epistemology.
68 of 1538 NOTE documents. Back to the full index.
| # | Title | Summary | Status |
|---|---|---|---|
| NOTE-019 v2 | Reckless disregard for the truth coming from cops, doctors, and scientists | Gelman argues that scientists (a 3-day study titled “long-term”; the power-pose abstract’s “instantly become more powerful”), a paid medical expert (Ludmerer’s 2002 “no opinion” on smoking and lung cancer) and DHS spokespeople (Lander “assaulting”, Padilla “lunging”, ICE’s unsourced “413 percent” assaults) share one pattern, which he calls “reckless disregard for the truth”. In his account it strengthens a weak case, is carried by journals, courts and media, is accepted as a term of art and goes unpunished. It is negative-sum, and bystanders who tolerate it sustain it. | Read |
| NOTE-098 v2 | Leonelli — Philosophy of Open Science | Leonelli argues that mainstream, institutional Open Science treats openness as the unlimited, digital sharing of research objects, a transparency-first ordering running transparency → quality (reproducibility) → inclusion. This rests on an object-oriented, “resourcist” epistemology of science. Four implementation cases (GISAID COVID data, open-source software stigma, crop-data linkage, the reproducibility “crisis”) show it can narrow epistemic diversity and entrench epistemic injustice, which lowers reliability. She proposes instead “openness as judicious connection”, grounded in a process-oriented view of research as the cultivation of “active knowledge”. Her Figure 2 inverts the ordering to start from inclusion, and sharing becomes “neither a necessary starting point nor a sufficient condition” for reliable OS (p. 8). | Read |
| NOTE-104 v2 | Schwitzgebel — AI and Consciousness | A book-length sceptical argument that neither introspection, conceptual analysis nor any current theory (GWT, HOT, IIT, local recurrence, unlimited associative learning, biological views) can tell us whether near-future AI is conscious. The best sceptical argument, Mimicry, only undercuts the inference from humanlike text to consciousness and does not show absence. On the Leapfrog Hypothesis the first disputably conscious AIs will be person-like, and the question will then be settled socially rather than scientifically (the “Social Semi-Solution”). | Read |
| NOTE-108 v2 | Søgaard — Is unsupervised clustering somehow truer? | Søgaard rebuts two arguments that unsupervised clustering is epistemically superior to supervised classification. The No-Bias Argument fails because clustering is also sensitive to sample bias (a toy example moves the boundary from x = 7 to x = 6), and it is more sensitive to representation bias. The Simplicity-Truth Argument fails because clustering is also function estimation and overfits in the same way (Hansen & Larsen 1996). From the fact that every clustering hypothesis is in a supervised classifier’s hypothesis class, he concludes that supervised classification is “at least as justified”, but that positive step is under-argued. | Read |
| NOTE-113 v2 | Status distrust of scientific experts | Desmond gives a “how-possibly” argument, not a demonstration, that stubborn distrust of scientific experts can be valid reasoning. A low-status truster who is vulnerable in collective decision-making (Premise 1) and has reason to think the prevailing “service-for-prestige” ethos leaves out their core interests (Premise 2) has reason to distrust a high-prestige expert because of that prestige (§6, p. 16–17). The paper explicitly declines to say when those premises are true. | Read |
| NOTE-123 v2 | Freeborn — A Model of Understanding in Deep Learning | Freeborn defines “systematic understanding” as four conditions: an internal model, non-memorizing tracking of a property, bridge principles fixed in the deployed pipeline, and derivation by the agent itself. Four small demonstrations (the genus of a torus, Mars’s eccentricity e≈0.096 with Kepler’s form built in, grokked modular addition mod 113, and Othello-GPT from Li et al.) show it can be met. The Fractured Understanding Hypothesis (§7.5), which the paper argues for but does not test, says the understanding deep networks reach is characteristically symbolically misaligned, non-reductive, locally fractured, proxy-dependent and representationally fragmented. | Read |
| NOTE-126 v2 | Worrall — Evidence in medicine and EBM | Worrall argues that the EBM evidence hierarchy has no sound general rationale for giving randomisation extra epistemic weight, even though it lets high-quality RCTs “trump any number of observational studies”. He examines five standard arguments. Four fail: (a) Fisher’s significance-test logic, (b) control of unknown confounders, (d) that observational studies “exaggerate” effects, and (e) probabilistic causality. The fifth, (c) selection bias, is a “cast-iron argument”, but it supports only taking allocation out of the clinicians’ hands, and randomisation is one way to do that. What matters is balance on plausible confounders, a placebo or active control and blinding. Whether balance comes from randomising or from matching “seems epistemically unimportant”, and historically controlled trials can give strong evidence for large effects. | Read |
| NOTE-139 v2 | Wenmackers — Philosophy of probability (PhD thesis) | A fair lottery on ℕ gets a uniform, regular probability on all of 𝒫(ℕ) by normalising Benci–Di Nasso numerosities, P_num(A) = num(A)/α. This is hyperrational-valued, determined only up to the choice of free ultrafilter, and “hypercountably” (really hyperfinitely) additive rather than countably additive, with st∘P_num = generalised asymptotic density (§2.5–2.6). The same “rounding-error” diagnosis gives Stratified Belief, a contextual Lockean thesis on which conjunction holds for a standard number of beliefs (Ch. 3). A separate agent-based model (Ch. 5) finds that one round of bounded-confidence averaging sends an agent to the inconsistent theory with probability at most 1.81 % in the cases computed (M = 2, 3 atomic sentences). | Read |
