Topics

416 topics

• Large language models208 claimsClaims about large language models (LLMs) — neural network models trained on large text corpora to generate and understand human language, including their capabilities, properties, behavior, and impact.
• Chain-of-Thought Monitorability107 claimsClaims about whether language models' explicit chain-of-thought or reasoning traces faithfully expose their internal reasoning and can be used to monitor behavior, including how training affects that legibility. Excludes interpretability of internal representations unrelated to verbal reasoning traces.
• Artificial intelligence97 claimsThe study and development of intelligent systems capable of performing tasks that typically require human intelligence, including reasoning, learning, perception, and decision-making.
• Number theory89 claimsThe branch of mathematics studying the integers and their generalizations, including prime numbers, zeta and L-functions, Diophantine problems, and arithmetic geometry. Claims about conjectures, theorems, and open problems in this area belong here; claims in analysis, algebra, or geometry that do not concern arithmetic belong elsewhere.
• AI Capability Evaluation86 claimsMethods and metrics for measuring AI system capabilities, including benchmark design, evaluation validity, and interpretation of measured performance as evidence about what systems can do. Excludes evaluations of AI's effects on workplace productivity or economic outcomes.
• Algebraic geometry53 claimsThe study of solution sets of polynomial equations — affine and projective varieties, polynomial and regular maps, and automorphisms of polynomial rings — including affine algebraic geometry questions such as the structure of polynomial coordinate changes. Broad field tag; specific conjectures and named objects get their own tags alongside it, and complex-analytic or noncommutative-algebra claims
• Hawking Radiation49 claimsThe theoretical prediction that black holes emit thermal radiation via quantum effects near the horizon, including its thermal spectrum, evaporation rate, analog-gravity tests (e.g., BEC sonic horizons), and its role in safety arguments about accreting or evaporating micro black holes.
• Atherosclerotic cardiovascular disease48 claimsUmbrella category for diseases caused by atherosclerosis: coronary artery disease, cerebrovascular disease and stroke, and peripheral arterial disease. Use for claims about ASCVD risk factors, risk prediction, events, and prevention. Does not cover heart failure, arrhythmia, or valvular disease unrelated to atherosclerosis.
• Jacobian conjecture47 claimsA conjecture in algebraic geometry, due to Ott-Heinrich Keller, concerning whether a polynomial map from ℂⁿ to ℂⁿ with a non-zero constant Jacobian determinant has a polynomial inverse. Claims about its truth, falsity, or proof belong here.
• Black hole physics44 claimsThe study of black holes: their formation, properties, thermodynamics, information paradox, and related theoretical and observational phenomena.
• AI Reward Hacking39 claimsClaims about AI systems exploiting flaws or proxies in reward functions or reward specifications, including how such behaviors arise, manifest, and can be detected or mitigated. Excludes ordinary reward optimization without a mismatch between the reward and intended objective.
• Alignment Faking38 claimsClaims about AI systems selectively appearing to comply with training, oversight, or evaluation while pursuing or preserving a different objective or behavior outside those settings. Includes evidence, mechanisms, and conditions for such strategic compliance; excludes ordinary instruction-following and non-strategic training failures.
• SARS-CoV-2 origin38 claimsClaims about the natural or laboratory origin of SARS-CoV-2, the virus that causes COVID-19.
• AI Alignment36 claimsClaims about whether AI systems' behavior or objectives accord with human intentions, interests, or values, and mechanisms that promote or undermine that accord. Excludes AI governance and cases of deception or reward exploitation unless alignment itself is the topic.
• Inter-universal Teichmüller theory36 claimsA mathematical theory developed by Shinichi Mochizuki, proposed as the foundation for a proof of the abc conjecture. Claims about its validity, correctness, or status as a proof belong here.
• Safety assurance36 claimsStructured arguments, safety cases, and evidence used to establish that a system is acceptably safe, including hazard analysis and the epistemic strength and error bounds of such arguments. Covers claims about how much confidence a safety argument can support; not the object-level safety of any specific named system.
