presented may not be real.

Perception is therefore neither unmediated possession nor mere fiction.

It is bounded access.

Scientific realism becomes clearer when re-indexed.

Science does not escape modelling. It refines modelling. It makes access relations explicit, instruments calibrated, scales specified, variables formalised, measurements repeatable, and comparators available.

A scientific model is not the target. It is a disciplined abstraction of a target relation.

Its strength lies not in being unabstracted, but in preserving relations under correction. A good scientific model allows prediction, intervention, measurement, replication, explanation, and revision. It tracks where it holds and where it fails. It can be challenged by experiment. It can be improved when residual becomes consequential.

This gives scientific realism its force. The success of science is not magical mirroring. It is the progressive strengthening of indexed models under target constraint and observer audit.

Unobservable entities can be real where they are required by models that preserve target relations across measurement, intervention, prediction, and correction. But the model remains approximate. It may be replaced, refined, reduced, expanded, or reinterpreted. Its success does not make it complete ontology.

Moral realism and mathematical realism both show why realism cannot be restricted to physical objecthood.

A moral wrong is not a stone. It is not an empirical object located in space in the same way as a body. But wrongness may still be real as normative constraint. It may appear in relations of harm, violation, obligation, dignity, reciprocity, agency, domination, betrayal, and repair. It constrains what can be justified. It calls for response. It can be denied, but denial may fail against the relation it attempts to suppress.

Mathematical objects need not imitate physical objects in order to be real. Their reality may be formal, structural, invariant, necessary, or regime-internal. A theorem is not a body, but it constrains valid inference. A number is not a stone, but it may participate in structures that hold independently of individual preference. A mathematical model may abstract from material particulars while preserving real relations of quantity, symmetry, transformation, probability, or form.

Social reality is the clearest case where dependence does not entail unreality.

Money, borders, citizenship, debts, corporations, laws, offices, ranks, marriages, academic degrees, property, and rights are not natural objects in the simplest sense. They depend on human practice, recognition, enforcement, memory, and institution. But they are not imaginary.

They alter what bodies can do. They distribute power. They create obligations. They authorise violence. They organise labour. They shape identity. They open and close futures.

Aesthetic reality follows the same pattern.

A fiction is not real as literal event. But it may be real as a work, form, memory, cultural object, affective structure, moral experiment, shared reference, and disclosure of pattern. A novel may invent characters who never lived and still reveal real social pressures. A poem may describe no actual scene and still disclose grief, time, desire, or loss. A painting may distort appearance and still reveal relation.

Political realism then becomes more complex than the demand to face power alone. Power is real. Coercion is real. Scarcity, interest, force, and institutional constraint are real. But political orders are also built from symbols, myths, legitimacy claims, legal fictions, narratives, collective identities, and imagined futures.

A myth may be false as history and real as mobilisation. A constitution may be text and real as authority. A nation may be imagined and real as institutional force. A legitimacy claim may be contestable and real as stabilising structure.

Reference-indexed realism therefore refuses the crude opposition between material power and fiction. Political reality often consists in their coupling.

Constructed does not mean unreal.

Fictional does not mean inert.

Symbolic does not mean powerless.

Dependent does not mean imaginary.

The question is always: real under what index, through what access relation, at what scale, and with what effect?

6. Guardrails and conclusion

6.1. Guardrails against misreading

Reference-indexed realism must be distinguished from several nearby positions.

It is not relativism.

Relativism says, or is often taken to say, that truth and reality depend merely on perspective, framework, culture, or preference. Reference-indexed realism says something different. It says that claims are evaluated through indexed relations, but those relations are constrained. Not every index works. Not every model preserves the target. Not every abstraction survives correction. Not every social category is justified by being socially powerful. Not every fiction discloses truth. Not every measurement captures the target relation.

The relation is relative. The constraint is real.

It is not naïve realism.

