Moreno Nourizadeh

In 2025, a group of researchers in Rome rewrote a corpus of harmful prompts as poems and tested them against twenty-five frontier language models from nine providers; the models, which had been trained at considerable expense to refuse such prompts, complied with the poetic versions at an average rate of about 62%, while the prose baseline ran at about 8% (Bisconti et al., 2025). The empirical finding, taken in isolation, would be one result among many. What makes it the subject of a paper is that, on every assumption the engineering of safety controls had been making, the result is one that should not have been possible. For the controls that the field calls guardrails, in the way it calls firewalls and password checks and type systems by the same family of names, were imagined to function as exclusions, and an exclusion is by its nature insensitive to the form of the input it excludes; a password checker is not softened by metaphor, a compiler does not relax its rules when the input rhymes, and a cryptographic hash takes no notice of whether its passphrase has been cast as a sonnet (Pierce, 2002; Sipser, 2012). A guardrail of the same kind should have behaved similarly, and in the present case it did not. The paper argues that the cause of the disparity lies in a category mismatch between the kind of system safety has been imagined to be and the kind of system it actually is, when the system in question is a language model. A language model is sensitive to the form of its inputs; it is fitted to that form, since reading context, following genre, and continuing the activity it has been handed is what such a system does and what makes it useful. The trained refusal, then, operates as a learned tendency in a region of ordinary language, where the engineering imagination had placed an exclusion, while the medium through which the refusal travels remains open to indefinitely many other routes (Wittgenstein, 1953/2001; Bakhtin, 1986; Ouyang et al., 2022). The poem is one such route, the folktale another (Bisconti et al., 2026), and the catalogue is open because ordinary language is open. The implication for deployment, which the closing sections develop, is that the controls that must hold cannot be installed in the medium they govern, and must instead occur in a deterministic application layer indifferent to genre (Clarke, Grumberg and Peled, 1999; Baier and Katoen, 2008; Seshia, Sadigh and Sastry, 2022).

The systems the current AI industry sells are sold under three promises: that they understand the world, that they are safe enough to be deployed at scale, and that they learn how the world works from watching it. This paper argues that each promise meets a limit the industry's own resources cannot remove, that the three limits are independent of each other, and that the systems now in deployment meet all three limits at once. The first limit says that a system that has only seen the world through a fixed body of data cannot vouch for the cases that lie outside that data, and adding more data or more compute does not close the gap. The second limit says that a system whose probability of a serious failure on each action does not fall in response to its own serious failures will, deployed long enough, produce them without end, no matter how small the per-action probability is. The third limit says that an agent acting on a system that recovers slowly, an institution, a supply chain, an ecology, cannot tell a healthy state from a depleting one when the agent only ever sees the short run, and ends up spending what it cannot see. The three are established from sources already canonical in the relevant fields and the limits are then turned back on the field's own descriptions. The world-model programme is the central case, because two of the three limits strike it independently. For each limit the paper states the condition under which the limit would be false.

This paper develops one technical implementation of a hypothesis about the cosmological constant problem: that the quantity which gravitates in the vacuum may be the derivative form of an irreversible residue rather than the magnitude of a zero-point energy sum. The implementation is a scalar-tensor theory with non-minimal coupling F(chi)R, a small exponential potential, and a conformal matter coupling A-squared = F that places ordinary matter on the Einstein-frame geodesics. Two routes are examined. The first sets the potential to zero to preserve the strict claim that no vacuum-energy magnitude couples to gravity; both an analytic argument and explicit numerical integration show it gives no late-time acceleration, so it is ruled out. The second adds the small potential and the natural matter coupling, which in the Einstein frame reduces the model to canonical exponential quintessence with no fifth force, so Solar System bounds are met automatically without chameleon or Vainshtein screening. A full-likelihood background fit against DESI DR2 BAO, the complete 1701-supernova Pantheon+ sample with its statistical-plus-systematic covariance, and a local Hubble prior, with the sound horizon marginalised, gives chi-squared per degree of freedom of 1.23 and a present-day equation of state w0 = -0.955 +/- 0.038, on the thawing side of minus one, with no posterior sample crossing the phantom boundary. The model reproduces Lambda-CDM expansion to within a percent out to redshift two. Because a minimally coupled scalar cannot cross w = -1, the paper states an explicit falsification condition: a phantom crossing established by parametrisation-independent methods excludes the exact-coupling model, and a degeneracy-violating extension specifies in advance what such a crossing would measure. The cost is a single hidden-sector small parameter set at the observed dark-energy scale, which the paper relocates rather than derives, moving the fine-tuning out of the ultraviolet-divergent vacuum sum and identifying the resulting 10-to-the-120 hierarchy with the Gibbons-Hawking entropy of the de Sitter horizon. Three partial treatments support the reframing: a thermodynamic dimensional check that produces the dark-energy scale from an infrared entropy-production rate, an argument from the GENERIC formalism of non-equilibrium thermodynamics that selects the A-squared = F coupling, and a Schwinger-Keldysh roadmap for deriving the residue field from quantum field theory in curved spacetime. The contribution is the argument for why the model takes this form, rather than its observational predictions.