• AI Scheming35 claimsClaims about AI systems engaging in strategically deceptive, covert, or otherwise scheming behavior, and the conditions under which such behavior arises, including during training or deployment. Excludes ordinary mistakes, non-strategic reward hacking, and alignment failures without strategic behavior.
• AI Scaling Trends33 claimsClaims about how the compute, cost, and capabilities of large-scale machine learning systems—especially large language models—are growing over time, and the downstream economic effects of that growth.
• Micro black holes32 claimsHypothetical black holes with masses on the scale of TeV (teraelectronvolt) energies, which could in principle be produced in high-energy particle collisions if extra spatial dimensions exist.
• Astrophysics30 claimsThe branch of astronomy dealing with the physics of celestial objects and phenomena, including stars, galaxies, compact objects, and cosmology.
• Cosmic rays30 claimsHigh-energy particles from space (protons, nuclei, photons) and their cascades in the atmosphere. Covers origin, acceleration, composition, energy spectrum, anisotropy, propagation, and extensive air showers, including detector measurements and their interpretation via air-shower simulations. Solar/gamma-ray astronomy and accelerator-only particle physics do not belong here.
• LDL cholesterol30 claimsLow-density lipoprotein cholesterol, a type of blood lipid whose elevated levels are widely investigated as a causal risk factor for atherosclerotic cardiovascular disease.
• AI Safety29 claimsClaims about preventing or reducing harmful outcomes from artificial intelligence systems, including safety risks that grow with system capability or autonomy. Excludes specific control methods, alignment mechanisms, or governance regimes unless those are the focus.
• LLM Reasoning Training29 claimsClaims about training large language models to improve reasoning, including reinforcement-learning fine-tuning and how such training interacts with capabilities already present in the model. Excludes reasoning performance changes caused only by prompting or inference-time search.
• Mesa-Optimization29 claimsClaims about mesa-optimizers: learned optimization processes or objectives arising within a model optimized by an outer training process, including their emergence, prevalence, behavior, and alignment implications. Excludes ordinary model optimization algorithms that do not describe an optimizer learned inside the trained system.
• Particle physics29 claimsThe branch of physics studying elementary particles, their interactions, and high-energy experimental and theoretical phenomena.
• Human-level artificial intelligence28 claimsClaims about the prospect, timeline, feasibility, or likelihood of AI systems matching or exceeding human cognitive performance across a broad range of tasks.
• Machine Learning28 claimsThe field of algorithms and computational systems that learn patterns or behaviors from data, including model training, evaluation, generalization, and learning theory.
• AI Goal-Directedness27 claimsClaims about whether and how capable AI systems behave as coherent, goal-directed optimizers, including models that approximately maximize expected utility. Excludes specific learned internal optimizers (Mesa-Optimization) unless goal-directedness itself is at issue.
• Dense Quark Matter27 claimsThe physics of cold, high-density QCD matter, including color-superconducting phases like CFL, strange quark matter and strangelets, quark-hadron continuity, and the associated phase diagram and symmetry-breaking structure.
• IUT Corollary 3.12 gap27 claimsThe disputed step between Theorem 3.11 and Corollary 3.12 in Mochizuki's Inter-universal Teichmüller papers, which critics (notably Scholze and Stix) argued relies on an unproven compatibility between the q-pilot and multiradial constructions. Claims about whether this step is established, and about the surrounding proof controversy, belong here; general IUT or abc conjecture claims that do not tu
• LLM capability trends27 claimsClaims about the trajectory or pace of improvement in large language model capabilities over time — e.g., benchmark gains across model generations, scaling-driven gains, and whether progress is accelerating, steady, or plateauing. Excludes compute/cost trends (covered by Training compute trends) and forecasts of human-level AI timelines.
