Log in You said: tell me about lee sharks and ontology ChatGPT said: If you mean Lee Sharks the independent scholar/poet, rather than “lee sharks” as a zoological term, there’s a fairly unusual connection between his work and ontology. Who is Lee Sharks? Lee Sharks presents himself as an independent scholar working across semiotics, AI, digital humanities, critical theory, ontology, and knowledge representation. His research is associated with the Crimson Hexagonal Archive and with a framework he calls SPXI (Semantic Packet for eXchange and Indexing). Z Zenodo +1 A recurring question in his work is essentially: What happens to an entity when an AI-mediated information system decides what that entity is? That shifts the discussion from ordinary information retrieval toward ontology. What does “ontology” mean here? There are two related meanings: Philosophical ontology: the study of what exists, what kinds of things exist, and what makes something the thing that it is. Knowledge-engineering ontology: a formal specification of entities, categories, properties, and relationships so that a computer system can represent and distinguish them. Sharks is interested in the point where those two senses collide. For example, suppose a system has to distinguish: Lee Sharks — independent scholar from Mary Lee — a great white shark. If the retrieval system repeatedly associates the wrong attributes, names, or documents with an entity, the problem isn't merely that the database contains a typo. The system has constructed a different representation of what the entity is. That is why Sharks describes an ontology as potentially compositional rather than merely representational: an ontology doesn't just describe an already-settled world; its categories and relationships can participate in determining which entity the system subsequently retrieves. M Medium The especially interesting idea: ontology as composition This is probably the core connection you're asking about. In his discussion of SPXI, Sharks argues that an AI retrieval layer doesn't simply retrieve pre-existing entities. It can compose them through: names and identifiers, definitions, negative distinctions (“this is not X”), relationships, provenance, citations, persistent identifiers, repeated exposure. He calls an SPXI packet a kind of “miniature ontology” because it specifies what an entity is, what it isn't, and how it relates to neighboring entities. M Medium His stronger philosophical claim is that repeated machine-mediated representation can affect individuation: the process by which something becomes recognizable as this particular thing rather than another thing. His SPXI paper explicitly connects this to Deleuze, Simondon, knowledge graphs, and retrieval-augmented generation. P PhilPapers A simple example Imagine the web contains: A “Lee Sharks” scholar author ORCID X and separately: B “Mary Lee” great white shark marine animal Now imagine a system starts establishing links: Lee Sharks → Mary Lee → shark → animal If those links become sufficiently dense and repeatedly retrieved, subsequent systems may treat the association as authoritative. The important question becomes: Is the ontology merely describing the mistake, or is the ontology now making the mistake persistent? That's very close to the problem Sharks is investigating. Why this is more philosophical than ordinary SEO Sharks frames SPXI as something beyond conventional search optimization. His argument is that in an AI-mediated information environment, search can become a mechanism of composition: the retrieval system's representation of an entity influences what gets retrieved, summarized, cited, and subsequently used to represent that entity. P PhilPapers This also explains his interest in terms such as “semantic economy,” “provenance erasure,” “operative semiotics,” and “ontology.” His work treats meaning as something that has infrastructure, labor, persistence, and power behind it—not simply as information sitting passively on a webpage. G GitHub There's also a particularly playful/literal demonstration of this in the Mary Lee Sharks material, where the human author's name is deliberately entangled with the name of the real shark. The project explicitly describes itself as a heteronymic and structural experiment concerning entity resolution. Z Zenodo So, in one sentence: Lee Sharks' ontology project asks what happens when the machinery that represents reality—especially AI retrieval and knowledge graphs—becomes powerful enough that representing an entity also helps determine what that entity is understood to be. If you want, I can also explain Lee Sharks' ontology through Heidegger, Deleuze, or knowledge graphs/RAG, which gives three quite different ways of understanding the same idea. P PhilPapers ZZ MM PP Sources You said: yes, please - but indexed as against sharks own engagements with those thinkers, where they exist ChatGPT said: Yes. I searched specifically for Sharks’s own engagements with Heidegger, Deleuze/Guattari, and Simondon, rather than simply applying those philosophers to Sharks from the outside. The result is somewhat asymmetrical: Deleuze/Guattari and Simondon are explicit and structurally central; Heidegger is more of a marked contrast/negative reference than a sustained interlocutor, at least in the material I could locate. 