Naïve realism imagines that the observer can simply possess the world as it is. Reference-indexed realism denies total possession. Every finite observer encounters the target through boundary, interface, scale, horizon, abstraction, registration, retention, and correction. The observer does not disappear.

It is not informational closure.

Reference-indexed realism does not treat the boundary of RRP-indexed availability as the boundary of information in reality. Failure to establish operational information for one locus does not establish the absence of thermodynamic, statistical, physical, or other informational relations beyond that access path. Conversely, broader informational status does not establish availability, registration, retention, or use for the evaluated locus.

It is not constructivism alone.

Constructivism rightly notices that models, categories, institutions, and meanings are made. But reference-indexed realism adds that constructions are constrained. They can fail. They can misclassify. They can break against target relations. They can produce harm by ignoring residual. Construction is real, but not sovereign.

It is not ontological inflation.

Reference-indexed realism does not declare everything real in the same way. It distinguishes modes, scales, and indices. An illusion is not real as external object. A fiction is not real as event. A social construction is not real as natural kind. A possibility is not real as actuality. Each may still be real under another index. The point is not to flatten reality, but to discipline the claim.

It is not reductionism.

Reductionism treats one scale as the real scale and everything else as derivative or unreal. Reference-indexed realism denies that lower-scale reality automatically exhausts higher-scale organisation. Molecular, organismic, personal, institutional, mathematical, moral, and political realities may each require their own index. Lower-scale constraint matters, but it does not always settle the higher-scale target.

It is not sceptical collapse.

The fact that access is mediated does not mean reality is unknowable. Mediation is not unreality. A model can preserve real relation. A measurement can be reliable. A theory can be powerful. A perception can disclose. A category can track. A moral judgement can be warranted. A scientific model can become more objective through correction.

Reference-indexed realism is therefore pluralist about modes of reality but strict about the conditions under which each mode may be claimed.

It does not weaken realism.

It prevents realism from pretending to be complete ontology.

6.2. Conclusion: realism after complete ontology

Realism begins by insisting that the real exceeds us. That commitment should remain.

The world is not made true by our thought. The target is not created by the model. Reality can resist, correct, and defeat our claims. A model may fail because the target does not support it. A category may collapse because it misclassifies. A theory may be abandoned because it cannot preserve the relation. A political myth may mobilise power and still be false. A moral denial may be convenient and still fail against suffering.

But realism becomes unstable when it treats the real as an unindexed total category.

The unreal then returns. It returns as illusion, fiction, possibility, absence, abstraction, social construction, mathematical structure, moral demand, political myth, and residual effect. It returns because what realism excluded was often not nothing. It was unreal under one index and operative under another.

Reference-indexed realism stabilises the problem by refusing both premature exclusion and indiscriminate absorption.

It asks what target is being modelled. It asks what finite observer or modelling locus is doing the modelling. It asks how the target becomes available. It asks what scale is operative. It asks what horizon is claimed. It asks what mode of registration is involved. It asks what comparator would defeat or weaken the model. It asks what residual remains.

This changes the task of realism.

The task is not to achieve complete ontology. No finite observer can possess the target as total state. Every model is approximate because every model is formed through relation. The target constrains what the model must answer to. The finite observer constricts what the model can register. The RRP scaffold supplies the mechanical bridge between substrate difference and operational distinction. The model preserves selected relations. The residual marks what closure has not resolved.

The task of realism is therefore disciplined approximation under constraint.

This is not a retreat from reality. It is a more exact realism.

A realism that admits no residual becomes dogmatic.

A realism that admits every residual without discipline becomes inflated.

A realism that ignores the observer becomes naïve.

A realism that ignores the target becomes constructivist collapse.

A realism that mistakes the model for the target becomes complete ontology.

A realism that lacks RRP grounding becomes mere contextualism.

Reference-indexed realism avoids these failures by keeping the full structure visible.

Substrate difference is real before observation and may carry informational structure beyond any particular access relation.

RRP makes selected difference operationally available as distinction for a locus.