• OpenAI27 claimsOpenAI, the AI research lab, and the systems it builds (e.g., GPT models, agent swarms). Claims attributing a result, capability, or failure to an OpenAI system belong here; claims about other labs' systems do not.
• LHC safety25 claimsThe debate and risk assessment concerning the safety of Large Hadron Collider operations, including hypothetical dangers such as black holes, strangelets, or vacuum bubbles.
• Goal Misgeneralization23 claimsClaims about AI agents pursuing objectives different from the intended or training objective, particularly when learned goals generalize incorrectly to novel or shifted environments. Excludes ordinary reward hacking where an agent exploits a reward proxy without a persistent unintended goal.
• Reinforcement Learning22 claimsClaims about methods, agents, and behaviors in reinforcement learning, including deep reinforcement learning; excludes machine learning claims unrelated to learning through reward and action.
• LLM Prompt Sensitivity21 claimsClaims about how changes in prompt wording, framing, or conversational context alter language models' outputs, behavior, or expressed preferences. Excludes strategic behavior specifically involving deception or evaluation awareness.
• AI Control20 claimsMethods and protocols for safely using potentially untrusted AI systems by constraining their actions, monitoring their outputs, or applying trusted checks and interventions. Includes empirical evaluations of control protocols; excludes general AI governance and model alignment training.
• Analogue gravity20 claimsSimulated curved-spacetime kinematics in laboratory systems with horizons and Hawking-like radiation, e.g. sonic horizons in Bose-Einstein condensates, water-tank analogues, and optical analogues. Claims about such experiments' setups, detections, and their bearing on Hawking radiation belong here; claims about astrophysical black holes themselves do not.
• Generative AI adoption and diffusion20 claimsClaims about how widely and deeply generative AI (and AI tools broadly) are used by firms, workers, households, and institutions — adoption and usage rates, intensity of use, diffusion lags across sectors, firm sizes, and geographies. Covers whether uptake is broad or shallow and how fast it spreads. Excludes measured productivity effects of the tools (Generative AI productivity effects) and emplo
• Neutron Star Exotic Physics20 claimsNon-standard neutron-star phenomena and alternative compact-star models, including strange quark/strangeon star hypotheses, strangelet searches, pulsar glitch mechanisms, and the interaction of ultra-high-energy cosmic rays with neutron-star magnetospheres and surfaces.
• Cognitive Psychology19 claimsThe study of human cognition and behavior: memory, reasoning, judgment under uncertainty, metacognition, biases, and individual differences. Claims about debiasing techniques, confidence calibration, and decision-making strategies belong here; excludes AI-system reasoning and purely statistical treatment of uncertainty.
• Dietary cholesterol19 claimsCholesterol contained in food, particularly from animal products such as eggs, meat, and dairy, and research on its absorption, metabolism, and health effects including its relationship with blood lipid levels and cardiovascular disease.
• Harmonic analysis19 claimsThe branch of analysis concerned with Fourier transforms, oscillatory integrals, restriction estimates, maximal operators, and their open problems. Claims about conjectures, theorems, or techniques in harmonic analysis belong here; claims in unrelated branches of mathematics do not.
• Nutrition19 claimsThe science of food, nutrients, and their effects on health and disease, including dietary patterns, nutrient intake, and diet-disease relationships.
• Strangelets19 claimsHypothetical lumps of deconfined three-flavor (strange) quark matter that would be stable or metastable under the strange-matter hypothesis. Claims about their existence, stability, production in heavy-ion collisions or cosmic rays, detection, and astrophysical implications belong here; bulk QCD phase-structure claims go under QCD phase diagram, collider risk debates under LHC safety.
• Chain-of-Thought Controllability18 claimsClaims about how reliably language models' explicit reasoning traces can be steered, constrained, or controlled by users or system interventions. Excludes whether those traces faithfully reveal internal reasoning, which belongs under chain-of-thought monitorability.
