This essay develops the philosophical motivation for the relational situational ontology that grounds this system. It takes seriously the concept of a world model — a representation of reality sufficient to support reasoning and action — and argues that the right ground for such a model is neither propositional nor essentialist but purely relational. The argument moves from Harman's object-oriented ontology through its limitations to Leibniz's relational alternative, then treats the problem of identity that any purely relational view must face, and arrives at topology and symmetry as the answer.
#World Models
The phrase "world model" circulates widely in contemporary AI discourse without a settled definition. The aspiration is clear: a representation of reality that enables anticipation, planning, and situated action — the kind of internal model that allows an agent to reason about the world rather than merely react to it. The old symbolic AI tradition pursued this through propositional systems, inserting structured assertions about the world and deriving consequences from them. That tradition is largely superseded but not forgotten, and its ambition remains the right one.
What the relational situational ontology proposed here shares with the world model tradition is that ambition: to represent reality in a way that can ground reasoning. Where it departs from the propositional tradition is the mechanism. Instead of a list of atomic propositions about intrinsic properties of things, we have a landscape of relations. Things are not defined by what they are in themselves but by what they connect to, how they relate, what position they occupy in a web of other things. This is not a difference in degree but in kind — it is a different ontological grammar.
The ontology we build is of agents, actions, locations, and situations. It can represent mathematics as easily as social arrangements. It models the same world at different scales and from different perspectives, because each situation carries its own perspective on the whole. This is why the model is perspectival and multi-scale without collapsing into relativism: there is a fact about every relation, even though no single perspective captures all of them.
#Harman's Object-Oriented Ontology
The deepest recent philosophical statement of a broadly relational view is Graham Harman's object-oriented ontology, which is a serious predecessor to what we propose, though it arrives at a different destination.
Harman's central move is to extend the logic of perception beyond minds. When a flame burns cotton, he observes, the flame does not encounter the full reality of the cotton. The cotton has softness to the touch, a characteristic smell, a color that only appears to eyes — properties that exist but that the flame never activates. The flame encounters only the combustible surface of the cotton, not the cotton in its entirety. Harman generalizes from this: every object encounters every other object through a subset of its properties, through what he calls the sensual qualities of the sensual object. Perception is not a privileged cognitive achievement that only minds have. It is a general structural feature of all material interaction: objects meet through surfaces of contact, not through complete mutual transparency.
This is a powerful move. It democratizes ontology, refusing to grant minds a special status as the only entities capable of representation. And it explains a basic phenomenological observation: things always seem to exceed our grasp of them. There is always more to the cotton than the flame encounters, more to a person than their friend knows, more to a legal case than the lawyer has assembled. Harman gives this excess a name and an explanation: withdrawal, grounded in the distinction between real and sensual objects.
But here the trouble begins. To explain why things always exceed appearances, Harman posits a real object behind every sensual presentation — the true, unknowable core of the thing, protected from any encounter by the veil of sensual qualities. His ontology is essentialist: at the heart of every object is something that appearances only partially reveal, something with real qualities that no observer ever directly accesses.
Two problems arise. The epistemological problem: if all contact is mediated by sensual qualities, and the real object is always withdrawn, how do we ever say anything true about real objects? Harman develops notions of allure and aesthetic experience as oblique access routes, but these are strained. The more fundamental problem is that the posited essences do no necessary explanatory work. The observation that things always exceed any particular encounter with them can be derived from a purely relational ontology, without residue, without invoking hidden natures. When essence is not needed to explain what it was invoked to explain, the principle of parsimony speaks clearly.
There is a further difficulty, harder to articulate but important. Essentialism has a history. The philosophical conviction that things have hidden true natures — natures that appearances only imperfectly disclose — has served as the engine of racial categorization, of purism, of a broad family of projects that sort things into authentic and inauthentic kinds. This history does not refute an essentialism, but it raises the bar for what it must earn. If the same explanatory work can be done without hidden essences, the burden falls on the essentialist to show why the essences are needed.
#The Relational Alternative
There is a different tradition, older and more comfortable with the language of science. Leibniz's monadology is the ancestor. Each monad — the fundamental constituent of reality in Leibniz's system — is an entity with a partial, perspectival glimpse on the infinite whole. Monads do not interact directly in Leibniz's system; he requires a pre-established harmony to coordinate their perspectives, which is an elaborate theological rescue operation. But the core intuition survives: a thing is constituted partly by its perspective on, and position within, the relational whole.
Modern physics expresses the same intuition without the theological machinery. When we say that an atom has mass, or that gravitational attraction falls with the square of distance, or that charge is conserved, we are describing relations between entities. We never arrive at the intrinsic essence of any entity — we only ever describe its relations with other things. Relationality is the natural language of science. An ontology that makes relations second-class — that reserves ontological primacy for intrinsic natures and treats relations as derivative — sits uncomfortably with the practice of science, which never isolates or characterizes any intrinsic nature at all. Aesthetically and methodologically, essentialism conflicts with science's actual epistemic situation.