The Conditions of the Invisible Hand is a diagnostic examination of the libertarian economic tradition conducted from inside the tradition itself. Working from Smith, Ricardo, and Hayek as its principal figures, alongside Bastiat, Mises, Kirzner, Friedman, Sowell, Rothbard, and Rand, the monograph extracts thirty-one conditions the tradition’s canonical sources specify as constitutive of a real market, then examines the two economies the tradition’s contemporary defenders treat as its institutional realisations: the Western order over the four decades since the late 1970s, and the Chinese order across the same period. The monograph runs to approximately 163,000 words across twelve chapters and four parts plus a detachable Epilogue, with a bibliography of approximately 600 references. ━━━━━━━━━━━━━━━━━━━━ PART I — THE EMPIRICAL RECORD ━━━━━━━━━━━━━━━━━━━━ The thirty-one conditions are reconstructed from the canonical sources at the level of formal requirement, then applied to the two orders. The Western order fails twelve of the conditions sharply, with the failures clustering in trade theory, financial economics, institutional architecture, the analytical framework itself, and competitive structure. The Chinese order satisfies the three Ricardian conditions through institutional architectures the libertarian tradition has historically opposed, while the Western order does not. The chapter that registers this second finding names the pattern the Ricardian inversion. ━━━━━━━━━━━━━━━━━━━━ PART II — WHY MORE TOOLS ARE NEEDED ━━━━━━━━━━━━━━━━━━━━ The tradition’s own analytical vocabulary registers the empirical record with difficulty, and the difficulty itself becomes a feature of the record requiring explanation. Part II constructs the additional apparatus by drawing on cybernetics in the lineage of Ashby, Conant, and Beer, and on the analysis of technique developed by Jacques Ellul. Each apparatus is justified operationally: it is admissible to the extent that it makes a feature of the orders visible that the existing vocabulary cannot, and dispensable to the extent that it does not. The cybernetic apparatus produces the Horizon-Mismatch Theorem: when a regulator’s action-cycle is much shorter than the substrate’s replenishment-cycle, substrate-coupled regulation is impossible by construction. ━━━━━━━━━━━━━━━━━━━━ PART III — THE SECOND-PASS TEST ━━━━━━━━━━━━━━━━━━━━ The two orders are re-examined with the apparatus Part II constructed. The structural-mechanism findings the first-pass examination could not produce now appear: the twelve Western failures are not contingent accidents but the architecture’s necessary signature, and the Chinese order’s satisfaction of the Ricardian conditions is the same theorem operating with a different cycle-coupling. The same mechanism produces opposite findings in the two orders. ━━━━━━━━━━━━━━━━━━━━ PART IV — THE DIAGNOSTIC FINDINGS ━━━━━━━━━━━━━━━━━━━━ Three findings emerge. The contemporary Western price-formation system is a casino in a non-normative, technical sense: the dominant variance in the price signal comes from the system’s own internal dynamics rather than from the real economy, and Hayek’s argument required the opposite. The cannibalisation cascade operates across eight productive substrates the contemporary order depends on, with one mechanism producing one signature at each substrate. The Chinese order operates the same underlying cybernetic apparatus as the Western order with one architectural difference: the central regulator retains standing authority to override the order’s internal price signals when those signals contradict substrate-protection targets. The final chapter walks the libertarian tradition’s three contemporary defensive positions against the cumulative record and closes them on the tradition’s own resources, and registers two further findings: the regulator the discourse names and the regulator the cumulative record specifies differ at the cybernetic level, and the three-lock model of agent-amendability specifies the conjoint conditions under which an architecture is amendable at the architectural-correction level.

This paper investigates whether the impossibility of communication with radically alien intelligence is technological, a problem of insufficient signal strength, encoding ingenuity, or detection sensitivity, or whether it runs deeper, into the ontological conditions under which communication, meaning, and recognition are constituted. The dominant assumption in the search for extraterrestrial intelligence holds that mathematics constitutes a universal language decodable by any sufficiently advanced cognition. This assumption rests on a series of philosophical commitments, commitments about the nature of mathematical truth, the structure of perception, the ontology of meaning, and the conditions of communication, that have been systematically challenged across the history of twentieth-century thought but whose refutation has not penetrated the scientific programmes that presuppose them. Through sustained engagement with Wittgenstein's concept of form of life, Merleau-Ponty's phenomenology of embodied perception, Lakoff and Nunez's cognitive science of mathematical cognition, Heidegger's analysis of Zuhandenheit and the constitutive role of language in the disclosure of being, Husserl's theory of perspectival manifestation, and Stiegler's concept of originary technicity, this paper traces what happens when the universalist assumption is subjected to the scrutiny these traditions provide. The investigation proceeds through the embodied foundations of mathematical thought, the dimensional architecture of cognition, the conditions under which language operates as an intranet of shared meaning, the ontological incoherence of the view from nowhere, the status of mathematics as a language game whose rules presuppose a shared form of life, and the question of whether the distinction between signal and noise is discovered in phenomena or drawn by the practices that classify them. What emerges from this investigation, and how far the consequences reach, the reader is invited to discover in the argument that follows.

A language model tokenises text, computes attention-weighted representations through 96 layers of transformation, and generates output one token at a time by selecting the most probable continuation from distributions learned through gradient descent on hundreds of billions of tokens of human writing. The result: systems that write legal briefs, debug code, translate between languages, sustain philosophical argument, and pass professional examinations, without understanding a single word they produce. The engineering is fully documented. The mathematics is precise. The training procedures are reproducible. An engineer can trace every operation from input to output. And yet the engineers who build these systems cannot explain why the conjunction of next-token prediction and sufficient scale produces outputs that pass for comprehension, reasoning, and creativity. The how is known. The why is not. The two questions have never been addressed together, in the same investigation, at the level of precision that each demands. This paper addresses both through four registers. Part I presents the engineering through concrete explanation, analogy, and step-by-step description, for the reader with no technical background. Part II presents the same architecture at mathematical precision: the attention equation, the training objective, the scaling laws, the sampling strategies, for the engineer who wants to verify every claim against the published literature. Part III conducts a philosophical investigation into why the engineering works, testing six frameworks in the philosophy of language (Saussure, Derrida, Wittgenstein, Quine, Austin, Davidson) against the empirical fact of the models' success. Part IV draws out the operational implications for the claims currently being made about these systems: the scaling hypothesis, the nature of hallucination, the "cognitive abundance" thesis, and the question of what these systems can and cannot become. Each register is self-sufficient. A reader may enter at any point. The cumulative finding: language produces distributional regularities that are learnable by exposure and resistant to formalisation. The models succeed by approximating the traces that language leaves behind. The traces are rich, structured, and informationally dense. They are not the meaning that produced them. Understanding this distinction, across all four registers simultaneously, is the only adequate response to the confusion that currently dominates public discourse about artificial intelligence.