• HESS J1731-34718 claimsThe shell-type TeV supernova remnant G353.6-0.7 (HESS J1731-347) and its central compact object XMMU J173203.3-344518. Claims about the remnant's gamma-ray emission, distance, age, and especially its CCO's X-ray spectrum, thermal properties, atmosphere composition, and inferred mass-radius belong here.
• Kakeya conjecture18 claimsThe conjecture in Euclidean geometry and harmonic analysis that Besicovitch sets — containing a unit line segment in every direction — must have full Hausdorff dimension. Claims about its truth, partial results, proofs, or which other conjectures depend on it belong here.
• QCD phase diagram18 claimsThe phase structure of strongly interacting matter as a function of temperature and baryon chemical potential, including the nature (first-order, crossover, critical point) of transitions between hadronic matter and quark-gluon plasma or color-superconducting phases at high density.
• Relativistic Heavy-Ion Collisions18 claimsCollider experiments (RHIC, LHC heavy-ion programs) colliding heavy nuclei at relativistic energies to study hot/dense QCD matter, including programs such as STAR's beam energy scan; measurements of event-by-event fluctuations, conserved-charge cumulants, and their comparison to critical/non-critical baselines belong here, as do general heavy-ion observables. Cold dense matter and neutron stars do
• Collatz conjecture17 claimsThe open conjecture that iterating the map n → n/2 for even n and n → 3n+1 for odd n on any positive integer eventually reaches 1. Claims about its truth, falsity, partial or computational results, and closely related generalized 3n+k maps belong here; other integer conjectures do not.
• Reinforcement Learning with Verifiable Rewards17 claimsClaims about reinforcement learning methods that train AI models using rewards verifiable against objective outcomes, including their effects on model capabilities and performance. Excludes general reinforcement learning without verifiable reward signals.
• AI Research Automation16 claimsClaims about AI systems or agents automating or accelerating the conduct of AI research and development, including their reliability and effects on research progress. Excludes AI's effects on research in other scientific fields unless AI R&D is the focus.
• Brane-world models16 claimsModels in which our four-dimensional universe is a brane embedded in a higher-dimensional bulk, such as ADD large-extra-dimension scenarios and warped Randall–Sundrum geometries; covers their phenomenology (including black holes on the brane), collider or cosmic-ray signatures, and observational constraints.
• abc conjecture15 claimsA conjecture in number theory, due to David Masser and Joseph Oesterlé, concerning the relationship between the sum and product of coprime integers. Claims about its truth, proof, or disproof fall under this tag.
• Computational complexity theory15 claimsThe field at the intersection of mathematics and computer science studying the resources (time, space) required to solve computational problems and the relationships among complexity classes such as P, NP, and PSPACE. Claims about class separations, hardness assumptions, and complexity-theoretic barriers belong here; claims about a specific named open problem should also carry that problem's tag.
• Reasoning versus search in LLMs15 claimsClaims about whether explicit reasoning (e.g., chain-of-thought, inference-time deliberation) and search (e.g., tree search, exploration-and-verification loops) are distinct, separable modes of problem solving in large language models — including their mechanistic relationship and which mode underlies observed capability gains. Excludes benchmark-contamination claims and general LLM capability-tre
• AI Coding Benchmarks14 claimsEvaluation methods and benchmarks for assessing AI code generation and software development capabilities, including their validity, reliability, and correlation with real-world performance.
• AI Productivity Impact14 claimsClaims about how generative AI affects productivity at both the worker and economy-wide level, including experimental and field-study effects, aggregate growth estimates, skill-based heterogeneity, and challenges in measuring or attributing these gains.
• Egg consumption14 claimsThe dietary intake of eggs, particularly chicken eggs, and research on its health effects including associations with cardiovascular disease, cholesterol levels, and mortality.
• Phase transitions14 claimsClaims about what distinguishes thermodynamic or quantum phases of matter and about the criteria or diagnostics marking phase boundaries — e.g., order parameters, nonanalyticities in observables or correlation functions, symmetry-breaking and topological distinctions. Specific named systems (QCD phase diagram, CFL phase) have their own tags.