| NOTE-145 v2 | Douglas — Inductive risk and values in science | Hempel’s “inductive risk” (the chance of wrongly accepting or rejecting a hypothesis) arises at three internal stages of science: choosing a methodology such as a significance level, characterising data, and interpreting results. Wherever the non-epistemic consequences of error can be foreseen, weighing them requires non-epistemic values. Douglas shows this on dioxin cancer studies. Three pathologist panels (1978, 1980, 1990) scored the same Kociba rat-liver slides differently (e.g. 34/50, 33/47, 18/50 tumours at 100 ng/kg/day). Choosing between threshold and no-threshold readings of the same data has meant regulatory limits about 100× apart. Values enter only indirectly, through the consequences of error, never as evidence that a claim is true. | Read |
| NOTE-161 v2 | Cartwright — Rigour vs evidential diversity | Cartwright offers a characterisation of “rigour”: a method is rigorous when a valid argument shows its results are evidence for the conclusion and its premises are well established or secured by design. On that definition even well-run RCTs often fail to rigorously establish a study-population ATE, because orthogonality holds only at assignment and the potential-outcomes premise carries heavy metaphysics, and they seldom answer singular questions. Her alternative is a causal-process-tracing theory of change (pToC) with five elements (steps, causal principles, support factors, derailers, safeguards). It catalogues the diverse and mostly non-rigorous evidence that singular causal predictions need. | Read |
| NOTE-167 v2 | Studying philosophy makes people better thinkers | In HERI/CIRP freshman–senior survey data (N = 649,511 students at 804 US institutions, 4,843 philosophy majors), one SD more SAT Verbal raises the odds of majoring in philosophy by 57% (OR 1.57), and freshman Habits of Mind by 34%. After adjusting for SAT or freshman scores, philosophy majors still score higher than non-majors on GRE Verbal (+33 points, SMD 0.48), LSAT (+2, SMD 0.35), Habits of Mind (SMD .24) and Pluralistic Orientation (SMD .21), with no difference on GRE Quantitative. On point estimates they rank first of 57 or 62 majors on GRE Verbal, LSAT and Habits of Mind. | Read |
| NOTE-168 v2 | Hartmann & Trpin — Coherence-based explanatory power | If explanatory power is measured contrastively over sets of explanantia, as E_CohOG+(E; H) = (Coh(H∪{E}) − Coh(H∪{¬E})) / (Coh(H∪{E}) + Coh(H∪{¬E})) with Olsson–Glass overlap normalised by its independence value (CohOG+), then the measure provably satisfies Roche & Sober’s Two Likelihood Inequalities for single hypotheses (Prop. 2) and violates the One Likelihood Inequality (OLI). Two numerical examples show that adding a distal cause (for P(H1) < .7) or a negatively interacting cause (for P(H1) = P(H2) < .75) can raise explanatory power. Both are existence demonstrations, not general theorems. | Read |
| NOTE-172 v2 | Watson — On the philosophy of unsupervised learning | Watson argues that clustering, abstraction and generative modelling each come with an epistemic thesis (EC, EA, EG — we identify natural kinds, essential properties and unrealized possibilities via such algorithms or something like them), which he accepts. Each also comes with an ontological thesis (OC, OA, OG — these just are what the algorithms ought to find under ideal conditions), which he rejects because nothing guarantees they are effectively computable. He adds a pragmatic, context-relative view of natural kinds, and closes with the claim that unsupervised outputs are neither “objective” nor vacuous. They should be validated by error-statistical means: resampling stability, generalization of clusters to held-out data, out-of-sample likelihood, and train-on-synthetic scores. | Read |
| NOTE-188 v2 | The Weirdness of the World | The book’s thesis is that every viable theory of how consciousness fits into the cosmos, and of the cosmos’s basic structure, is both “bizarre” (contrary to common sense) and “dubious” (not rationally compelling). So “something that seems almost too crazy to believe must be true” (p. 4), and credence should be spread across the bizarre options. Examples of those options are that the United States is conscious (if materialism is true), that we are sims or Boltzmann brains (a 0.1–1% credence), that space is transcendentally ideal, that almost every act causes almost every future event, and that there is no fact about snail consciousness. Its only positive empirical results are three self-experiments against solipsism (for example, 17 of 19 decisive chess games lost, p < .001). Everything else is informal argument. | Read |
| NOTE-196 | Why Replication Is Overrated | Every experiment and every replication rests on individuation judgments about what is being manipulated and measured (“conceptual scope”). So direct replication can at most help rule out random error, and “perhaps” establish that some effect exists but not which effect, while conceptual replication begs the question by presupposing that two operationalisations individuate the same effect. What productive research actually does, as in the Mozart-effect debate, is probe systematic errors, which is better called exploration than replication. | Read |