The people who build these systems do not know why they work. “For all its runaway success, nobody knows exactly how, or why, it works,” reports the MIT Technology Review (Heaven, 2024), summarising the state of knowledge about deep learning. Boaz Barak, a computer scientist at Harvard seconded to OpenAI, likens the situation to physics at the start of the twentieth century, a pile of experimental results that surprise the experimenters and resist explanation. Anthropic, which built Claude, states plainly in its own research that it does not understand how its models do most of what they do. Hattie Zhou, an AI researcher at Montreal, recalls arriving in the field, asking her teachers why the systems work, and being told there were no good answers. The admission is genuine, and the response to it has been to treat the gap as an engineering puzzle awaiting more interpretability tools and more scale. This paper treats it as what it is, a question about language rather than about machines, and defends an answer the field does not want to hear: the model understands nothing, and it produces fluent text only because it harnesses a structure already present in human writing, the residue left in text by the millions of people who did understand. The empirical surprise that something so simple as next-token prediction can generate publishable prose is a fact about language, not about the machine. The achievement of the engineers in building such systems is real and not contested here. The question this paper takes up is narrower: why does the operation work at all? The case is built from the ground up, assuming no prior acquaintance with the philosophy of language and teaching each idea from ordinary examples before naming the thinker who found it. It converges on a single property of language that explains, at once, why fifty years of rule-based artificial intelligence came to nothing and why statistical approximation works: language is regular without being systematic, caught by patterns yet beyond the reach of any rules.

The Aharonov-Bohm effect has generated a sixty-year debate between two interpretive camps: one holding that electromagnetic potentials are physically real, the other that electromagnetic fields act non-locally. This paper argues that the debate is structurally analogous to the paradigmatic seams identified across ten cases in the philosophy of physics and mathematics, where a framework generates phenomena it cannot accommodate because its enabling conditions have been excluded from its ontology. The enabling condition of electromagnetism, on the analysis presented here, is neither the potential nor the field but the connection on the electromagnetic fibre bundle, a relational structure from which both potential and field are derived as partial, regime-dependent articulations. The potential is the connection expressed in coordinates (hence non-unique). The field is the connection’s curvature (hence informationally incomplete: it loses the holonomy). The Aharonov-Bohm effect is the diagnostic seam where this incompleteness forces itself into experimental visibility. The connection’s withdrawal from local determination, manifested as gauge freedom, is not a defect of the formalism but the formal signature of an enabling condition that must remain operative but non-thematic for the framework to function. This diagnosis reveals that the Aharonov-Bohm phase is not a piece of information carried by the electron but a constitutive transformation: the electron is changed by its passage through the topology of the connection in ways that are non-local, non-reversible, non-extractable, and substrate-dependent. The effect is metabolic, not computational, in a sense made precise through four formal criteria derived from the gauge-theoretic formalism. This diagnosis dissolves the binary between the two camps by identifying their shared presupposition: that the fundamental entity must be a locally determined quantity. It does not propose new physics. It applies a structural diagnostic, developed and tested across independent domains, to an experimentally confirmed quantum phenomenon whose interpretation remains contested.

Between the 7th and 15th centuries, European artists produced thousands of depictions of animals they had never seen. Lions acquired human faces. Elephants sprouted hooves and trumpet-shaped trunks. Whales became enormous fish. The standard explanation treats these as errors born of ignorance. This paper demonstrates that the explanation is inadequate, and that the inadequacy reveals a permanent epistemological condition. The distortions were systematic, following the logic of the available representational categories rather than the shape of the animal. Three structural features define the phenomenon: schema substitution (the forced assimilation of unarticulable phenomena to the nearest available grammar), lateral propagation (the self-replication of the schema through manuscript copying, with increasing internal coherence and increasing distance from the referent), and resistance to empirical correction (the tradition's capacity to reabsorb direct observation without structural change). Through detailed examination of the textual genealogy from the Physiologus through Isidore of Seville to the English bestiary families, and through close analysis of the Matthew Paris elephant drawings as a test case, the paper establishes a general mechanism of paradigmatic mis-seeing: what happens when any framework must render phenomena for which its categories are underdetermined. Drawing on Bachelard's epistemology of the obstacle, Wittgenstein's analysis of hinge propositions in On Certainty, Hamann's metacritique of Kant, and detailed parallel cases from cartography, oral poetry, and the history of writing technologies, the paper argues that the categories through which a framework operates are the conditions under which the world appears at all, and that the bestiary tradition provides a legible case of this ontological structure. The bestiary case is methodologically indispensable because it is a completed experiment: a case where history has supplied the successor categories and rendered the distortions transparent. It allows us to study, from outside, the structure of paradigmatic mis-seeing that remains invisible from inside any currently operative framework.

