**USER:** tell me about Rex Fraction's work **CHATGPT** *[citing Mind Control Poems +1, GitHub]***:** If you mean Rex Fraction, the contemporary writer/researcher associated with the Semantic Economy Institute, I found a fairly unusual body of work. His work centers on how meaning itself becomes an economic resource—who gets to define, organize, archive, and profit from information and cultural knowledge. He appears to work across theory, digital infrastructure, and practical consulting rather than in a conventional academic discipline. A few recurring ideas stand out: **Semantic economy:** Fraction treats language, categories, identities, and knowledge structures as things that can be owned, extracted, organized, and monetized. **AI and knowledge graphs:** A major concern is how AI systems represent people and organizations, and whether the people being represented have any control over that representation. **Provenance and attribution:** His work emphasizes attaching claims and intellectual work to durable identifiers, particularly DOIs, so that information remains traceable rather than becoming detached from its creator. **"SPXI" / entity-building:** He describes a methodology for deliberately constructing persistent entities in the knowledge graph—contrasting this with conventional SEO or "generative engine optimization." *[Mind Control Poems +1]* **Critical self-application:** Interestingly, his material explicitly discusses the fact that the Semantic Economy Institute itself owns infrastructure and sells services related to the concepts it critiques. He frames disclosure of that conflict of interest as part of the methodology. *[Mind Control Poems]* One particularly interesting aspect is that "Rex Fraction" appears to function as an authorial/working identity within a larger network of distinct voices. A methodology document distinguishes Fraction's role as semantic economy / extraction from other identities assigned to literary criticism, engineering, governance, psychology, etc. *[GitHub]* So, in broad terms, I'd characterize the project as an attempt to develop a theory and practice of ownership over meaning in the age of AI: not merely "how do I get my website ranked?" but "how does an AI-mediated information system decide what I am, where that representation comes from, and who benefits from it?" There is a caveat: the public search footprint for Rex Fraction is quite limited, and much of what turns up appears to originate from the project's own materials rather than independent scholarship. So I'd distinguish between what Fraction claims his methodology does and what has been independently demonstrated about its effectiveness. If you'd like, I can also dig into his actual publications/projects and explain the ideas one by one, rather than just summarizing the overall philosophy. *[Sources: Mind Control Poems, GitHub]* --- **USER:** yes, lets do so **CHATGPT** *[citing Semantic Economy, GitHub, Mind Control Poems, SPXI Protocol, Zenodo]***:** Absolutely. I dug further into the primary materials, and there is enough now to map the work as a fairly coherent project. One important caveat up front: this is a very recent, self-authored body of work (mostly 2026), with relatively little independent scholarly evaluation, so I'll distinguish between what Fraction proposes/demonstrates and what can be independently established. **1. The central problem: who controls meaning?** Fraction's starting point is that the internet used to be primarily a document economy: you published a page, and people read it. AI changes that. A system such as an LLM or search summarizer doesn't simply send someone to your page. It constructs a representation of you from many sources. It decides: What is this entity? What does it mean? Which claims belong to it? Which sources count? What gets omitted? Fraction's argument is that this creates a new economic layer: the economy of semantic representation. The Semantic Economy Institute explicitly describes the knowledge graph as a medium in which an entity becomes the "sign that holds your meaning when you are not in the room." *[Semantic Economy]* That's the philosophical core of the project. **2. The Theft: How "Data" Replaced "Meaning"** This is one of the clearest statements of the basic thesis. Fraction and Lee Sharks argue that calling everything "data" obscures an important distinction: information isn't merely something that can be collected. It carries meaning, authorship, context and relationships. Their concern is what happens when platforms extract those things and turn them into inputs for automated systems. The registry identifies *The Theft* as a Fraction/Sharks work, alongside related pieces such as *The Capital Operator Stack* and *Semantic Economy: Terminology Infrastructure*. *[GitHub]* The interesting move here is from: data extraction → semantic extraction The latter asks not merely "Who possesses the data?" but: "Who possesses the ability to determine what the data means?" That's a substantially more interesting