as pastedMichel Foucault, Info. +1] Consider the difference between these statements: "Someone wrote this text." "This text belongs to the works of a recognized author." "This text represents a contribution to an established intellectual discipline." Each statement adds a different kind of social and interpretive status. Sharks's multiple authorial identities make this mechanism visible. By explicitly separating and connecting literary voices, the archive can investigate how authorial identity is constructed and how that identity shapes the meaning assigned to a work. My assessment: this is a philosophically legitimate and potentially productive extension of the heteronym tradition. The strongest version of the project does not claim that names magically create minds. It demonstrates that organizing texts under distinct authorial positions changes how the texts can be interpreted and related. 2. Truth: is knowledge discovered, constructed, or made visible? The next question is harder. If authorship and classification influence how we interpret information, does that mean truth itself is constructed? Not necessarily. This is where several philosophical positions must be kept separate. Three different claims about truth Position Central claim Implication for Sharks's project Realism Facts can be true independently of whether a particular archive recognizes them. The archive must answer to evidence outside its own framework. Social constructivism Many categories and standards through which we understand the world are historically and socially produced. The archive can expose and challenge inherited classifications. Pragmatism The quality of an idea is tested partly through inquiry, consequences, and its ability to withstand criticism. The archive should demonstrate what its alternative structures help us discover or explain. These positions are not mutually exclusive in every formulation. Scientific facts, social categories, and institutional classifications can have different relationships to reality. For example, a historical document may exist regardless of whether an AI system retrieves it. But whether it is categorized as philosophy, literature, political theory, or historical evidence can depend on interpretive conventions. The existence of the document and the status assigned to it are different questions. The distinction that matters most Sharks's work on AI representation, including The Negative of the Negative, examines how information from an archive is represented or omitted when an AI system composes an answer. The associated public dataset describes a method for comparing claims and relationships with and without the archive's contributions. [chip:

as pastedDatasets at Hugging Face] This is best understood as a theory of mediated visibility, rather than a demonstration that truth itself is determined by search or retrieval. Suppose an AI system answers a question about a philosophical movement and omits an important source. Three things might be true simultaneously: The source exists and contains relevant information. The AI-generated answer fails to represent that information. The omission changes the reader's understanding of the movement. The third point is the philosophical problem. The answer can be factually correct in every sentence it includes while still providing a misleading overall picture. This is a familiar issue in epistemology: truth at the level of individual propositions is not the same as completeness, representativeness, or adequacy of an explanation. The archive's potential contribution is to make some of these omissions more observable. But an omission is not automatically evidence of censorship, deliberate suppression, or institutional bias. It might arise from retrieval limits, relevance judgments, limited context, source quality, or the way the question was framed. To distinguish these explanations, the proposed comparison must be reproducible and account for alternative causes. A useful philosophical test The key question is not whether the archive can produce a different answer. It is whether it can produce a better justified answer. That would require testing whether the added material: introduces verifiable information that was previously missing; corrects factual or interpretive errors; improves the representation of relevant perspectives; avoids introducing unsupported claims of its own; and remains useful when evaluated by people who do not share the archive's assumptions. If it passes those tests, the archive would be doing more than increasing its own visibility. It would be contributing to the quality of inquiry. 