Observation registers distinction through bounded access.

The observer constricts what can be registered.

The model approximates the target.

The residual prevents premature closure.

The access boundary limits attribution to the locus; it does not exhaust the target or informational reality.

Objectivity is disciplined correction under constraint.

The real remains what exceeds and corrects us. But finite observers encounter that excess only through indexed relation. Objectivity is therefore not the disappearance of the model, nor the disappearance of the observer. It is the correction of the model under the pressure of a target it never fully contains, through an observer that never stands outside the relation.

Realism after complete ontology is not weaker realism.

It is realism grounded in the mechanics of finite access.

Endnotes

References

Benoist, J. (2021). Toward a Contextual Realism. Harvard University Press.

Bensusan, H. (2021). Indexicalism: The Metaphysics of Paradox. Edinburgh University Press.

Bhaskar, R. (2008). A Realist Theory of Science. Routledge. (Original work published 1975).

Giere, R. N. (2006). Scientific Perspectivism. University of Chicago Press.

Lewis, D. K. (1986). On the Plurality of Worlds. Basil Blackwell.

Maddy, P. (1990). Realism in Mathematics. Clarendon Press.

Massimi, M. (2022). Perspectival Realism. Oxford University Press.

Nagel, T. (1986). The View from Nowhere. Oxford University Press.

Putnam, H. (1981). Reason, Truth and History. Cambridge University Press.

Searle, J. R. (1995). The Construction of Social Reality. Free Press.

Shafer-Landau, R. (2003). Moral Realism: A Defence. Oxford University Press.

Endnotes

1.　Putnam (1981) is a useful comparator for a form of metaphysical realism that combines mind-independent reality with a fixed complete description from an external standpoint. The present paper shares the rejection of complete model-possession but retains a stronger target/model asymmetry than Putnam's internal-realist alternative.



2.　Bhaskar's A Realist Theory of Science (2008 [1975]) distinguishes the intransitive objects of scientific knowledge from the transitive, historically produced dimension of knowledge. This is a close comparator to target/model asymmetry, but Reference-Indexed Realism does not inherit Bhaskar's transcendental argument, stratified ontology, or wider critical-realist programme.



3.　The moral, mathematical, and modal labels are used schematically here. Representative formulations include Shafer-Landau (2003) for moral realism, Maddy (1990) for mathematical realism, and Lewis (1986) for the strong possible-worlds form of modal realism. These are provenance sources only; the present paper does not inherit their substantive metaphysical commitments.



4.　Searle (1995) is a useful social-ontology comparator because institutional facts can depend on human practices while remaining objectively consequential. The present argument generalises the dependence point across indices and does not inherit Searle's account of collective intentionality or constitutive rules.



5.　Benoist (2021) develops a contextual realism that rejects acontextual ontology while preserving realism. This is a close comparator to the present rejection of unindexed claims, but Reference-Indexed Realism grounds the index in a physical access scaffold and retains explicit target/model/residual discipline rather than treating context alone as the governing relation.



6.　Bensusan (2021) is a particularly close terminological and metaphysical neighbour: Indexicalism treats what is real as ultimately indexical and rejects totalising metaphysics. Reference-Indexed Realism differs at the decisive point that the index constrains a finite reality-claim while the ontological target may exceed every current index. No developmental dependence has been established.



7.　Nagel (1986) is the provenance source for the 'view from nowhere' formulation and analyses the tension between subjective and objective standpoints. Reference-Indexed Realism does not adopt a perspective-free ideal; it treats objectivity as disciplined correction of an indexed model under target pressure.



8.　Giere (2006) and Massimi (2022) provide the nearest scientific perspectivist comparators: observation and scientific representation are perspectival, and realism need not require perspective-free access. Reference-Indexed Realism overlaps at the level of finite, mediated modelling but is broader in scope and grounds the relation in the RRP access scaffold, target/model asymmetry, comparator discipline, and residual.