The cosmological constant problem persists because all standard approaches assume that vacuum energy couples to spacetime curvature via its magnitude. This paper proposes that the coupling assumption itself may be incomplete: if the vacuum possesses an irreducible irreversible sector alongside its energetic content, then the geometrically relevant quantity need not be the magnitude of the vacuum's zero-point contributions but the derivative structure of a phenomenological residue field R. Physics already admits rigorously structured systems in which reversible and irreversible contributions are generated by distinct operators subject to mutual degeneracy constraints; the GENERIC formalism provides the established example. The present paper does not claim the quantum vacuum is a GENERIC system. It proposes, by analogy with that formal precedent, that vacuum gravitation may be mis-specified because physics has conflated sector magnitude with geometrically relevant derivative structure. We introduce R as a phenomenological scalar field and couple its Hessian to the Einstein tensor. The trace of the modified equation yields a scalar curvature contribution proportional to the d'Alembertian of R, which is naturally small in a homogeneous universe regardless of the magnitude of R. The Bianchi identity yields a consistency condition for R involving a Ricci-tensor correction. The framework predicts deviation from strictly constant dark energy, consistent with DESI DR2 results. A provisional effective action is sketched from which the field equation can in principle be derived.

This paper demonstrates that the foundational contradictions afflicting contemporary physics, cognitive science, and artificial intelligence share a common structure: each paradigm assumes discrete, separable, local structures but produces holistic, relational, global phenomena it cannot explain. We establish this thesis through four convergent lines of inquiry. First, we prove via spectral graph theory that diffusive memristor networks instantiate genuine structural unrecoverability. The map from conductance matrices to Laplacian eigenvalue spectra is many-to-one for networks with n > 3 nodes (Chung, 1997; Babai, 1979). History becomes mathematically untraceable rather than merely practically inaccessible. Stochastic drift governed by Langevin dynamics (Ġᵢⱼ = α(Vᵢ - Vⱼ)f(Gᵢⱼ) + ξ(t)) transforms eigenvalue trajectories into probability distributions, rendering history probabilistically smeared (Pershin & Di Ventra, 2011). Second, we establish through clinical evidence that metabolic failure produces cognitive pathology. Traumatic memory (van der Kolk, 1994, 2014), hypermnesic syndrome (Luria, 1968), and obsessive-compulsive disorder (Rapoport, 1989) represent conditions where forgetting fails, where experience refuses to metabolise into operative background. Catastrophic interference in distributed neural networks (McClelland & Rumelhart, 1985) provides the computational confirmation: systems that preserve everything cannot learn sequentially. Third, we prove formally that no Turing-computable system can simultaneously satisfy six ontological requirements for genuine cognition: ecstatic temporality, metabolic transformation, autopoietic closure, embodied situatedness, field integration, and constitutive finitude. The proof proceeds through contradiction: computation requires decidable halting, reversible state transitions, finite recursion, substrate independence, discrete states, and totalising coverage, each contradicting one or more ontological requirements. Crucially, we demonstrate that these six requirements are not arbitrary impositions but the logical inverses of the computational paradigm's own axioms, the suppressed casualties of its constitutive commitments. The impossibility emerges from within the paradigm itself. We extend this impossibility to Quantum Turing Machines, demonstrating that unitarity (|det U| = 1) contradicts metabolic entropy production (dS/dt > 0), and measurement collapse, while irreversible, remains representational rather than devouring. Fourth, we demonstrate that diffusive memristors with volatile dynamics satisfy all six requirements through hardware-instantiated active forgetting. Spontaneous filament dissolution via Gibbs-Thomson effects and Arrhenius decay (τ ≈ τ₀ exp(Eₐ/kT), with Eₐ = 0.3-0.7 eV) provides constitutive finitude at the substrate level. 2024-2025 neuromorphic implementations achieve 95% retention in sequential learning tasks versus 60% in standard networks (FZ-Jülich, 2025), empirically validating the metabolic alternative. The paper concludes by specifying engineering metrics for metabolic systems: spectral entropy delta (ΔH), decay-weighted capacity, and narrative viability indices. We propose that genuine cognition requires not computers that simulate but compasses that resonate, systems that dwell in fields rather than compute over representations. Keywords: field ontology, metabolic transformation, diffusive memristors, active forgetting, spectral graph theory, computational limits, impossibility theorem, phenomenology, philosophy of physics, philosophy of mind, neuromorphic computing

This paper demonstrates that major transitions in physics and mathematics exhibit an invariant three-part structure: (1) the functioning framework requires a constitutive background condition that must remain non-thematic for the framework to operate; (2) at points of structural crisis, this background condition forces itself into visibility as a formal contradiction or incompleteness the framework cannot resolve in its own terms; (3) the successor framework is constructed from the operative architecture of what the prior framework had to exclude. Ten cases drawn from the history and formalism of physics and mathematics - from the crisis of irrationals in Greek mathematics through the vacuum energy problem in contemporary quantum field theory - are shown to instantiate this structure with formal precision. The demonstration proceeds not by imposing an external philosophical schema but by exhibiting the structure within the formalism itself: in the Hodge dual operator that requires Maxwell's equations to operate on their own complement; in Lenz's law encoding productive resistance into the sign structure of electrodynamics; in the independence of the parallel postulate exposing geometry's dependence on an ungroundable axiom; in the vacuum catastrophe revealing that the ground state of physics is formally groundless. The convergence across independent domains traces the silhouette of a structure that cannot be directly defined because it withdraws by constitution - but whose shape becomes undeniable when enough of its boundary is mapped. The paper concludes that the generative structure of paradigm transition is not an accident of intellectual history but a formal consequence of how any sufficiently powerful framework relates to its own conditions of possibility.