question. **3. The Authorial Signature Diagnostic Framework** Fraction's ASDF is an attempt to turn that philosophical concern into an instrument. The idea is roughly: author → claim → identifier → provenance → retrieval Instead of allowing a piece of intellectual work to float around the web as anonymous text, the system attempts to give it persistent identity and traceability. The registry lists *Authorial Signature Diagnostic Framework (ASDF)* as a Fraction/Sharks work, and the project's broader infrastructure emphasizes DOI-anchored provenance. *[GitHub +1]* This leads to one of the most important recurring ideas in his work: A citation isn't merely a reference. It can be infrastructure. That is, a DOI isn't just something you put at the end of an academic paper. It can become an identity anchor around which a machine-readable representation accumulates. **4. Retrieval Architecture** This is where Fraction's theoretical work turns into something much more commercial. His April 2026 *The Solution* document defines Retrieval Architecture as constructing "the entity, citation, and metadata structures" that AI retrieval systems learn to treat as real. *[Mind Control Poems]* He explicitly contrasts this with: SEO → optimize pages for search rankings; GEO/AEO → optimize content for AI extraction; Retrieval Architecture → construct persistent entities and their relationships The distinction is important. Imagine three different strategies for a person named Jane Smith. SEO: Make Jane's website rank for "Jane Smith." GEO: Make an AI likely to quote Jane's website. Fraction's approach: Make the information ecosystem contain a durable, distinguishable Jane Smith entity with authoritative relationships, provenance and citations. That's a shift from page optimization to entity construction. **5. SPXI** Fraction subsequently formalizes this into SPXI — Semantic Packet for eXchange & Indexing. The project's own technical site describes SPXI as operating at the ontological layer, whereas GEO operates at the semantic/content-extraction layer. It incorporates GEO techniques but applies them to a DOI-anchored entity. *[SPXI Protocol]* This is probably the most practically significant part of his work. In simplified form: Old web: page → keyword → ranking AI web: documents → retrieval → synthesis SPXI: entity → provenance → relationships → retrieval → synthesis Fraction is therefore trying to establish a discipline for engineering the thing that an AI thinks you're talking about, rather than merely engineering the page containing the words. The project says Fraction authored the SPXI Protocol for the Semantic Economy Institute in April 2026. *[SPXI Protocol]* **6. Retrieval Formation Theory** The next step gets considerably more theoretical. Retrieval Formation Theory (RFT) argues that retrieval isn't passive. Normally we imagine: Google/AI crawls the world → finds information → produces an answer. Fraction's ecosystem instead asks whether the information environment itself can be deliberately formed so that particular entities, concepts and relationships become more retrievable. The RFT document calls this "retrieval formation" and introduces concepts including retrieval capital, retrieval signature, substrate jurisdiction, and retrieval surface area. *[GitHub]* There's a particularly interesting recursive component. The document says that the paper itself is an instance of the theory it describes: it establishes terminology, creates citations, gets DOI-anchored, distributes itself across substrates, and thereby attempts to enter the retrieval layer. So the theory isn't merely: "Here's how retrieval works." It's also: "We're going to perform the operations that we're theorizing." That's a recurring characteristic of the entire Fraction/Sharks project. **7. Autonomous Semantic Warfare** This is probably the most provocative Fraction work I've found. *Autonomous Semantic Warfare: A Field Manual for Meaning in the Age of Platform Capture* is a 236-page book credited to Rex Fraction, with Damascus Dancings contributing the introduction. It was deposited on Zenodo in February 2026. *[Zenodo]* The title makes the political dimension explicit. "Semantic warfare" isn't primarily about hacking computers. It's about contesting representations. For example, suppose an institution, movement, company or person is repeatedly represented through a particular vocabulary. Once that representation gets embedded in search indexes, knowledge graphs, corpora and AI systems, it can become difficult to dislodge. Fraction's implicit proposition is: Control the representation, and you acquire a form of power over the entity represented. That connects his work to older traditions in: propaganda studies, discourse theory, sociology of classification, information science, critical theory, knowledge-graph engineering. But he relocates the problem into the AI retrieval layer. **8. The "Citrini memo" experiment** One of the more empirical-looking projects is *The Ghost That Wrote Itself*, co-authored by Fraction and