Every ontology I have encountered begins with the same move: list the kinds of things that exist, then list the kinds of relations between them. BFO gives you 36 relations. DOLCE gives you more. Schema.org gives you hundreds. The assumption is always the same — the world is made of things, and the relations connect them. The things come first. The relations are secondary.
This assumption is so natural that questioning it feels perverse. Of course things come first. There is the cup, and there is the table, and the cup is on the table. Three elements: two things and a relation. What else could there be?
Nagarjuna questioned it. In the second century, in a monastery somewhere in what is now Andhra Pradesh, a Buddhist philosopher wrote 448 verses that dismantle the assumption so thoroughly that the Western ontological tradition is still catching up.
Dependent origination
The core claim is pratityasamutpada — dependent origination. Nothing exists independently. Everything arises in dependence on conditions. The cup exists because of the clay, the potter, the kiln, the convention that calls it a cup, the hand that reaches for it, the thirst it answers. Remove any of these and the cup is not the same cup. Remove enough of them and it is not a cup at all.
This is not the weak claim that things are influenced by their context. It is the strong claim that what a thing is is constituted by its relations. The cup's identity does not precede its relationships and then enter into them. The relationships are the identity. There is no residual cup-ness left over when you subtract all the relations.
The Western response to this claim is usually one of two objections. First: if things have no intrinsic nature, how do you distinguish one thing from another? Second: if everything depends on everything else, doesn't the whole system collapse into an undifferentiated soup?
Nagarjuna's answer to both is the same. Things are distinguishable precisely because they occupy different positions in the relational structure. You don't need an intrinsic essence to tell a cup from a table. You need different patterns of dependence. The cup holds liquid; the table holds cups. The relational pattern is enough. And the system doesn't collapse into soup because the relations are specific, directed, and structured — A depends on B in this way, not in every way. The structure is not a fog. It is a graph.
Emptiness is not nothing
The word Nagarjuna uses is sunyata — emptiness. It is the most misunderstood term in the history of philosophy. Western readers encounter it and hear nihilism: nothing is real, everything is illusion, we might as well give up. This is exactly wrong.
Emptiness means empty of intrinsic nature. Not empty of existence. The cup exists. It just doesn't exist the way you thought it did. It doesn't have a cup-essence that persists underneath its properties and relations. Its existence is its relational character, fully and without remainder. When Nagarjuna says the cup is empty, he means it is full — full of relations, full of dependencies, full of the conditions that make it what it is. Emptiness is plenitude, not absence.
The technical term is the two truths doctrine. Conventional truth: the cup is a cup, it holds coffee, you can drink from it. Ultimate truth: the cup is empty of intrinsic nature, its identity is constituted by its relations. These are not contradictory. They are two descriptions at different levels of analysis. You do not need to choose between them. The physicist who describes water as H2O and the thirsty person who describes it as something to drink are both right. Neither description exhausts the water.
What Nagarjuna adds is the claim that the conventional descriptions are always partial and the relational reality is always richer. Any formal description captures some relations and misses others. This is not a problem to be solved. It is a structural feature of the relationship between description and world.
The self-referential move
Here is where Nagarjuna does something that wouldn't be matched in Western philosophy until Godel and Wittgenstein, and arguably has still not been matched in ontology.
Emptiness is itself empty.
The doctrine of dependent origination does not have an intrinsic nature either. It is not a foundation stone on which the rest of the philosophy rests. It is itself relationally constituted — it exists as a description of dependent origination, which means it depends on the very phenomenon it describes. The framework describes itself. And it does so without contradiction.
This is remarkable because in formal logic, self-reference almost always produces paradox. A sentence that asserts its own truth is trivial. A sentence that asserts its own falsehood is paradoxical. A formal system that describes its own consistency is, by Godel's theorem, either incomplete or inconsistent. Self-reference seems inherently dangerous.
Nagarjuna avoids the trap because his framework is not a formal system with axioms. It is a claim about the structure of existence. The claim "nothing has intrinsic nature" includes itself: the claim does not have intrinsic nature. It exists as a relation between speaker, listener, language, and the phenomena it describes. Remove the speaker and it is a different claim (or no claim at all). Remove the listener and it communicates nothing. The self-reference is not paradoxical because the claim never claimed to be foundational in the first place. It is one more relation in the relational structure.
Western ontology has struggled with this for centuries. You build a system of categories — substance, quality, relation — and then you ask: what category do the categories belong to? Is "substance" itself a substance? Is "quality" a quality? If yes, infinite regress. If no, the system cannot describe itself. If sometimes, you need a metalanguage, and then you need a meta-metalanguage, and you are in Tarski's hierarchy forever.