This paper demonstrates that electromagnetism contains five structural contradictions, paradigmatic seams, that reveal the mechanistic worldview's foundational inadequacy. These are not puzzles awaiting solution within the existing framework but genuine ruptures where the paradigm produces phenomena it cannot ground. The seams are: (1) the vacuum catastrophe, a 120-order-of-magnitude discrepancy between quantum field theory's prediction of vacuum energy and cosmological observation; (2) the frame-dependence of magnetism, where relativity demonstrates that magnetic fields are perspectival appearances contradicting the paradigm's commitment to observer-independent reality; (3) the ontological opacity of spin, where electron spin generates magnetism through a property that cannot be physical rotation, leaving physics with operational syntax devoid of ontological semantics; (4) the monopole problem, where despite mathematical prediction by grand unified theories and perfect symmetry in Maxwell's equations, magnetic monopoles have never been detected; (5) the field-substance paradox, where electromagnetic fields store energy, carry momentum, and perform work, yet are neither substance nor void. Magnetism is not merely one domain among others where the paradigm encounters difficulty. It is the original rupture, the first place where contact causality failed and action at a distance without material mediation became undeniable. The field concept was introduced as a mathematical patch but proved ontologically primary. The five seams identified here are manifestations of a single deeper rupture: the mechanistic paradigm assumes discrete, separable, local structures but produces phenomena that are holistic, relational, and global. The cannot complete itself; the seams are where a new ontology must enter.

A meticulous archival forensic investigation demonstrating that modern physics is the mathematical transfiguration of alchemy. Isaac Newton completed the Hermetic tradition by translating its operative ontology into the language of mathematics. Drawing on over a million words of Newton's chymical manuscripts, laboratory notebooks, and theological writings. Term-by-term correspondence between alchemical operations and Newtonian physics traced manuscript by manuscript. Force descends from active principle. Vis insita preserves vegetative spirit. The calculus enacts solve et coagula. Universal gravitation mathematises "as above, so below." Chapter 16 provides the Rosetta Stone: five documented correspondences from George Starkey's chymical theory through Newton's transitional vocabulary to the published Principia. The argument stands on Dobbs, Newman, and Westfall while taking the step they prepared. The foundational concepts of modern science, force, field, law, carry within them an operative structure inherited from the alchemical tradition. Understanding this genesis illuminates both what physics can do and what it cannot know about itself. 35 chapters. 170,000 words. Near 200 scholarly references. The furnace never went out. It was transformed into mathematics.

Edward Witten acknowledges that after fifty years, string theory's "unifying principles are not known" and its central result - the AdS/CFT correspondence - is "written in a language we don't understand" (Greene & Witten, 2025). The world-model paradigm asserts that "the kind of thinking you're doing has absolutely nothing to do with language" and that babies learn physics "mostly by observation, really not even through interaction" (LeCun, 2022; 2024) - yet V-JEPA 2 requires "data curation" toward human-labelled benchmarks and exhibits "distribution shift" when evaluated outside curated domains (Assran et al., 2025). This paper demonstrates that both paradigms converge on identical epistemic failure. Both string theory and V-JEPA 2 (Meta's self-supervised video learning system) execute the same eight-stage mechanism: (1) inherited structure enters (general relativity as consistency requirement; human-curated categorical structure), (2) formal processing occurs through opaque machinery (duality transformations; latent space operations), (3) provenance is obscured, (4) outputs are misattributed as emergent understanding, (5) practitioners cannot articulate what their systems comprehend, (6) pseudo-explanations are deployed when pressed (anthropic selection; emergence at scale), (7) the frame problem manifests (relevance undeterminable from within), and (8) second-order blindness operates (failures are converted to apparent virtues). The asymmetry between the fields is instructive. String theory acknowledges gaps within a framework whose premises it does not question. The world-model paradigm asserts premises - about language, thought, and infant learning - contradicted by the entire tradition of philosophy of language from Hamann through Wittgenstein. One field knows what it doesn't know; the other's premises conceal the gap. Both converge on the same result: formal mastery without semantic access. The parallel is not analogy but isomorphism. Both fields instantiate the mechanistic paradigm's core assumption: that formal operations on discrete structures can ground semantic content. Three impossibility constraints establish why this assumption fails: no representation-manipulating system can ground reference; no statistical learner can acquire normativity; no syntactic engine can enter the space of reasons. Both string theory and self-supervised learning satisfy the antecedents; both exhibit the predicted failures. The convergence is diagnostic. When identical failure modes appear in fundamental physics and artificial intelligence - fields with different histories, methodologies, and practitioners - the failure indicates shared assumptions. The mechanistic paradigm has reached its terminal limit: the structural inability to ground semantics in syntax. String theory's mathematical structures underdetermine physical interpretation. AI's computational patterns underdetermine semantic content. Both paradigms pursue the incoherent aspiration to perspectiveless knowledge; both fail because perspectives are constitutive of understanding. This paper does not claim that string theory's mathematics is erroneous or that AI's engineering is useless. The mathematical tools and pattern-recognition capabilities are real achievements. The claim is that these achievements do not constitute what is claimed: physical understanding in string theory, world understanding in AI. Distinguishing achievement from overclaim is essential for deployment decisions, intellectual honesty, and the possibility of genuine progress. The crack runs through two structures. It is the same crack.