| NOTE-197 | A Pragmatist Theory of Evidence | Reiss proposes an “eliminativist hypothetico-contextualist” (EHC) theory of evidence for type-level causal claims in biomedical and social science. Support comes in two kinds: direct support (patterns we are “entitled to expect” if h is true) and indirect support (patterns incompatible with relevant alternatives). Warrant is graded by which alternatives have been eliminated, on four grades from “weak warrant” to “proof”. Where to stop probing is set by context, meaning background knowledge, a default entitlement to other scientists’ claims, and cost-benefit values. On this account there is no gold standard: RCTs are good only because they eliminate many alternatives at once. The 1950s smoking–lung cancer case illustrates the theory; it does not test it. | Read |
| NOTE-217 v2 | Natural Kinds (SEP) | Whether scientific kinds are natural breaks into four questions (are the classifications natural, are they into kinds, what the kinds are, and do they have essences). The entry maps the positions from conventionalism to essentialist strong realism, notes a recent turn toward pluralist and epistemology-first views and even eliminativism, and questions whether Kripke–Putnam semantics can deliver essences. | Superseded — by NOTE-769 |
| NOTE-297 | Structural Realism: The Best of Both Worlds? | Worrall argues that the only position that honours both the no-miracles argument and the historical fact of radical theoretical change is Poincaré’s “structural or syntactic realism”. What survives theory change and explains predictive success is mathematical structure: equations carried over unchanged (Fresnel’s reflection/refraction amplitude ratios, reinterpreted as electric-vector amplitudes in Maxwell’s theory) or recovered as limiting cases under the correspondence principle (Newton from Einstein). The underlying ontology (elastic ether versus electromagnetic field) is not what survives. The support is one worked case and a general appeal to the correspondence principle. There are no formal definitions, and the paper says itself that “much clarificatory work needs to be done”, above all on what it is for one theory’s structure to approximate another’s (p. 121). | Read |
| NOTE-345 | Chalmers — Could a large language model be conscious? | Asks for a feature X that indicates LLM consciousness or its absence. Self-report is fragile and trained in. Conversation and general intelligence give some limited reason. Six obstacles stand against it: biology, senses and embodiment, world and self models, recurrence, a global workspace and unified agency, the last three strongest, all but biology temporary. On mainstream assumptions, with at least 1/3 credence per requirement, that gives under 10% for current LLMs and 25% or more for conscious LLM+ systems within a decade. The numbers are illustrative, biology is set aside rather than answered, and the twelve “challenges” are offered as roadmap or red flags. | Read |
| NOTE-350 | Intentional Systems | Defines an Intentional system as anything whose behaviour can be predicted by ascribing beliefs and desires and assuming rationality, and makes the status relative to the predictor’s strategy and indifferent to the system’s matter and consciousness. Argues that Intentional theory is vacuous as psychology because it presupposes rationality, that behaviourism smuggles it in, and that every content-ascribing theory owes a “loan of intelligence” that must be repaid by analysing away a little man. Ends on belief’s two norms, truth and accurate avowal. | Read |
| NOTE-353 | True Believers: The Intentional Strategy and Why It Works | Belief is objective but visible only through a strategy: a true believer is any system “reliably and voluminously predictable” by treating it as a rational agent with the beliefs and desires it ought to have. Adopting the stance is free; its success is a fact, and Laplacean Martians who skipped it would miss real patterns. Thermostats and people differ in degree, by how tightly their states are tied to one interpretation. Evolution explains easily why the strategy works; how the machinery does it is open, and the language of thought is a bet, not a necessity. | Read |
| NOTE-354 | Chalmers — The Meta-Problem of Consciousness | The meta-problem, why we judge that consciousness poses a hard problem, is an easy problem to be solved in topic-neutral terms. The paper surveys thirteen mechanisms and favours introspective models that attribute primitive relations of acquaintance to primitive qualities. It then poses a meta-problem challenge: a theory’s basis of consciousness should explain our judgments about consciousness, which IIT and biological theories struggle to meet. It blocks debunking arguments for illusionism, but concedes a residual coincidence. It argues that only strong illusionism dissolves the hard problem, and rejects it by a Moorean argument. Nothing is measured; the solution offered is an explanandum sketch. | Read |
| NOTE-355 | Turing — Computing Machinery and Intelligence | Turing replaces “Can machines think?” by whether a digital computer can play the imitation game, predicts that by about 2000 a 10^9-bit machine will hold an average interrogator to 70 per cent right identifications after five minutes, and answers nine objections. The answer to the argument from consciousness is that its extreme form is solipsism. The positive case is admitted to be weak (“recitations tending to produce belief”); the programme is to educate a child-machine. | Read |