The alternative is to begin with relations and derive everything else. If a thing has infinitely many relations with other things, then any one encounter with it activates only a finite subset. The thing is always in excess of any particular encounter, not because of a hidden essence, but because most of its relations are not present in any given interaction. Withdrawal from full disclosure follows directly from infinite relational connectivity. You do not need a real object behind the sensual object. You need only the observation that every relational entity has more relations than any one situation can contain.
Consider a person you know well. You believe you understand them. Then you see them in a new context — with a stranger, in an unfamiliar setting — and something appears that your relationship with them never elicited. This is not the emergence of a hidden soul. It is the overdetermination of relational character: the same person enters different relations in different situations, and each relation draws out something that others did not. The surplus is real; its explanation is relational, not essentialist. What you thought was familiarity was partial, not because the person conceals a hidden core, but because any relationship activates only a slice of an infinitely ramified relational being.
#The Problem of Identity
If things are nothing but relations, the question of identity is urgent. What holds a thing together across time, if there is no essence carrying it forward? The essentialist answer is simple: the soul, the substance, the real object — something internal and persistent grounds identity through change. In a purely relational ontology, none of these is available. How, then, does a thing remain the same thing through growth, transformation, damage, repair?
The answer lies in topology.
Topology is the study of properties preserved under continuous deformation. A sheet of paper and a crumpled sheet have the same topology: you can deform one into the other without tearing or gluing, and the fundamental relational structure — which regions connect to which — is preserved. A donut and a sphere do not have the same topology: no continuous deformation maps one to the other, because they differ in a fundamental connectivity property (the donut has a hole; the sphere does not). A bowl and a cup without a handle are topologically equivalent; adding a handle to a cup creates a new topological feature that cannot be removed by continuous deformation.
The proposal is this: a thing's identity persists through changes in its relational arrangement when that arrangement is a continuous deformation of its previous arrangement. Identity is topological invariance through transformation. The marble worn smooth is still the same marble because its connectivity — its topological structure — has not changed, even though its surface geometry has. The body that has aged, grown, healed, metabolized, is still the same body because the topological structure of its organization — the connectivity of systems, organs, surfaces, functions — has been maintained through all those geometrical changes.
This account fits the way we actually use identity. We say the river is the same river even though its water is entirely replaced, because the topological structure of banks and flow patterns persists. We say the institution is the same institution through changes of personnel, budget, and mandate, because the relational structure that constitutes it — the web of roles, rules, communications, dependencies — is continuously deformed rather than discontinuously ruptured.
Symmetry offers the same intuition from a different angle, and connects it explicitly to physics. A symmetry is a transformation that leaves an object's fundamental character unchanged. A sphere is invariant under rotation: however you rotate it, it remains the same sphere. The conservation laws of physics — conservation of mass, energy, momentum, charge — are all expressions of symmetries: properties of the universe that are invariant under specified transformations (time translation, spatial translation, Lorentz boosts). Emmy Noether showed that every conserved quantity corresponds to a symmetry; conservation and symmetry are the same phenomenon described differently.
Identity is the same structure. Something persists as the same thing when it has a symmetry — a pattern of invariance — that its changes respect. What makes a person the same person through decades of change is not a hidden soul but a symmetry: a pattern of relational invariance that is preserved through the ongoing transformations of living. When the symmetry fails — when the pattern breaks — we perceive change of kind, not merely change of state. We say the person has become someone else, or has ceased to exist.
#Birth, Death, and Ontological Honesty
This account of identity carries a consequence that is, if anything, a point in its favor: it makes the boundaries of existence tractable rather than mysterious.
Essentialism struggles with birth and death. If a thing is constituted by a hidden essence, when does that essence arrive? When does it depart? The problems of personal identity — when does a fetus become a person, when does a body stop being the person it was, what is the relationship between the infant and the adult who will have no memory of it — are hard for essentialist views precisely because they treat identity as binary. Either the essence is present or it is not. But there is no principled way to say when it arrives or departs without importing assumptions from outside the ontology.
In a relational topology, birth and death are not binary events requiring arbitrary stipulation. Birth is the emergence of a topologically stable relational configuration — the establishment of a new pattern of invariance that wasn't there before. Death is the dissolution of that pattern: the topology changes, the invariance conditions fail, and what was a persistent relational identity disperses into other configurations. These events can be graded and perspectival: different observers, operating with different grain of description, may identify them at different moments. This is not vagueness in the ontology; it is accuracy about what happens. Boundaries are real but often fuzzy; the relational topology acknowledges this rather than resolving it by fiat.