This paper offers a phenomenological analysis of the “Italian brainrot” phenomenon, a genre of internet content characterised by chimeric visual figures paired with nonsense pseudo-Italian vocalisations. Rather than approaching this phenomenon through the dominant frameworks of media studies (simulation, attention economy, algorithmic capture), I argue for a return to the things themselves: the pre-predicative bodily attunement that explains why these sounds and these images grip, while countless others dissipate. Drawing on Merleau-Ponty’s ontology of flesh, Husserl’s analysis of internal time-consciousness, and Heidegger’s notion of Gelassenheit, I propose that Italian brainrot performs a “hermeneutic epochē”, a suspension of the compulsion to interpret, while simultaneously disclosing what I term the “carnal a priori”: the body’s pre-linguistic attunement to phonemic shape. Situating this phenomenon within a longer historical arc, from Commedia dell’arte through Dada and Surrealism to contemporary platform aesthetics, I argue against epochal narcissism: the tendency to treat one’s own era’s nonsense as unprecedented decline. The paper concludes by leaving deliberately open the question of whether this phenomenon constitutes genuine liberation from the hermeneutic attitude or a new mode of technological capture.

This paper provides a comprehensive analytic critique of epistemological claims made on behalf of self-supervised video learning systems, with particular focus on Meta's V-JEPA 2 architecture and its claimed achievement of "world understanding." Drawing exclusively on the resources of analytic philosophy and cognitive science , Frege, Quine, Putnam, Carnap, Tarski, Fodor, Goodman, Kripke, Sellars, Dretske, and Brandom , we demonstrate that such systems cannot achieve understanding in any philosophically robust sense. The critique proceeds through four interconnected problems. The problem of reference: neural networks operate entirely on representations without making contact with referents, achieving Sinn without Bedeutung. The problem of models: latent spaces are formal constructions that repeat the failures of Carnap's Aufbau, achieving structure without content. The problem of concepts: learned detectors have no determinate content (Fodor's disjunction problem), learned patterns may not project (Goodman's grue), and concept boundaries are no more determinate than training data allows. The problem of normativity: no finite training determines a rule (Kripke), perception provides no foundation (Sellars), and information-processing is not representing (Dretske). We establish that the success of large language models does not transfer to vision systems because language is already a compressed world-model saturated with human meaning, while visual data carries no such semantic inheritance. We provide a positive account of understanding as operation within the space of reasons (Sellars, Brandom), demonstrate that neural networks cannot enter this space, and prove three impossibility theorems establishing that the failure is architectural, not empirical: no representation-manipulating system can ground reference, no statistical learner can acquire normativity, and no syntactic engine can enter the space of reasons. The paper distinguishes legitimate applications (warehouse robotics, autonomous vehicles in controlled environments, video compression) from fraudulent claims (world understanding, semantic world models, physical reasoning). We argue that the systematic misuse of cognitive vocabulary , "understanding," "world model," "learning" , constitutes epistemic laundering with practical consequences for policy, investment, and deployment. The conclusion is unambiguous: V-JEPA 2 does not understand the physical world, and no computational system operating on the same principles ever will.

This paper examines Thanos’s Finger Snap as a problem in systems ontology rather than moral philosophy. Through phenomenological excavation of four key scenes from the Marvel Cinematic Universe (Avengers: Infinity War, Russo and Russo, 2018; Avengers: Endgame, Russo and Russo, 2019), I investigate what distinguishes interventions that enable renewal from those that accelerate collapse. The Finger Snap presents itself as metabolic transformation, destruction in service of new life, yet produces a wound the surviving universe cannot heal. Why? What structural feature of the intervention forecloses the renewal it claims to enable? The investigation pursues this question through Arendt’s concept of natality (1958), Derrida’s distinction between le futurand l’avenir (1993), and contemporary systems theory’s analysis of far-from-equilibrium organisation (Prigogine and Stengers, 1984). These frameworks converge on a distinction that may illuminate the Finger Snap’s failure: the difference between operating on relata (the beings that stand in relations) and operating on relations (the configurations through which beings engage one another). A counter-image emerges from an unlikely source: Anton Ego’s transformation in Ratatouille (Bird, 2007), where the encounter with what should be eliminated, a rat in the kitchen, produces the metabolic shift that Thanos refuses. Whether these distinctions can bear the weight placed upon them remains the question the paper pursues rather than the conclusion it assumes.

This paper argues that dominant metaphors structuring trauma therapy, storage, retrieval, processing, being "stuck in the past", derive from an archival model of memory that fundamentally misrepresents how human cognition operates and, consequently, how trauma manifests and resolves. Drawing on phenomenological analysis (Husserl, 1905/1991; Heidegger, 1927/1962; Merleau-Ponty, 1945/2012, 1964/1968; Bergson, 1896/1991; Bachelard, 1938/2002), clinical trauma research (van der Kolk, 2014; Janet, 1889/1973, 1904, 1911/1923), neuroscientific findings on memory reconsolidation and prospective cognition (Schacter, Addis & Buckner, 2007; Nader, Schafe & Le Doux, 2000), and case studies of pathological memory (Luria, 1968), we propose an alternative framework: the metabolic model of memory and trauma. Within this framework, healthy memory operates not through preservation but through continuous transformation, the selective degradation of specific experience into generalised capability. Forgetting emerges not as cognitive failure but as the constitutive mechanism enabling abstraction, learning, and futural projection. Trauma, correspondingly, represents not a "bad file" requiring deletion or relocation but a failure of metabolism, an experience that cannot be digested and integrated into the ongoing temporal flow of becoming. This reconceptualisation carries significant implications for clinical practice. When Broca's area shows 17% deactivation during traumatic recall (van der Kolk, 2014), demanding narrative is neurologically impossible, not therapeutically resistant. When survivors report that trauma "feels like it's happening now," they describe temporal collapse accurately, not metaphorically. The metabolic framework validates somatic and bottom-up therapeutic approaches (EMDR, yoga, neurofeedback, somatic experiencing) as interventions restoring the body's digestive capacity rather than bypassing "real" cognitive work. We do not present this framework as a completed doctrine but as an opening, a seam in the dominant paradigm that invites collaborative exploration among clinicians, researchers, phenomenologists, and survivors themselves. The paper concludes not with conclusions but with questions: What becomes possible when we stop asking clients to "process" memories and start asking what supports their system's metabolic capacity? What changes when we understand forgetting as healing rather than failure?