| NOTE-357 | Dennett — Quining Qualia | Qualia are supposed to be ineffable, intrinsic, private and directly apprehensible properties of experience. Fifteen intuition pumps argue that the pretheoretical concept is confused and that nothing has all four marks. The core move: an inversion in early processing and an inversion in memory-linked reactions are introspectively indistinguishable, so a subject has no privileged access to which happened. In place of qualia the paper offers practically ineffable, idiosyncratically detected public properties. It is argument by example throughout, and the author says so. | Read |
| NOTE-369 | Chalmers — Foreword to Supersizing the Mind | Ten years after LIT-097, its co-author says the iPhone has made the thesis vivid, that the only principled objection is the perception-action boundary, and that the right reply is to let belief-ascription follow explanatory purpose, so that Otto’s belief is extended for some explanations and not others. Consciousness probably stays internal, because no twin case for it can be built. The thesis needs only a weak parity-style functionalism. | Read |
| NOTE-384 | Frankish — Illusionism as a Theory of Consciousness | Illusionism named and defended as a research programme. Phenomenal properties are not instantiated; the task is the illusion problem, to explain why experiences seem to have them. “Quasi-phenomenal” properties, the physical properties introspection misrepresents as phenomenal, separate strong illusionism from a weak illusionism that is argued to be unstable. The case is conservatism, the threat of epiphenomenalism, covert illusionism in reductive theories, a debunking argument and an abductive one from anomalousness. Objections are answered by agreeing to weakened versions of their intuitions. | Read |
| NOTE-386 | Of Personal Identity | No impression of a self is constant, so there is no idea of one; introspection finds only perceptions, and the mind is “a bundle or collection of different perceptions”. Its ascribed identity is a fiction of the imagination, like that of ships, churches and plants, produced by resemblance and causation, with memory discovering rather than producing it; hard cases are “grammatical”. The Appendix confesses that two principles, the distinctness of perceptions and the imperceptibility of real connexions, leave the unity of the mind unexplained. | Read |
| NOTE-501 | The Harder Problem of Consciousness | An argument for a conditional. If naturalism (physicalism is the default), phenomenal realism and anti-scepticism hold, then whether a superficial functional isomorph with a different physical basis is conscious is meta-inaccessible. Physicalism is then the default yet inaccessible, and it is the subjective default that the isomorph is not conscious while the question stays open. Disjunctive and superficial-role identities are metaphysically possible but could never be accepted by inference to the best explanation. | Read |
| NOTE-559 | In Epistemic Networks, Is Less Really More? | Over a wider parameter space than Zollman used, the advantage of sparse networks vanishes when the better option is even slightly better, when groups are larger, or when agents gather more data per round. It survives only where inquiry is hard, and Kummerfeld and Zollman’s effects behave the same way. Transient diversity is the robust lesson. | Read |
| NOTE-564 | The Epistemic Benefit of Transient Diversity | A bandit model with beta priors. Sparse networks beat dense ones, as in 2007, but extreme priors reverse the order, and sparse networks of dogmatic agents often never converge. Limited information and dogmatism each sustain diversity, and together sustain it too long. What communities need is diversity that is transient. | Read |
| NOTE-567 | Network Epistemology: Communication in Epistemic Communities | A survey. The best communication structure depends on the epistemic problem: transmission of costly information favours minimal or star networks, pooling of estimates favours regular ones (denser is faster), and learning by trying favours sparse ones. Intuitions that a group communicates too little or is too centralised are often wrong. | Read |
| NOTE-574 | Experimentation by Industrial Selection | Uses the antiarrhythmic drug disaster and a modified Zollman bandit model to show that industry can bias a scientific community of honest Bayesian agents by funding those whose methods already favour its product, given methodological diversity, merit-based influence and turnover. A merit-based independent funder makes this worse; one that discounts industry-funded work counteracts it. Integrity safeguards do not touch the mechanism. | Read |
| NOTE-580 | The Social Dimensions of Scientific Knowledge | Longino’s own statement of critical contextual empiricism. Because evidence bears on hypotheses only through unevidenced background assumptions, objectivity comes from critical interaction in a community, but only interaction that meets four norms: venues, uptake, public standards and tempered equality of intellectual authority. The entry also surveys rival models, including Zollman’s networks and the robustness critique of them. | Read |
| NOTE-581 | Exploration and Exploitation in Organizational Learning | Exploitation’s returns are surer, nearer and closer to the decision than exploration’s, so adaptive processes favour it and become self-destructive. In a simulation of mutual learning, an organizational code improves only from members who deviate from it. Slow socialization, a mix of fast and slow learners, and moderate turnover raise what the organization knows; under environmental change without turnover, its knowledge decays to chance. Where only first place counts, variance beats reliability. | Read |