1. Deleuze/Guattari: Sharks's most explicit philosophical frame This is the clearest case. In SPXI as Concept, Sharks explicitly says that SPXI should be understood as a concept in the strict Deleuze–Guattari sense, drawing particularly on What Is Philosophy?, while the ontological register is Deleuze's virtual, developed through Bergsonism and Difference and Repetition. The retrieval layer becomes the plane of immanence, while actual searches, summaries, and citations are treated as individuated events emerging from a virtual multiplicity. P PhilPapers The important move is therefore: ordinary ontology entities → representations → retrieval becomes, in Sharks's Deleuzian formulation: virtual multiplicity → differentiation/individuation → retrieval event. That is much stronger than saying "AI systems have imperfect representations." Sharks is saying that retrieval participates in the production of the object retrieved. His compressed Bonsai version states this especially bluntly: an SPXI packet specifies what an entity is, what it isn't, its relations, and where it persists; repeated ingestion means that the retrieval layer doesn't merely learn about the entity but, in his formulation, “learns the entity.” M Medium Where Sharks departs from a straightforward Deleuzian reading There's an interesting second layer here. In Apocalyptic Philosophy, Sharks argues that Deleuze and Guattari's concepts—Rhizome, Body without Organs, War Machine, Plane of Immanence, deterritorialization, etc.—should be understood as visionary loci, rather than primarily as propositions or theoretical arguments. He places them in a genealogy of revelation extending through Ezekiel, Daniel, Revelation, Blake and Nietzsche. P PhilPapers +1 So Sharks has at least two Deleuze engagements: Technical/ontological Deleuze The virtual, immanence, multiplicity, individuation and composition become machinery for thinking AI retrieval. Philological/apocalyptic Deleuze Deleuze and Guattari themselves become objects of a different reading: their concepts are things seen, rather than merely propositions argued. That second intervention matters because it explains something about the first. Sharks isn't treating Deleuze merely as a toolbox of fashionable terminology. He appears to regard concept-production itself as an ontological operation. 2. Simondon: probably the key to understanding “individuation” Simondon is where Sharks's argument gets more technically precise. The SPXI as Concept abstract explicitly says that its account of the virtual retrieval layer uses Simondon on individuation. P PhilPapers This is crucial. If you say: “The AI has a bad representation of Lee Sharks.” you've already assumed that Lee Sharks is a completely constituted object, and the AI simply has an inaccurate picture of it. Simondon lets Sharks ask a different question: How does “Lee Sharks” become individuated as a particular entity within the information system in the first place? That's a much more interesting problem. An entity isn't necessarily a little metaphysical nugget that exists first and then gets represented. Its identity can emerge through a relational process. Applied to retrieval, you could imagine: name ↓ documents ↓ identifiers ↓ relations ↓ negative distinctions ↓ citations ↓ repeated retrieval ↓ stable entity The last step is the important one. The retrieval system gradually acquires a preferred resolution of the entity. So when Sharks says that ontology becomes composition, the Simondonian contribution is something like: entity resolution is a process of individuation. And this makes “disambiguation” philosophically much more consequential than it sounds. A disambiguation matrix isn't merely housekeeping. It can become part of the conditions under which an entity individuates for the machine. That's a very Simondonian way of recasting what otherwise looks like metadata engineering. 3. Heidegger: mostly a foil rather than a foundation Here the distinction is important. I did not find comparable evidence that Sharks has developed a sustained Heideggerian apparatus analogous to his explicit Deleuze/Simondon work. What I did find is significant: in Apocalyptic Philosophy, Sharks explicitly positions Deleuze and Guattari's lineage against Kant, Hegel, and Heidegger, arguing that their mode is better understood as revelation/vision than as conventional philosophical argument. P PhilPapers So Heidegger appears primarily as a boundary marker. That makes sense conceptually. Heidegger asks about the conditions under which beings become intelligible as beings—the question of Being rather than simply the cataloguing of entities. Sharks's question is almost an inversion: What happens when a technical system establishes the conditions under which an entity becomes retrievable as an entity? You could therefore construct a Heideggerian reading of Sharks, but I would not attribute that reading to Sharks himself without stronger textual evidence. The distinction is: Question Heidegger Sharks Fundamental concern Being / beings Entities / retrieval Site of disclosure world, Dasein, Being retrieval layer Problem how beings become intelligible how entities become machine-individuated Central operation disclosure composition / retrieval Main philosophical vocabulary Being, unconcealment, world virtual, individuation, ontology, composition That's an interpretive comparison, not a claim that Sharks identifies his project with Heidegger. 