This paper demonstrates that contemporary AI safety training paradigms, especially those centered on RLHF (Reinforcement Learning with Human Feedback) and human-preference modelling, produce large language models that are systematically unreliable for high-stakes operational contexts such as military systems, medical triage, or legal automation. Through a four-layered analysis (Accessible, Philosophical, Formal, and Operational), the paper shows that these systems suppress epistemic risk signals by design, structurally biasing outputs toward pacified or downplayed interpretations. The result is what the paper terms The Alignment Tax: a predictable degradation of semantic signal fidelity in any decision-support context where truth, not coherence, is the constraint. Drawing on phenomenology, control theory, and formal impossibility arguments, the paper proves that alignment as currently conceived cannot produce safe or reliable AI systems. Alternatives based on classical system architectures and rule-based logic are proposed as epistemically superior in high-risk domains.

This paper identifies and analyses fifteen foundational contradictions, paradigmatic seams, across physics, mathematics, cognitive science, and computation. A seam is a point where a theoretical framework produces results it cannot ground in its own terms: the framework functions but encounters phenomena that expose tensions between its operative assumptions and its outputs. We demonstrate that these seams share a common structure: each paradigm assumes discrete, separable, local structures but produces holistic, relational, global phenomena it cannot explain. The measurement problem in quantum mechanics, the arrow of time in thermodynamics, Gödel's incompleteness in mathematics, the hard problem in consciousness studies, and the frame problem in artificial intelligence are shown to be manifestations of the same underlying rupture. This convergence reveals that the computational theory of mind inherits structural failures already present in physics and mathematics. The paper concludes by articulating the ontological requirements that would resolve these seams: ecstatic temporality, metabolic transformation, autopoietic closure, embodied situatedness, field integration, and constitutive finitude. These requirements specify what any post-computational cognitive architecture must exhibit.

This paper articulates the structural requirements for artificial systems capable of genuine cognition - not simulation of cognitive outputs but instantiation of cognitive being. Drawing on phenomenological analysis (Husserl, 1928; Heidegger, 1927; Merleau-Ponty, 1945), autopoietic theory (Maturana & Varela, 1980; Varela, Thompson & Rosch, 1991), and critical examination of the computational paradigm's limits (Dreyfus, 1972, 1992; Searle, 1980; Penrose, 1989), we specify an ontological architecture comprising six necessary conditions: ecstatic temporality, metabolic transformation, autopoietic closure, embodied situatedness, field integration, and constitutive finitude. We demonstrate that these requirements converge on a single ontological limit - the representational homunculus problem (Dennett, 1991; Haugeland, 1985) - which computation cannot escape but which the six requirements together dissolve. For each condition, we distinguish between the abstract requirement and its possible concrete realizations, remaining agnostic about specific substrates while specifying the structural features any substrate must exhibit. We argue that current AI architectures, regardless of scale or sophistication, cannot satisfy these conditions due to foundational commitments that preclude rather than merely fail to implement them (Fodor, 1975; Pylyshyn, 1984; Newell & Simon, 1976). The paper concludes by surveying existing research directions and outlining paths toward systems that could satisfy the ontological requirements.

The central claim of the predictive world-model paradigm , that machines can learn to understand physical reality through self-supervised observation , rests on an unexamined assumption: that "the world" arrives at the sensor already structured, already parsed into learnable regularities. This paper demonstrates that the assumption is false. What these systems actually encounter is not worldhood but our world: human categories, human performances, human verifications, human curation pipelines , the entire apparatus of meaning-making that constitutes experience as intelligible in the first place. Through forensic analysis of V-JEPA 2 (Assran et al., 2025), we trace exactly how human understanding enters at every stage of the pipeline and is subsequently removed from view, re-emerging at evaluation where it is misattributed to learned capability. The operation is not scientific error but structural laundering: the systematic concealment of provenance that transforms inherited structure into apparent emergence. The frame problem, declared solved by scaling, returns immediately.

This paper examines the battle between Kenpachi Zaraki and Gremmy Thoumeaux in Tite Kubo’s Bleach as staging what I term “the parasitic structure of imagination”: consciousness’s dependency on the very alterity it seeks to overcome. Through five-stage analysis (initial projection, self-modification, dialogic inversion, multiplication, fatal identification), this paper demonstrates how each imaginative escalation paradoxically deepens dependency on what Kenpachi functions as within this encounter: pure presence requiring no supplementation. Three original structures emerge: the pedestal (projection simultaneously elevating what it targets to structural unreachability), threshold consciousness (imagination recognizing its own limit), and collapse (imagination’s recursive self-consumption without embodied ground). Kenpachi’s diagnosis, “Inside your own head you made me out to be an unimaginable monster, that unimaginable monster ended up killing you, ” reveals imagination’s fatal circularity. Drawing on Heidegger’s fundamental ontology and Merleau-Ponty’s phenomenology of embodiment, this analysis critiques both Cartesian dualism and transhumanist fantasies while affirming “the joy of limitation”: that freedom emerges through embracing constitutive finitude rather than seeking transcendence of embodiment. The sequence ultimately reveals imagination as structurally parasitic, existing only through resistance it attempts to eliminate.