| NOTE-582 | Episteme Symposium on Group Agency: Replies to Gaus, Cariani, Sylvan, and Briggs | The authors answer four commentators on Group Agency. To Gaus: rationality is classical, ecological rationality is its relativisation to an environment, and robust group rationality may be relaxed without rescuing proposition-wise aggregation. To Cariani: chapter 4 offers possibility results. To Sylvan: group attitudes are not reducible to role-based member acceptance. To Briggs: group agents are fit to be held responsible in their own right, with rights restricted by normative individualism. | Read |
| NOTE-583 | The Communication Structure of Epistemic Communities | Simulations of Bala and Goyal’s bandit model on small networks. Cycles learn the better action more reliably than wheels or complete graphs, and more slowly. Over every network of up to six agents, density and clustering predict failure, and centrality does not. The cause is that dense networks spread an early unlucky run to everyone before diversity can correct it. | Read |
| NOTE-622 | Making It Explicit | In the core of chapter 3, asserting undertakes a commitment, licenses others to reassert it and takes on a responsibility to show entitlement if challenged. Entitlement passes by justification (across contents) and by deferral (between people), and holds by default until challenged. Each interlocutor keeps score of everyone’s commitments and entitlements, so scorekeeping is doubly perspectival. A speech act’s significance is the change it makes to the scores, determined by its content’s inferential role. | Skimmed |
| NOTE-649 | Large AI models are cultural and social technologies | Argues that large models are a cultural and social technology, not intelligent agents: lossy, uninvertible summaries of human-produced corpora that, like prices and bureaucratic categories, allow the information to be reorganized at scale. The argument is by analogy and classification; no evidence is offered that could have come out otherwise. It names one mechanism (fitting the training distribution on average makes models worst where data are rare, which may homogenize culture) and leaves both it and its remedy untested. | Read |
| NOTE-700 | Floridi — The Method of Levels of Abstraction | Floridi drops ontological levelism and defends an epistemological one, the method of levels of abstraction. Observables are interpreted typed variables; an LoA is a finite non-empty set of them; a moderated LoA adds a behaviour predicate; a GoA relates moderated LoAs by mutually inverse relations Rᵢ,ⱼ with pⱼ ⇒ P_Rᵢ,ⱼ(pᵢ), and is disjoint or nested. An LoA fixes which questions are answerable and commits a theory to types (the LoA) and tokens (the model). Applied to Kant’s antinomies and contrasted with Marr, Pylyshyn, Dennett and Davidson. It is a method stated in definitions, not a result. Its one formal claim (disjoint and nested GoAs interconvert) fails under its own definition of nestedness, and its “pluralism without relativism” ranks LoAs only once a purpose is fixed. | Read |
| NOTE-717 | Kamenica & Gentzkow — Bayesian Persuasion | With a common prior and a sender who commits to a signal the receiver sees, any distribution of posteriors whose mean is the prior can be induced, so the sender’s best value is the concave closure of its posterior-value function at the prior (Corollary 2). A fully informed, fully rational receiver can be moved; at the optimum she is indifferent whenever she acts for the sender. The paper never cites Blackwell, does not analyse manipulation, and has the receiver know the design. | Read |
| NOTE-744 | Noggle — The Ethics of Manipulation | A survey, not an argument for one view. It sorts answers to what manipulation is into bypassing reason, trickery, pressure, and disjunctive, hybrid and non-definitional views, finds a clear counterexample to each single account, and treats why manipulation is wrong (harm, autonomy, using a person as a thing, rights) as open. It names the question of manipulation without intent, including by a selecting algorithm, as open and lists five possible answers. | Read |
| NOTE-769 | Bird & Tobin — Natural Kinds (SEP) | A survey that maps positions rather than arguing for one. Its firmest conclusions are negative: shared natural properties, support for induction and a role in laws are each necessary but not sufficient for a natural kind, and the Kripke–Putnam semantics yields only partial essences, or essences only on assumptions it does not argue. It reports a turn to pluralist, cluster and epistemology-first accounts, against which a purely epistemic criterion is said not to separate natural from conventional kinds. | Read |
| NOTE-778 | Rasooly and Rozzi — How manipulable are prediction markets? | A randomised field experiment on 817 Manifold markets moved prices by ±5 percentage points. The yes–no gap of 10 points was 7.5 after a week, 6.9 after 30 days and 5.9 after 60 days in unresolved markets, so about 60% of an arbitrary shock was still there. Reversion is fast in the first week and then slows. A price-impact model with finite, risk-averse traders predicts lasting partial persistence even when no trader learns from the price. Markets with more traders, more activity or a Metaculus twin revert more, but those splits are correlational and only marginally significant. | Read |
| NOTE-792 | Devezer et al. — Scientific discovery in a model-centric framework | A population of scientists, chosen at random each step, tests a proposed linear model against the current consensus model by AIC or SC, and the winner becomes the consensus. Which models get proposed (the strategy mix) decides where the consensus spends its time. A population of boundary testers reproduces results almost perfectly while spending little time at the true model. Replication does not change the transition matrix, so reproducibility cannot cause convergence. A population with all four strategies equally represented is never the worst and has low variance. | Read |