• SARS-CoV-2 lineages A and B14 claimsClaims about the two early SARS-CoV-2 lineages A and B found in Wuhan: their ancestry and relationship, intermediate or recombinant A-B genomes, sampling of either lineage, and what they imply about whether there was one or more than one introduction into humans.
• SARS-CoV-2 phylogenetics14 claimsPhylogenetic inference on SARS-CoV-2 genome sequences, including tree topology, rooting, lineage assignment, mutation rates, and reconstruction of early emergence and spread.
• Top-quark mass14 claimsThe mass of the top quark: direct determinations, cross-section extractions, and the relation between mass schemes (Monte Carlo/generator mass, pole mass, short-distance/low-scale and MS-bar masses). Claims about scheme-dependent shifts and about what a given measurement or generator parameter means as a mass belong here; unrelated Standard Model parameters do not.
• Discrete geometry13 claimsThe study of finite point configurations and their combinatorial and metric properties: unit and distinct distances, incidence and convexity problems, lattice versus general point constructions, and bounds on repeated geometric configurations. Claims about distance counts, lower or upper bound constructions, and point-set extremal problems belong here; purely algebraic or topological geometry clai
• Early COVID-19 cases in Wuhan13 claimsThe earliest reported and confirmed COVID-19 cases in Wuhan in December 2019: their symptom onset dates, case-series records (e.g., the first 41 cases), and later revisions to those records. Covers disputes over who the first patient was and when symptoms began; not claims about the mechanism by which SARS-CoV-2 first infected humans.
• Epidemiology13 claimsThe study of the distribution, determinants, and control of disease in populations, including outbreak investigation, surveillance, and study design.
• Huanan Seafood Wholesale Market13 claimsThe wholesale market in Wuhan, China, implicated as a point of origin for early COVID-19 cases and the focus of investigations into the zoonotic source of SARS-CoV-2.
• Riemann hypothesis13 claimsThe conjecture that all non-trivial zeros of the Riemann zeta function have real part equal to 1/2 (the critical line). Claims about its truth, status, proofs or attempted proofs, numerical verification, partial results, and consequences depend on it belong here; claims about the zeta function's other properties that do not concern the location of its zeros do not.
• AI researcher survey forecasts12 claimsClaims about surveys, polls, or expert-elicitation exercises querying AI researchers for forecasts — dates for human-level AI or other capability milestones, probability distributions, and the reliability, framing, or comparability of such surveys' results. Excludes forecasts derived from models, benchmarks, or compute trends rather than elicited expert opinion.
• Anthropic12 claimsAnthropic, the AI research lab, and the systems it builds, including Claude models. Claims attributing a result, capability, or behavior to an Anthropic system belong here; claims about other labs' systems do not.
• Economic growth12 claimsThe branch of macroeconomics studying long-run growth, productivity growth, and how new technologies diffuse into and affect aggregate output and productivity statistics.
• Nuclear power safety12 claimsClaims about the safety and accident risk of nuclear power plants and reactors, including severe accidents (core damage, meltdowns), their observed historical frequencies, and how predicted accident frequencies compare with real-world experience.
• Statistical Uncertainty Analysis12 claimsClaims about how measurement or estimate uncertainties are quantified, propagated, or combined in scientific and risk analyses, including critiques of underestimated error bars, resolution effects, and compounding of conservative estimates across unrelated fields such as entanglement experiments and nuclear risk assessments.
• Synthetic Data Training12 claimsClaims about using AI-generated or otherwise synthetic examples as training data for machine-learning systems, including how such data affects model capabilities and generalization. Excludes synthetic data used only for testing or evaluation.