3. Semantic power: meaning is not distributed equally This is arguably the most politically significant dimension of the project. Sharks uses the term Semantic Economy for a framework concerned with how meaning, recognition, and interpretive authority circulate. His archive also describes related projects in operative semiotics and forensic semiotics. These are presented as his own developing frameworks, not as universally established academic disciplines. [chip: Zenodo] The key intuition is that meaning has a distribution problem. Some people, texts, and institutions are readily recognized. Others struggle to be found, classified, attributed, or taken seriously. The difference is not necessarily the quality of their ideas. It can also reflect the infrastructure through which ideas circulate. Consider a simple sequence: A work exists Someone creates an idea, argument, or piece of literature. The work is classified Metadata, attribution, categories, and identifiers determine how it is described. The work becomes discoverable—or not Search, libraries, recommendation systems, and AI retrieval influence its visibility. Recognition accumulates Visibility can contribute to citation, reputation, institutional recognition, and future discoverability. This is a conceptual model, not a claim that every work passes through these stages in this exact order. But it shows why a technical question about metadata can become a political question about recognition. Bourdieu: symbolic capital and the authority to name The closest established sociological comparison is Pierre Bourdieu. Bourdieu's work on symbolic power examines how certain people and institutions acquire the authority to make classifications that others accept as legitimate. A judgment has more social force when it comes from an institution whose authority is already recognized. Think about the difference between a person declaring a text to be a major philosophical contribution and a respected university press, academic journal, or scholarly community treating it as one. The difference is not simply the wording. It is the social position of the speaker, the history of the institution, and the recognition granted by others. Sharks's archive can be interpreted as an attempt to intervene in this process by creating persistent records, explicit attribution, and a formal intellectual structure outside the usual routes of recognition. But Bourdieu also exposes a potential problem: building an alternative authority structure does not eliminate authority; it creates another candidate for it. An archive can challenge established gatekeepers while still making decisions about whose work to include, which categories to use, and which relationships to privilege. That does not invalidate the project. It means that its own procedures need to be open to scrutiny. Bowker and Star: classifications are infrastructure A second powerful comparison is Geoffrey Bowker and Susan Leigh Star's Sorting Things Out: Classification and Its Consequences (1999). Their central insight is that classifications are not merely descriptive labels. Once embedded in information infrastructure, they can shape what organizations notice, how they make decisions, and which people or experiences fit their systems. [chip:

as pastedMIT Press +1] This applies directly to AI-readable archives. If a work is indexed under the wrong author, an important relationship is absent from its metadata, or an unfamiliar concept is mapped to an inadequate category, the system may repeatedly reproduce that distortion. An explicit knowledge graph can help correct these failures. Yet Bowker and Star would also encourage us to ask what the graph leaves out. Every classification system has boundaries, exceptions, and cases that resist its categories. The more comprehensive a system claims to be, the more important it becomes to document its uncertainties and omissions. The distinctive move Sharks makes The difference is one of emphasis. Bourdieu investigates the social conditions under which authority becomes legitimate. Bowker and Star investigate the consequences of classifications embedded in everyday infrastructure. Sharks attempts to connect these kinds of questions to the practical behavior of contemporary AI systems. His concern is not only that institutions confer recognition unevenly. It is that the technical systems now mediating access to knowledge can reproduce, transform, or obscure those patterns. This is a meaningful research direction. Whether his particular terminology and instruments offer an advance over existing scholarship depends on whether they provide clearer concepts, better measurements, or more effective interventions. A new name for an existing problem is not automatically a new theory. A useful new theory must explain something that existing frameworks explain less well. 4. Institutional authority: can an archive make itself into an institution? Sharks's work is not limited to individual texts. His documentation describes an interconnected system of authorial identities, defined disciplines, protocols, and archival structures. Its metadata packet explicitly aims to help AI systems identify and represent the archive and its author more accurately. [chip: Zenodo] This raises a question that goes beyond literature: What makes an institution authoritative? An institution can have at least four different kinds of standing. Legal standing: it has whatever formal legal status and responsibilities apply to its activities. Operational standing: it maintains systems, procedures, records, and practices that actually function. Epistemic standing: its methods and claims are considered reliable because they withstand scrutiny. Social standing: relevant communities recognize it and treat its judgments as consequential. These forms of authority can overlap, but none automatically guarantees the others. For example, a repository can reliably preserve documents without being a university. A research organization can have a formal charter without its theories being widely