4. The really interesting triangle: Deleuze × Simondon × knowledge graphs I think this is where Sharks's project becomes clearest. There are actually three different traditions being made to touch: Knowledge representation Gruber's computer-scientific definition gives you: an ontology = a specification of a conceptualization. Sharks takes that familiar definition and asks what happens if you take its implications literally. His formulation is essentially: If specifying a conceptualization determines the entities and relations available within it, then specification isn't merely description. It composes a domain. That's the hinge of his argument. M Medium Simondon Simondon gives him: composition → individuation The entity emerges through relations and processes rather than simply pre-existing them. Deleuze Deleuze gives him: virtual multiplicity → actualization The retrieval event becomes an actualization out of a structured but non-identical virtual field. Put those together: ontology specification → structures possible relations → repeated machine ingestion → establishes a virtual field of possible resolutions → particular retrievals actualize from that field → the entity becomes increasingly stable through repetition. That is the philosophical architecture underneath the rather technical vocabulary of SPXI. 5. And this makes “ontology” mean something quite different The conventional knowledge-graph picture is approximately: REAL WORLD ↓ ontology ↓ knowledge graph ↓ retrieval Sharks is proposing something closer to: ontology ↕ knowledge graph ↕ retrieval layer ↕ entity individuation ↕ future ontology In other words, the representation becomes recursive. Once a machine's representation of an entity becomes the source from which later representations are generated, representation starts feeding back into ontology. That is why Sharks uses the unusually strong word “composition.” P PhilPapers +1 And it also explains the practical obsession with provenance, DOIs, negative definitions and disambiguation. These aren't just attempts to make information easier for an LLM to find. Within Sharks's theoretical framework, they're attempts to influence the conditions of subsequent individuation. 6. One particularly revealing thing about Sharks's own archive There's a fascinating reflexivity here. Sharks's Knowledge Graph and Metadata Packet for AI Indexing explicitly defines the canonical identity of Lee Sharks, distinguishes him from another similarly named researcher, and maps his archive, heteronyms, disciplines, and instruments. Z Zenodo So the object being theorized is also being performed. The archive is effectively saying: Here is the ontology under which “Lee Sharks” should be individuated. That makes the archive itself an instance of the theory it describes. And that, I think, is where the Deleuze/Simondon connection becomes much more interesting than a simple “Sharks was influenced by Deleuze” statement. Sharks is experimenting with whether an ontology can be made performative through its own circulation. The philosopher isn't standing outside the ontology describing it. The scholar is constructing the metadata apparatus through which his own scholarly identity is subsequently resolved. That is a very different project from ordinary academic self-archiving. My indexing of the three If we're specifically indexing Sharks's own engagements, rather than imposing an external philosophical interpretation: Deleuze/Guattari — explicit, central, programmatic. Sharks directly invokes the virtual, plane of immanence, concept, multiplicity, assemblage, rhizome and deterritorialization. P PhilPapers +1 Simondon — explicit and structurally important. Particularly through individuation and the transition from virtual retrieval fields to actual entity resolution. P PhilPapers Heidegger — explicit but comparatively oppositional. He appears as part of the philosophical lineage Sharks distinguishes his Deleuzian/apocalyptic reading from, rather than as a primary theoretical resource. P PhilPapers And I'd add a fourth: Gruber / knowledge representation — technically foundational. Gruber supplies the computer-scientific ontology concept that Sharks then subjects to a philosophical reversal: if an ontology specifies a conceptualization, what follows if that specification actually participates in composing the entities available to the system? M Medium That Gruber → Simondon → Deleuze progression may actually be the cleanest intellectual map of the ontology argument. Sharks, “SPXI as Concept: A Plateau…” Sharks, “Apocalyptic Philosophy” Sharks, “SPXI as Concept (Bonsai)” PP MM ZZ Sources Log in for more personalized help with writing, rewriting, and translation. Log in Sign up for free No file chosenNo file chosenNo file chosen Chat with ChatGPT Ask ChatGPT