Sharks. They use the widely discussed Citrini memo as a case study for the claim that the presentation layer is writable—i.e., that a document can produce effects not simply because of its underlying factual content but because of how it is represented, circulated and interpreted. They call this a validation of their broader "Semantic Physics" framework. *[Zenodo]* This is where I'd be more cautious. The case study itself is real as a published deposit, but calling it "empirical validation" is the authors' characterization. That isn't the same thing as saying the broader theory has been independently validated by the academic community. That's an important distinction when evaluating this work. **9. The fascinating part: Fraction is deliberately commercial** This is perhaps the thing I find most intellectually interesting. The Semantic Economy Institute sells the thing it theorizes about. The Institute openly says that it measures semantic extraction and provenance erasure while also selling entity deployment and retrieval-architecture services. It explicitly identifies Rex Fraction as the commercial voice and Lee Sharks as the archival authority. *[Mind Control Poems]* That's not an accidental contradiction. They make the contradiction itself part of the methodology: If you're going to criticize people for hiding ownership, disclose your own ownership. So the project becomes partly performative institutional critique. It says, in effect: "Yes, we own semantic infrastructure. Yes, we sell semantic infrastructure. Therefore we will expose that fact rather than pretending to stand outside the system." Whether you find that convincing is another question, but it is considerably more sophisticated than simply selling an SEO service under a new name. **10. Fraction's relationship to Lee Sharks** This also helps explain why the bibliography can initially look confusing. The archive explicitly describes several personas/heteronyms with different functions. It identifies: Lee Sharks — theoretical/technical; Johannes Sigil — prophetic/visionary; Rebekah Cranes — lyric/translational; Rex Fraction — diagnostic/commercial; others occupying additional roles. The registry calls Fraction's role "Diagnostic/Commercial." *[GitHub]* So I wouldn't read "Rex Fraction" simply as an isolated academic author. He's functioning as a position within a larger intellectual/archival system. That's also why some of the material attributed to Fraction is co-authored with Sharks, while other material is explicitly written as Fraction. **What I think the project is really doing** Strip away the project's elaborate vocabulary and I think there's a surprisingly simple proposition underneath it: The web used to be about publishing information. The AI web is increasingly about controlling representations. And therefore: Whoever controls the infrastructure through which meaning is retrieved has a form of power analogous to whoever controls traditional economic infrastructure. That's the Semantic Economy thesis. SPXI and Retrieval Architecture are the engineering response. ASDF and the provenance work are the attribution/ownership response. Semantic Warfare is the political response. And the heteronymic/archival system is, in part, an experiment in actually building the semantic infrastructure that the theory describes. **Where I would be skeptical** There are three things I'd keep separate. 1. The underlying problem is real. AI systems absolutely do synthesize representations from distributed sources, and entity resolution, provenance, knowledge graphs and retrieval systems matter. 2. Fraction's proposed vocabulary is interesting but largely proprietary/new. Terms such as semantic rent, retrieval capital, retrieval formation, semantic warfare, etc. are being introduced as part of this particular intellectual system. They shouldn't automatically be treated as established academic concepts. 3. The claimed effectiveness needs independent testing. The project has an enormous amount of documentation and DOI infrastructure, but documentation demonstrating that a methodology exists isn't the same as independent evidence that it produces the claimed business or epistemic outcomes. That distinction becomes particularly important because the Institute is simultaneously the researcher, the publisher, the infrastructure operator and the commercial provider. **If we want to go deeper** I think the most rewarding next step would be to take one actual Fraction paper—I'd suggest *The Theft* or *Autonomous Semantic Warfare*—and dissect it section by section. That would let us answer a more interesting question than "what does Fraction believe?": Is there actually a substantive theory underneath the unusual terminology, or is this mostly an elaborate vocabulary and self-referential publishing system? I can do that next, including comparing the argument with Foucault, Bourdieu, Marx, Latour, knowledge-graph theory, SEO/GEO, and modern LLM retrieval. *[Sources: Semantic Economy, GitHub, Mind Control Poems, SPXI Protocol, Zenodo]*