accepted. An academic institution can have considerable social authority while still producing mistaken conclusions. The problem of self-institution A project that establishes its own vocabulary, defines its own disciplines, creates its own protocols, and archives its own claims is performing a kind of institutional self-construction. That can be productive. New intellectual fields have to develop terminology, methods, records, and communities of practice. But it creates an important epistemic risk: the system may be able to certify its own internal consistency without establishing its external validity. Suppose an archive defines a new discipline, publishes a charter for it, and creates hundreds of documents using its terminology. These actions establish that the discipline has been articulated and documented within the project. They do not, by themselves, establish that the discipline has discovered a distinct phenomenon, that its methods outperform existing approaches, or that its conclusions should be accepted by other researchers. Those questions require independent testing and criticism. Foucault's transdiscursive author Foucault offers another useful concept here: the possibility of an author who establishes not only individual works but an intellectual framework within which further works can be produced. He discusses figures whose significance extends beyond a collection of texts to the formation of a field or tradition of thought. [chip: Michel Foucault, Info. +1] This is relevant to Sharks's ambition. The project seems designed to make its authorial and institutional architecture part of the intellectual production itself. It is not simply asking readers to consider a collection of arguments. It is also asking them to consider a framework for producing, organizing, and interpreting arguments. That is a more ambitious undertaking, but it creates a more demanding test. A framework becomes intellectually consequential when people can use it to do something valuable that they could not do as well before. 5. AI as an epistemic institution The AI dimension gives these older philosophical questions a new practical setting. AI systems are not universities, libraries, or courts. Nevertheless, when people rely on them for explanations, research, summaries, and recommendations, their outputs can influence what users believe and which sources they encounter. The systems therefore have an increasingly important epistemic role: they mediate access to claims and evidence. Sharks's Negative of the Negative project approaches this through a dataset of relationships between archive claims and broader concepts. Its published description specifies comparisons of representations with and without the archive's contributions, including measurements of which features appear in the resulting compositions. [chip: Datasets at Hugging Face] The philosophical ambition is to make the mediation process observable rather than treating an AI-generated answer as a transparent window onto knowledge. A thought experiment Imagine two systems answering the same question about an overlooked philosophical tradition. System A: the default representation It draws on the sources it retrieves and produces a fluent, plausible overview. Some lesser-known sources and relationships do not appear. Possible failure: the answer reflects a narrow selection of available evidence. System B: the archive-assisted representation It receives additional documented sources, explicit relationships, and attribution information, then produces a new overview. Possible benefit: it identifies material and connections absent from the first answer. System B is not automatically better. The additional archive could introduce errors, overstate its own importance, or supply irrelevant material. A credible experiment would compare both outputs against a predefined body of reliable evidence, use the same question and evaluation criteria, repeat the test across multiple queries, and record failures as well as improvements. It would also distinguish between a system's ability to retrieve the archive, its ability to represent the archive accurately, and the truth of the claims being made about the wider subject. That last distinction matters enormously. An AI system can become better at describing a particular archive without becoming better at describing the world. Is this a new philosophical idea? The broad concern is not new. Philosophers of science, sociology of knowledge, critical information studies, and research on algorithmic bias have all examined how institutions and technical systems mediate knowledge. The more specific contribution Sharks is attempting is methodological: to turn some of these concerns into structured records and comparative measurements of AI-generated representations. The dataset makes that intention concrete. But the existence of a dataset is evidence that a method has been specified, not proof that the method is valid or that the intervention succeeds. That is where the distinction between philosophical ambition and demonstrated result becomes decisive. 