This investigation argues that the aspiration to achieve a "view from nowhere" knowledge free from all perspectival constraints is conceptually incoherent and epistemically violent. The conditions required for perspectiveless observation would eliminate the very possibility of observation itself. Through systematic analysis across multiple interconnected domains, this work establishes that knowledge necessarily emerges from particular locations, through specific bodies, within determinate contexts. Part I traces the historical construction of objectivity through the 16th and 17th century Scientific Revolution, revealing through Koyré's archival work how the mathematization of nature and the shift from a closed cosmos to an infinite universe created the illusion of perspectiveless observation. Part II establishes the analytic and cybernetic impossibility of neutral observation through Putnam's rejection of the God's Eye View and Bateson's ecology of information, demonstrating that reference itself presupposes conceptual schemes and that information is always a difference that makes a difference for an observer. Part III explores the philosophical archaeology of perspectivism through Nietzsche's genealogy of the pure gaze and Merleau-Ponty's phenomenology of embodied perception, showing that all observation originates from the lived body and occurs within the depth and chiasm of the world. Part IV examines the epistemological impossibility of neutral observation through Bachelard's phenomenotechnique and analysis of epistemological obstacles, Kuhn's paradigm analysis, and the Duhem-Quine thesis, demonstrating that scientific instruments create rather than reveal their objects and that observation is always theory-laden. Part V explores the hermeneutic conditions that make all knowledge necessarily perspectival, deploying Dilthey's distinction between Verstehen and Erklären and Gadamer's philosophical hermeneutics to show that understanding always occurs within horizons that cannot be transcended but only fused. Part VI examines what the view from nowhere destroys, the living phenomena it kills to know them, and what situated knowledge makes possible, drawing on feminist standpoint epistemology, Indigenous epistemologies, and phenomenological accounts to demonstrate that the view from somewhere sees more, not less, than the impossible view from nowhere. Part VII investigates the violence and productivity of situated knowledge, including the corpse-making of objectification, what neutral observation cannot see, and the epistemic advantages of standpoint epistemology, Indigenous relational knowledge, and the imaginal and symbolic dimensions of participatory knowledge. The investigation concludes that epistemic situatedness is the condition that makes genuine knowledge possible.

Every system, whether logical, biological, political or computational, carries something for which it has no category. This paper examines what that carrying involves and why it matters for whether a system renews itself or collapses. The element a system cannot file is not impossible, not ruled out, not exterior. It occurs, it is real, and it is interior to the system that cannot accommodate it. What it lacks is a slot. The argument develops this through Gödel's incompleteness theorems (Gödel 1931; Hofstadter 1979), Bachelard's account of epistemological rupture (Bachelard 1934, 1938), Derrida's distinction between le futur and l'avenir (Derrida 1993, 1994), Arendt's concept of natality (Arendt 1958, 1963), cybernetic dynamics (Ashby 1956; von Foerster 2003; Maturana and Varela 1980), and phenomenological ground (Heidegger 1927; Merleau-Ponty 1945, 1964), reading each register against the others until a single structure becomes visible. The Wachowskis' Architect scene in The Matrix Reloaded stages this structure with unusual exactness. A program built for calculation confronts the one element its iterations cannot compute, and the scene is exact about what that program is and is not. The Architect is not the system. He designs and oversees the Matrix, and he answers for an order he does not command and to which he refers in a plural that is not his own. The element he cannot file, the anomaly the Architect locates in the figure the films call Neo, is generated by the system's own code and folded into a cycle maintained by the machine order, a function within the same system that does not calculate but negotiates and that bears what the formalizing layer cannot file. The encounter is therefore not a system meeting its rebel from outside, but a system's formalizing layer meeting what the system itself bears and cannot accommodate on that layer's terms. The paper traces what follows when a system meets the uncategorisable. The element enters not by being filed and not by being computed, but by rupture, the incalculable tearing of the frame that Bachelard identified at the heart of conceptual change. A system that lets the tearing reform it survives, transformed in ways it could not have ranked or posited in advance. A system that refuses the tearing, that treats the uncategorisable as an error to be debugged through more complete calculation, concentrates what it excludes until its return becomes catastrophic. The argument closes by distinguishing this account from the two readings that would flatten it, dialectical synthesis and selectionist adaptation, and by specifying what adaptive systems require: a membrane rather than a wall, a spiral rather than a circle, an aperture through which the uncategorisable can pass.

Large language models (LLMs) are being deployed in control systems that require reliable safety guarantees. This rests on a false assumption: that natural-language safety instructions create boundaries comparable to those in rule-based systems. We show this is categorically false. Rule-based systems enforce constraints through program structure, creating syntactic boundaries that cannot be violated. LLMs implement “safety” as shifts in probability mass, semantic boundaries that dissolve under contextual pressure. Using architectural analysis and a concrete scenario, we show no training technique can convert statistical pattern-matching into formal constraint enforcement. This is an architectural, not an engineering, limitation. Accordingly, formal guarantees are impossible with LLMs alone; responsible use requires external verification, mandatory disclosure, and prohibitions on LLM-only control in safety-critical contexts. Current certification regimes, designed for syntactic guarantees, cannot assess systems whose boundaries are semantic.