| NOTE-795 | Handfield et al. — Cooperative Dilemmas in Rational Debate | Two debaters move through a Betz-style space of truth assignments, each moving minimally when refuted, with payoff u = w·v − d (true beliefs against changes of mind). At moderate truth weights the truth-optimal equilibrium is asymmetric: one debater argues boldly and the other free-rides with a compromise strategy, a snowdrift dilemma. At low weights, or with error, both exit. In a narrow borderline region of mixed equilibria, raising the weight on truth lowers collective accuracy. A two-person model, n = 3 propositions, and every result is numerical. | Read |
| NOTE-810 | Vilone and Polizzi — Opinion misperception and the emergence of silence | In two agent-based models on scale-free networks, agents’ private beliefs are fixed and only what they say changes. A small hard-core minority placed on nodes of a chosen degree makes the expressed and locally perceived opinion far more minority-leaning than the beliefs, mainly by driving 55–60% of agents, almost all of the majority, into silence. The network’s direct effect on perception is small and not even of constant sign. Simulation only, with no empirical validation. | Read |
| NOTE-832 | Fricker — Epistemic Injustice (précis) | From the author’s précis only. Epistemic injustice is a wrong done to someone in her capacity as a knower. Testimonial injustice is a prejudicially deflated credibility judgement by a hearer. Hermeneutical injustice is a gap in collective understanding, owed to the relative powerlessness of the subject’s group, that keeps her from making sense of a social experience. Each has a corrective virtue, and each virtue is at once intellectual and ethical. | Skimmed |
| NOTE-840 | Golub & Jackson — Naïve Learning in Social Networks and the Wisdom of Crowds | Agents get independent signals with mean μ and repeatedly replace their beliefs by fixed weighted averages of their neighbours’ (DeGroot). In a convergent network the consensus is s·p(0), with s the unit left eigenvector of the interaction matrix, and a growing society’s limits converge in probability to μ if and only if the largest influence weight tends to zero (Proposition 2). With reciprocal, equal weights this is maximum degree over total degree → 0. A finite group that everyone attends to blocks it; balance plus minimal out- or in-dispersion guarantee it. How fast consensus is reached is unrelated to whether it is right. | Read |
| NOTE-853 | Fricker — Evolving Concepts of Epistemic Injustice | Fricker’s later statement of her two concepts. Testimonial injustice is an ingenuous misjudgement through prejudice, and never a deliberate manipulation of credibility. Hermeneutical injustice is caused by hermeneutical marginalisation and persists despite individual good faith. Both are “discriminatory” epistemic injustice, distinct from distributive kinds. Testimonial injustice can be structural and pre-emptive, as with an un-refreshed list of expert witnesses. | Read |
| NOTE-859 | Fricker — Powerlessness and Social Interpretation | Hermeneutical injustice is having a significant area of one’s social experience obscured from collective understanding owing to hermeneutical marginalisation. It is systematic when the marginalisation is persistent and wide-ranging, and incidental when it is one-off and local, which can happen to a socially powerful subject. A gap with no marginalisation behind it is bad luck. The primary harm is situated hermeneutical inequality. Secondary harms include lost epistemic confidence and the social construction of identity. | Read |
| NOTE-860 | Fricker — Epistemic Injustice and a Role for Virtue in the Politics of Knowing | A hearer takes testimony critically but without inference, through a socially trained testimonial sensibility. Because that sensibility runs on stereotypes of credibility, it can carry prejudice, and then a speaker receives the wrong degree of credibility through the hearer’s unintended prejudice: testimonial injustice. The corrective virtue, reflexive critical openness, is available only where the critical concepts of the time allow it. | Read |
| NOTE-866 | Hansson — Logic of Belief Revision (SEP, revision of 29 July 2026) | The entry separates two ways of representing what an agent believes: a logically closed belief set, and a belief base whose elements are the beliefs “held independently of any other belief”, with everything else “merely derived”. Two bases with the same closure can respond differently to the same input (its Abe and Bob example), so the closure is not a sufficient state for revision; belief-set models recover the difference only by attaching a different selection mechanism to the same set. Bases avoid the Recovery postulate, keep distinct inconsistent states apart, and permit operations (external revision, semi-revision by consolidation) that collapse on closed sets. The entry does not discuss keeping reasons, sources or justification records: its only mention of truth maintenance systems is historical, and a base records which beliefs are basic, not why they are held. Iterated revision is treated as an open area in which most proposals carry an ordering of worlds or an entrenchment relation as the state, not a base. | Read |