More importantly, things can come into and out of existence — and can split, merge, and transform — without ontological crisis. A cell divides: one topological pattern becomes two. Two streams merge: two become one. A person fragments into conflicting identities under severe psychological pressure; the topology frays. These processes are natural in a relational ontology, whereas essentialism tends to make them paradoxical, demanding answers about where the essence went or which of the two descendants carries the original.
#What This Grounds
The relational situational ontology built in this system rests on this philosophical foundation. When the formal ontology defines objects by their position in actor × method × domain space — by their relations along three axes — it is expressing the relational commitment: things are what they are by virtue of their connections, not by virtue of hidden properties. When situations are described as dynamic intersections that activate a subset of each entity's relational character, it is expressing the perspectival consequence: no situation exhausts any of the entities it involves. When identity across sessions, projects, and actors is maintained through structural continuity rather than persistent state, it is expressing the topological commitment: what persists is not a stored soul but an invariant pattern.
The world model, on this account, is not a picture of intrinsic natures. It is a map of a relational landscape — partial, perspectival, always incomplete, always in excess of any one engagement with it.
#Badiou: Inconsistent Multiplicity and the Count-for-One
The relational ontology developed above takes relations as primitive and derives entities from their position in a relational web. It answers Harman by dissolving essence into topology, and it answers Leibniz by discarding pre-established harmony in favor of situated perspectives. But it has not yet addressed the question that Alain Badiou places at the center of ontology: what is the status of multiplicity itself, before any structuring operation carves it into identifiable units?
Badiou's ontological project, developed across Being and Event (1988; English translation 2005) and Logics of Worlds (2006; English translation 2009), makes a claim that sounds outrageous until you follow the argument: ontology is mathematics. Specifically, ontology is axiomatic set theory (ZFC). This is not a metaphor. Badiou means that the formal theory of pure multiplicity — sets composed of sets, with no ur-elements, no atoms, no intrinsic natures — is the only discourse adequate to being qua being. Everything else (physics, biology, politics, art) is a regional ontology, a discourse about some particular domain. Only set theory speaks of being as such: pure multiplicity without predication.
Three features of Badiou's framework bear directly on the relational situational ontology.
#Mess as Pre-Ontological Ground
Badiou distinguishes inconsistent multiplicity from consistent multiplicity. Inconsistent multiplicity is what exists before any structuring operation — pure multiplicity without identifiable units. It is not chaos in the colloquial sense (randomness, disorder); it is something more radical: multiplicity that has not yet been submitted to what Badiou calls the count-for-one, the operation that carves discrete units from the continuum of being. Inconsistent multiplicity is unthinkable in a precise sense: you cannot point to any part of it without already counting that part as one, thereby imposing structure on what is, by definition, pre-structural.
In the relational situational ontology, this corresponds to what TTT (the Turing Tape Theory developed in the foundations) calls Mess — pure relationality without identifiable entities. Mess is the pre-condition that entity-creation presupposes. Before any belonging is appended to the ledger, before any entity is differentiated from any other, there is the undifferentiated relational ground from which differentiation proceeds. Mess is not an entity in the ontology. It cannot be, because entities are constituted by belongings, and Mess is what obtains before the first belonging. It is like the blank Turing tape before the first symbol: real, necessary for the operation of the machine, but not itself a symbol.
The count-for-one — Badiou's term for the operation that produces identifiable units from inconsistent multiplicity — corresponds to entity creation in the relational ontology. When a belonging is appended to the ledger, it simultaneously creates an entity (the thing that now has this belonging) and differentiates that entity from the pre-ontological ground. The ontology operates on the structured side of the count-for-one. It presupposes that entities have already been counted, that belongings have already been assigned, that the structuring operation has already occurred. Mess is what it operates on, not what it operates within.
This is not a deficiency. Every formal system begins somewhere, and the somewhere is always already structured. Badiou is explicit about this: set theory itself cannot describe inconsistent multiplicity, because the axioms of ZFC already impose the count-for-one (every set is a set of sets, already counted). The inconsistent multiple is what the axioms retroactively posit as having been there before the axioms took hold. The relational ontology makes the same move: Mess is what must have been there before the first belonging, even though the ontology can only describe what comes after.
#Infinity as Openness
Badiou inherits from ZFC the axiom of infinity: infinite sets exist. This is not a concession to pathology or a reluctant admission that some things are very large. It is constitutive. Infinite multiplicity is the natural state of being; finitude is the exception, produced by specific structuring operations that bound and delimit. The count-for-one produces finite situations from infinite being. But the infinite is always there underneath, always in excess of any particular count.
In the relational ontology, infinity appears not as an entity but as the absence of a closure axiom on the quality graph. The quality vocabulary — the set of all qualities that can characterize belongings — is finite at any snapshot. There are hundreds of qualities in the current system: authored-by, located-in, identified-by, tagged, and so on. But there is no axiom saying "these are all the qualities that will ever exist." New domains produce new qualities. When the system encounters a domain it has never modeled before — genomics, contract law, liturgical music — new qualities emerge to describe the relations that matter in that domain. The quality graph grows without bound.