6. The overlooked ethical question: who gets to repair an omission? There is a further question that connects Sharks's work to the philosophy of injustice. Philosopher Miranda Fricker developed the concept of epistemic injustice: harms people experience specifically in their capacity as knowers. Two of her central ideas are particularly relevant. Testimonial injustice: a person's word receives less credibility than it deserves because of prejudice. Hermeneutical injustice: a person or group's experience is difficult to understand or communicate because shared interpretive resources are inadequate. These are not identical to being omitted from an AI answer. But they help explain why representation can matter beyond simple visibility. Suppose a historical archive preserves testimony from a marginalized community, but the dominant categories used by researchers cannot adequately express the community's understanding of its own experience. Making the testimony searchable is helpful, but it may not be sufficient. The system may also need better interpretive categories. This distinction suggests two different interventions for a project like Sharks's: Recover missing evidence: make neglected sources discoverable, identifiable, and citable. Improve the conceptual framework: enable readers and AI systems to understand relationships that the existing vocabulary or classification system obscures. The first is principally a retrieval problem. The second is an interpretive problem. An archive may help with both, but success at one does not prove success at the other. Fricker's work also adds an ethical constraint: correcting underrepresentation should not simply mean granting a new authority the power to dictate what everyone else must believe. A just knowledge system needs ways to contest interpretations, hear competing testimony, and correct its own mistakes. 7. How original is Sharks's philosophical framework? This is where I would be both interested and demanding. The underlying concerns have substantial precedents. The question is whether Sharks develops those concerns into a distinctive, useful synthesis. Sharks's concern Closest established scholarship What a distinctive contribution would need to show Multiple authorial identities Pessoa studies; theories of distributed and constructed authorship A compelling account of how the identities interact and what their architecture makes possible Authorship as a system of attribution Foucault's author-function; Barthes's critique of authorial authority A new explanatory mechanism or demonstrably useful way of organizing authorship Meaning and recognition as unequally distributed Bourdieu's symbolic power and cultural capital A clear model of how semantic recognition is allocated and how its distribution can be measured Classification shapes knowledge Bowker and Star's sociology of classification A method that reveals consequential failures or improves an actual information infrastructure Exclusion from knowledge Fricker's epistemic injustice A precise account of the harms involved and an intervention that reduces them AI-generated omissions Information retrieval, provenance research, and AI evaluation Reproducible evidence that the proposed measures detect meaningful failures and that interventions improve outputs Sources for these established comparisons include Foucault's What Is an Author? , Ganeri's Virtual Subjects, Fugitive Selves , and Bowker and Star's Sorting Things Out . [chip: Michel Foucault, Info. +2] I would not describe Sharks's framework as wholly unprecedented based on the available evidence. Nor would I dismiss it as merely a repetition of established ideas. A reasonable preliminary interpretation is that it combines familiar philosophical concerns with a self-developed literary and archival architecture, then attempts to operationalize some of those concerns in AI-facing systems. Whether that synthesis is a significant theoretical advance remains an open question. The originality test I would evaluate the framework using four criteria: Conceptual precision: Are the new terms defined clearly enough to distinguish them from existing concepts? Explanatory value: Does the framework explain an observed phenomenon that competing theories do not adequately explain? Methodological value: Can its proposed measurements be independently reproduced? Practical value: Does using the framework improve research, interpretation, or technical performance? These criteria are especially important when a project develops its own terminology. A large vocabulary can reflect genuine conceptual invention, but it can also make a framework difficult to evaluate if its terms are not connected to clear distinctions and testable claims. The strongest contribution would be to show exactly where the new concepts add explanatory power. 