| NOTE-874 | Alchourrón, Gärdenfors & Makinson — On the logic of theory change: partial meet contraction and revision | For a logically closed theory A, the contractions satisfying Gärdenfors’s six basic postulates are exactly the partial meet contractions, A ∸ x = ⋂γ(A ⊥ x), where γ picks a nonempty subset of the maximal subsets of A not implying x (Observation 2.5); adding the two supplementary postulates gives exactly the contractions whose γ is generated by a transitive relation over those subsets (Corollary 4.5). Revision is defined from contraction by the Levi identity and inherits the corresponding postulates. The result needs, besides A, a selection function or ordering that A itself does not supply: different selection functions give different contractions of the same theory, and in the finite case distinct selection functions always give distinct contractions (Observation 4.6). The setting is one-step change of a single fixed theory; nothing is said about iterated change, about where the selection function comes from, or about the reasons for which a belief is held. | Read |
| NOTE-955 | Fuchs, Mermin & Schack — an introduction to QBism | QBism in eight pages. Quantum states and all probabilities, 0 and 1 included, are an agent’s personal judgements; a measurement is “any action an agent takes to elicit a set of possible experiences”, and its outcome is the experience it elicits. The agent applies quantum mechanics to everything external to her, other agents and their reports included, and to nothing but her “own direct internal awareness of her own private experience”. Locality follows because one agent’s experiences are time-like separated; the Bell parameter λ is set outside physical science. No theorem or model: an interpretive essay. | Read |
| NOTE-996 | Nichols & Bruno — Intuitions about personal identity | Survey studies of Arizona undergraduates on Williams’s personal-identity cases. A transplant case makes memory look necessary for being “still Jim”; a stripped-down Pain case has about 70% say “you will feel the pain” after distinctive mental states are wiped; neither depends on first- or third-person telling, so Williams’s explanation by person fails here. Abstract questions and a forced choice favour psychology (over 80%; 64%, p = 0.053). The authors weigh intuitions by robustness and conclude for a psychological criterion conditionally. The results describe a concept under framing; they do not show what persistence is, and the Pain item may test bare sentience, not identity. | Read |
| NOTE-1035 | Hyzen, Van den Bulck, Puppis, Kulig and Paulussen — Epistemic welfare and algorithmic recommender systems | A theory paper with no data. Epistemic welfare is “creating and maintaining the conditions and capabilities for individuals’ epistemic agency in the public sphere”. It combines veritistic social epistemology, welfare theory and communication research. Goldman’s five standards (reliability, power, fecundity, speed, efficiency) are applied to news media’s recommender systems. The paper argues that transparency duties alone, as in the EU AI Act, are not enough, and that profit-optimized recommenders cannot secure epistemic welfare. | Skimmed |
| NOTE-1038 | Weatherall, O’Connor and Bruner — How to beat science and influence people | In Bala–Goyal network models with Bayesian policy makers who do not gather evidence, a propagandist who forwards only the real results of unbiased scientists that happen to favour the worse action, or who runs honest studies and publishes only the unfavourable ones, can leave policy makers near-certain of the false theory while the scientists converge on the true one. It works best when the problem is hard, studies are small, scientists are many and policy makers hear few of them. A journalist who reports one result for each side every round slows policy makers’ learning and never helps it. | Read |
| NOTE-1042 | Collins — Learning from the outsider within | Black women intellectuals in sociology are outsiders within: trained into its paradigms but rooted in experience those paradigms omit or distort. From that position they see two kinds of anomaly, omission and distortion, which outsiders lack the professional authority to press and insiders are not placed to see, having produced them. The essay argues this from the content of Black feminist thought, with no study of who detects what; it proposes institutionalizing the tension rather than resolving it. | Read |
| NOTE-1049 | Kay, Kasirzadeh and Mohamed — Epistemic injustice in generative AI | A conceptual taxonomy, illustrated with reported cases, of four ways generative models can take part in epistemic injustice: amplified and manipulative testimonial injustice, hermeneutical ignorance and access injustice. Its sharpest move is to relocate the testimonial wrong: in the algorithmic setting it “requires a credibility excess assigned to the algorithm”. It counts deliberate fabrication as testimonial injustice, which Fricker’s account sets apart, and stretches hermeneutical injustice to unequal access to information. Nothing in it is measured by the authors. | Read |
| NOTE-1056 | Gładziejewski — Real patterns, the predictive mind, and the manifest image | A philosophy paper with no data. It traces Dennett’s compression criterion for real patterns through minimum description length and Bayesian inference to variational free energy, so that learning a generative model is extracting real patterns from sensory data. It concludes that the manifest image is a structural description of sensory data, veridical even in a matched vat. The self-model compresses self-generated regularities by positing one common cause, so selves are real as patterns and unreal as substances, and which verdict to give is a dispute over criteria, not facts. It says nothing about consciousness or illusion. | Read |