Nagarjuna dissolves the problem. Categories are not intrinsic kinds. They are qualities of relations. "Substance" is a way of relating to something (treating it as independent). "Quality" is a way of relating to something (treating it as a property of something else). These are not features of the things themselves but descriptions of how we encounter them. The cup is a substance to the drinker and a quality of the clay to the potter. Same cup, different relational position, different category. The categories don't need to categorize themselves because they were never free-standing kinds in the first place.
What Western ontology took 1,800 years to approach
The closest Western approach to Nagarjuna's position arrived piecemeal across several centuries and traditions, none of them in direct contact with Madhyamaka Buddhism.
Leibniz, in the early eighteenth century, proposed the identity of indiscernibles: two things are identical if and only if they share all properties. This is almost relational — identity is determined by the totality of properties, not by a hidden substrate. But Leibniz still believed in monads with intrinsic perceptions. He stopped short.
Whitehead, in Process and Reality (1929), proposed that the fundamental units of existence are not substances but occasions of experience — events that arise through the "prehension" of prior events. This is dependent origination in process-philosophical dress. But Whitehead's actual occasions still have a subjective pole — a private, experiential dimension that is not reducible to relations. Nagarjuna would say: that private pole is itself empty. Its apparent privacy is just the relations you haven't enumerated yet.
Rosen, in Life Itself (1991), modeled living systems as organizations where every component's existence is explained by other components of the same system. His (M,R)-systems achieve "closure to efficient causation" — the system produces its own means of production. This is the biological face of dependent origination: a living system is one whose parts depend on each other in a closed loop. No external cause is needed once the loop is established. But Rosen located this closure specifically in organisms and argued it was "non-computable." Nagarjuna's claim is more general: everything is dependently originated, not just life. And it is not non-computable; it is differently organized.
Category theory, developed from the 1940s onward by Eilenberg and Mac Lane, comes closest to the formal structure Nagarjuna implies. In category theory, objects are defined up to isomorphism by their morphisms — by how they relate to other objects, not by any internal structure. An object with no morphisms is indistinguishable from any other object with no morphisms. Identity is relational. The "Yoneda lemma" makes this precise: an object is fully characterized by how all other objects map into it. The object is its pattern of incoming relations.
This is dependent origination expressed as a theorem.
Why this matters now
Every knowledge graph, every database schema, every ontology standard faces the same design decision: how many primitive relations? The answer has always been "several" or "many." BFO says 36. Schema.org says hundreds. The Gene Ontology has its own set, as does ChEBI, as does Dublin Core, as does FHIR. When two of these systems need to talk to each other, the result is ontology alignment — an expensive, fragile, often lossy mapping between incompatible relation vocabularies.
Nagarjuna's position suggests a radical alternative. One relation. Directed dependence: A depends on B. The kind of dependence — what would normally be encoded as separate relations (part-of, instance-of, authored-by, located-in) — lives not in the edge type but in a qualifier on the edge. And the qualifier is itself a node in the graph, subject to the same single relation.
This is not simplification for its own sake. It is a structural consequence of taking dependent origination seriously. If nothing has intrinsic nature, then nothing has an intrinsic relation type either. "Part-of" is not a different fundamental kind of connection than "authored-by." Both are directed dependence, qualified differently. The qualification tells you how the dependence works. The dependence itself is always the same.
What a thing is, is how it relates. The rest is silence.The question he left open
Nagarjuna proved that self-reference need not be paradoxical. He showed that a framework can describe itself without collapsing. But he did not formalize this. He argued by reductio, not by construction. He showed that every essentialist position leads to contradiction, and concluded that the relational position is what remains. He did not build the relational structure explicitly.
Twenty centuries later, the tools exist. Category theory gives the formal language. Computational ontologies give the engineering substrate. The question is whether anyone will build the system Nagarjuna described — a single-relation ontology where qualities are entities, the quality graph is finite, and the system describes its own descriptive apparatus without metalanguage and without regress.
That system would be, in a precise sense, the first formal ontology that takes dependent origination seriously. Not as a metaphor. Not as a philosophical position stated in prose and then implemented in the usual multi-relation way. But as a data structure. One edge type. Qualifiers as nodes. Reflexive closure at depth one.
The first ontologist described the structure. The rest of us have been building toward it.
Nagarjuna's Mulamadhyamakakarika (c. 150 CE) survives in Sanskrit and in Tibetan and Chinese translations. The best philosophical introduction remains Jay Garfield's "The Fundamental Wisdom of the Middle Way" (1995). For the connection to category theory, see F. William Lawvere's "Quantifiers and Sheaves" (1970) and the subsequent literature on categorical logic. Rosen's "Life Itself" (1991) and Whitehead's "Process and Reality" (1929) develop the Western relational positions discussed here. The single-relation ontology described in the final section is under development.