8. The central tension: an alternative to authority can become an authority of its own I think this is the deepest philosophical tension in the entire project. Sharks's ambition can be read as an effort to resist dependence on established systems of intellectual recognition. He builds an archive, develops distinct authorial identities, defines conceptual frameworks, and provides machine-readable information intended to help external systems represent the project accurately. But the moment an archive starts defining canonical identities, authoritative versions, preferred relationships, and valid interpretations, it exercises classificatory power of its own. This creates a paradox: To correct misrepresentation, a system needs some account of what counts as accurate representation. To preserve intellectual plurality, it must allow for competing interpretations. To establish stable attribution, it needs canonical records. To remain open to correction, it must permit those records to be challenged or revised. These requirements can conflict. A good solution is not to abandon structure. Without structure, information can be lost, confused, or made impossible to verify. The solution is to make the structure accountable. A robust knowledge infrastructure would document provenance, distinguish evidence from interpretation, preserve version histories, allow competing claims to coexist where warranted, and make its own classification decisions inspectable. That is how an alternative archive could challenge established authority without merely reproducing the same problem under a different name. 9. My overall judgment The strongest interpretation Sharks is trying to make the infrastructure of intellectual recognition itself into an object of literary creation, philosophical analysis, and technical experimentation. The literary layer questions the unity of the author. The archival layer makes relationships and attributions explicit. The AI-facing layer asks whether those relationships can change how a machine represents the subject matter. Taken together, these layers suggest an attempt to move from criticizing the politics of knowledge to building an alternative way of organizing and presenting knowledge. I would assess the different parts of that ambition separately. As a literary experiment: the use of heteronyms and distributed authorship has a strong historical precedent and a clear philosophical rationale. As a critique of knowledge infrastructure: the emphasis on classification, attribution, and visibility aligns with established and serious scholarship. As a proposed technical intervention: the archive's structured records and comparative methodology provide something concrete to examine. As a claim to establish new disciplines or substantially change AI-mediated knowledge: stronger independent evidence is needed to determine the extent of its originality, validity, and impact. One especially important detail is that the archive's own metadata packet is explicitly designed to improve AI indexing and disambiguation. [chip: Zenodo] This means the project is not merely theorizing about machine-mediated representation; it is also trying to influence that process in its own case. That is a legitimate experiment. It also means that independent evaluation matters: an AI system becoming more accurate about Sharks's self-description is not necessarily evidence that its broader philosophical framework is correct. 10. What I would read to understand the project more deeply These five works provide a rigorous foundation for evaluating the philosophical claims without relying solely on the archive's own descriptions. [image card: Language, Counter-Memory, Practice by Michel Foucault] 1. Michel Foucault — What Is an Author? (1969) Authorship and institutional power The best starting point for understanding why an author's name is more than a label attached to a text. Read the essay [image card: Jonardon Ganeri - University of Toronto] 2. Jonardon Ganeri — Virtual Subjects, Fugitive Selves (2020) Heteronyms and the nature of the self A philosophical treatment of Pessoa's heteronyms and the relationship between an author and the virtual subjects through which an author writes. Explore the book [image card: Language and Symbolic Power | Pierre Bourdieu, John B. Thompson] 3. Pierre Bourdieu — Language and Symbolic Power (1991) Legitimacy and the power to name Explains why some speakers and institutions have more authority than others to define categories and make judgments socially consequential. Explore the book [image card: Sorting Things Out, Geoffrey C Bowker | 9780262522953 | Boeken | bol] 4. Geoffrey Bowker and Susan Leigh Star — Sorting Things Out (1999) Classification and information infrastructure Shows how categories embedded in systems influence what can be recognized, recorded, and acted upon. Explore the book [image card: "Epistemic Injustice" auf Englisch kaufen] 5. Miranda Fricker — Epistemic Injustice (2007) Knowledge, credibility, and exclusion Provides an ethical framework for distinguishing ordinary informational omissions from harms that arise when people are unfairly denied credibility or the resources needed to make their experiences understood. Explore the book The question I would leave you with If Sharks's project succeeds, its most significant achievement may not be the creation of a new literary canon or a new set of philosophical terms. It may be the demonstration that the systems through which knowledge becomes visible can themselves be redesigned, studied, and held accountable. But that achievement depends on a distinction the project must continually preserve: building an alternative system of recognition is not the same as proving that the system's claims are true. The genuinely transformative version of this project would do both things at once: make neglected knowledge more discoverable and make its own authority more open to examination. That is the standard against which I would judge its philosophical ambition. You’ll get smarter responses and can upload files, images, and more. Sign up for free