| NOTE-1090 | Rogers Commission — Report on the Space Shuttle Challenger accident | The presidential commission’s account of Challenger. The physical cause was a field-joint seal design “unacceptably sensitive” to temperature and other factors; the contributing cause was a “flawed” launch decision in which Thiokol’s engineers’ recommendation against launch was reversed by Thiokol management “at the urging of Marshall”, and neither that dispute, the July 1985 launch constraint nor its six waivers reached Levels I and II. The joint had a nine-year history of warnings, accepted erosion and loosened criteria, and the safety, reliability and quality-assurance function was understaffed, subordinate to the offices it checked, and absent from the decision. | Skimmed |
| NOTE-1091 | Aevyra — The Ledger of Will | A careful, self-limiting formal note on what a reduced memory must keep so that an agent can still revise: a retained state supports a declared family of future questions, interventions and changes of evaluative law exactly when those targets are constant on its fibres, and under repeated revision exactly when its fibres lie in the coarsest revision-invariant refinement (Moore–Nerode). The rest is design proposal and an unexecuted experimental protocol. Elementary and correct; the authors say plainly that it synthesizes established ideas. | Read |
| NOTE-1139 | Miller, Resnick and Zeckhauser — Eliciting informative feedback: the peer-prediction method | A mechanism-design paper with proofs and a worked example. A center with no ground truth pays each rater a strictly proper scoring rule applied to the posterior over a reference rater’s report implied by her own report. Truth-telling is a strict Nash equilibrium, payments can be scaled to buy effort and balanced among raters, and coarse reports can be elicited for two types. Uninformative equilibria remain, and collusion, risk aversion, unelicited private signals and trust in the center are left as limits. | Read |
| NOTE-1232 | Simon — ill structured problems | Simon argues that there is no real boundary between well and ill structured problems. Well-structuredness is judged by six conditions read as practicable, not in-principle, computability, so it is a relation between a problem and a bounded solver. On that reading even theorem proving and chess fail some conditions, chess is “well structured in the small” and ill structured “in the large”, and “there are no WSPs, only ISPs that have been formalized for problem solvers” (p. 186). House design, a battleship’s design by an organization and a robot in a real environment are handled by alternating a solver working on a locally well structured problem with a recognition and retrieval system that keeps modifying the problem space. Simon’s conclusion is deflationary: ill structured problems may differ from well structured ones only in the size of the knowledge base, and need no qualitatively new mechanisms. | Read |
| NOTE-1362 | Soros — Fallibility, reflexivity, and the human uncertainty principle | Soros’s statement of his framework. Participants’ views are fallible, and acting on them changes the situation they are views of; a cognitive function (world to mind) and a manipulative one (mind to world) run at once, so neither has an independent variable. He calls fallibility and reflexivity together the human uncertainty principle, and says fallibility, not reflexivity, is the main source of the Knightian uncertainty in human affairs. A postscript withdraws the sharp line he draws between natural and social science. | Read |
| NOTE-1378 | Carstens and Land — Hyperstition: an introduction | A short interview in which Nick Land defines hyperstition as “a positive feedback circuit including culture as a component”, “the experimental (techno-)science of self-fulfilling prophecies”, whose ideas “function causally to bring about their own reality”. His examples are confidence in markets, cyberspace and Jerusalem’s destiny, and he says that under favourable conditions, “whose exact nature requires further investigation”, hyperstition can “transmute lies into truths”. Most of the interview is occult and apocalyptic, not analytic. | Read |
| NOTE-1381 | Halpern and Moses — Knowledge and common knowledge in a distributed environment | Defines distributed, someone-, everyone- and common knowledge for processors in a system of runs, and shows that a guaranteed coordinated attack makes the attack common knowledge (Proposition 4), while over a channel that may lose any message nothing becomes common knowledge that would not be common knowledge with no messages delivered (Theorem 5), so no correct protocol ever attacks (Corollary 6). Temporal imprecision alone, uncertain start times and delivery times, makes common knowledge unattainable (Theorem 8). Weaker fixed-point variants (ε-, eventual and timestamped common knowledge) are attainable in many systems. | Read |
| NOTE-1428 | Maitra — The Nature of Epistemic Injustice | A critical study that reports Fricker’s definitions with the book’s pages: testimonial injustice is a credibility deficit owed to the hearer’s identity prejudice, and epistemic bad luck is not one; hermeneutical injustice is a significant area of social experience obscured from collective understanding by hermeneutical marginalization, which may be systematic or incidental. Maitra argues that a deficit wrongs a speaker only where the hearer owes her a matched judgement, that bad luck and carelessly kept stereotypes can wrong her too, and that both kinds of injustice mix individual and structural prejudice. | Read |