This is analogous to a Turing tape: finite at any step of computation (only finitely many cells have been written), but potentially infinite (there is no bound on how many cells may eventually be written). The ontology's constructive finiteness — we store finite data in SQLite, we process finite documents, we run on machines with finite memory — is a practical constraint, not an ontological commitment. The ontology does not assert that reality is finite. It asserts that any particular structuring of reality (any particular situation, any particular snapshot of the quality graph) is finite, while the space of possible structurings is not. Infinity lives not in any entity but in the openness of the system to new qualities, new belongings, new situations that no current vocabulary anticipates.
Badiou would recognize this as a consequence of his own framework: every situation is finite (the count-for-one produces bounded presentations), but being itself — the inconsistent multiplicity that every situation structures — is infinite. The gap between the situation and being is irreducible. In the relational ontology, the gap between the current quality graph and the space of possible qualities is the same irreducible gap.
#Situation as Consistent Multiplicity
Badiou's most productive distinction, for our purposes, is between inconsistent multiplicity (being qua being, before structure) and consistent multiplicity (being structured by a count-for-one into a situation). A Badiouian situation is a structured presentation: multiplicities that have been counted-for-one and organized by what Badiou calls the state of the situation — a meta-count that groups the groups, a second-order structuring that organizes the first-order units into a legible arrangement.
The relational ontology's situation (Definition 11 of ontology-objects) is Badiou's consistent multiplicity. Entities — already counted-for-one from Mess — are organized by belongings (the structuring relation) within a domain (which plays the role of the state of the situation, providing the second-order organization that makes the first-order structure legible). The parallel is not metaphorical. The structures are isomorphic in the following sense: both begin with an unstructured ground (inconsistent multiplicity / Mess), both apply a structuring operation (count-for-one / belonging-creation) to produce identifiable units (elements of a set / entities), and both organize those units into structured wholes (situations / situations) governed by a meta-structure (state of the situation / domain).
The difference is in the formalism. Badiou's ontology is set theory: the primitive is membership, and everything is built from sets of sets with no ur-elements. The relational ontology is a category with one primitive morphism (belongs-to), and everything is built from entities and their belongings. Set-theoretic membership and categorical belonging are different formal primitives, but they serve the same structural role: they are the minimal relation from which all other structure is derived. Both achieve the same move: structure from multiplicity, situations from mess, entities from the pre-ontological.
#The Event and Productive Inconsistency
There is a fourth connection, less structural and more dynamic, that links Badiou directly to the system's treatment of productive inconsistency (developed in Paper 4a and formalized in ontology-governance-situation).
Badiou's event is what disrupts a situation. An event is something that belongs to the situation — it is composed of elements that the situation contains — but that the state of the situation cannot account for. The state, which is supposed to organize and classify every element of the situation, encounters something it cannot classify: a configuration of existing elements that exceeds the state's organizational capacity. The event is not an invasion from outside. It is an internal excess, a multiplicity that the situation's own structuring operation produced but cannot recognize.
In the relational ontology, a fork in the directed acyclic graph (the constitutional ledger) is an event in Badiou's sense. A fork occurs when an append — a new belonging, a new structuring assertion — cannot be validated by the current constitutional scope. The existing governance structure (the state of the situation) lacks the capacity to classify this new element. Rather than rejecting it (which would be censorship, the suppression of an event) or accepting it uncritically (which would be dissolution of the existing structure), the system forks: it creates a new tag partition, a new branch of the ledger, where the unclassifiable append can exist alongside the existing structure without either destroying or being destroyed by it.
This is productive inconsistency: the structural capacity to hold both the pre-event situation and the post-event fork without collapsing either. The event does not destroy the situation; it creates a new situation alongside it. The two situations share a common history (the ledger up to the fork point) but diverge in their structuring principles (the constitutional scope that governs each branch). This is not contradiction in the logical sense — both branches are internally consistent. It is inconsistency in Badiou's sense: the impossibility of a single state accounting for all the multiplicities that the situation contains.
Badiou's framework illuminates why productive inconsistency is not a bug but a feature. If a system could always account for everything it contains — if the state of the situation were always adequate to the situation — then nothing genuinely new could ever appear. Every novelty would be classifiable in advance, which means it would not be novel. The capacity for events — for internal excess that exceeds the current structure — is the capacity for genuine change. The relational ontology, by building forks into the ledger rather than treating them as failures, encodes this capacity formally.
ontology · 2026-03-03 · source: voice memo (17m) · zach + claude
Ontology 08 — The Relational Ground — 2026 — Zachary F. Mainen / HAAK