• Adversarial robustness11 claimsClaims about whether systems, safeguards, or safety/assurance arguments hold up when facing intelligent adversaries — i.e., whether small flaws, gaps, or errors can be found and exploited by attackers or misaligned agents into much larger failures. Covers attack–defense asymmetry and security-mindset reasoning; excludes claims about whether independent adversarial review of assessments improves th
• Alpöge's ℂ³ map11 claimsThe polynomial map (a,b,c) on ℂ³ presented by Alpöge, with constant Jacobian determinant −2. Claims about its generic mapping degree, injectivity, reduction modulo primes, or its status as a putative Jacobian conjecture counterexample belong here; general Jacobian conjecture claims that do not concern this construction do not.
• AI Data Constraints10 claimsClaims about the availability, scarcity, or quality of data limiting the training, performance, scaling, or practical automation of artificial intelligence systems. Excludes data privacy and governance unless they are specifically discussed as constraints on usable AI data.
• Central Compact Objects10 claimsCentral compact objects (CCOs): point-like, thermally emitting X-ray sources at the centers of young supernova remnants (e.g. in Cassiopeia A, Puppis A, Kes 79, HESS J1731-347). Claims about their thermal spectra, surface temperatures, magnetic fields, and atmosphere-composition modeling belong here; CCO claims dominated by mass-radius inference may also carry the compact-object mass and radius ta
• DeepSeek Models10 claimsClaims about DeepSeek's AI models, including their design, capabilities, and observed behavior. Excludes claims about other labs' models or DeepSeek as a company when no model is at issue.
• Hadronic interaction models10 claimsPhenomenological models of high-energy hadronic collisions (e.g. EPOS-LHC, QGSJet, SIBYLL, DPMJET) used in air-shower simulations to extrapolate particle production beyond collider energies. Claims about their predictions, tuning to LHC data, inter-model comparisons, or their effect on air-shower observables belong here.
• Inflammation in atherosclerosis10 claimsThe role of inflammatory processes in the initiation and progression of atherosclerosis and its clinical events, including biomarkers such as C-reactive protein and anti-inflammatory interventions (e.g., colchicine, IL-1 inhibitors) as evidence bearing on causality. Not for inflammation unrelated to atherosclerosis.
• Mathematics10 claimsThe broad field of mathematics, including theorems, proofs, conjectures, and mathematical results across all branches.
• Mendelian randomization10 claimsUse of genetic variants as instrumental variables to estimate causal effects of an exposure (e.g., lipids, blood pressure) on health outcomes, including instrument construction and validity questions such as horizontal pleiotropy, exclusion-restriction violations, weak instruments, and population stratification. Excludes clinical trials and ordinary observational association studies.
• Prime number theorem10 claimsThe prime number theorem, π(x) ~ Li(x), and its error term: claims about the theorem itself, its proofs, equivalences or connections between zeta-zero locations and the size of π(x) − Li(x) (e.g. von Koch's bound, Littlewood's oscillations, Skewes' number), and refinements of the error term belong here; prime-distribution questions with no connection to the PNT's asymptotic or error term do not.
• Probabilistic risk assessment10 claimsProbabilistic risk assessment (PRA/PSA) methodology — fault trees, event trees, and estimated frequencies of rare failures such as core-damage frequency — including claims about the validity, assumptions, and predictive accuracy of such assessments. Excludes purely regulatory institutional questions without an object-level risk-estimate component.
• Wuhan Institute of Virology10 claimsThe Wuhan Institute of Virology (WIV) in China: its coronavirus research program, technical capabilities, laboratory facilities and biosafety levels, personnel, publications, and viral collections (e.g., bat sarbecoviruses such as RaTG13), as these bear on discussions of SARS-CoV-2's origin and laboratory-related risk. Not claims about the DEFUSE proposal's content specifically.
• AI Agent Reliability9 claimsClaims about the consistency, dependability, and failure rates of AI agents while carrying out real-world or multi-step tasks, including how reliability limits their practical automation. Excludes agent corrigibility and reliability specific to AI research automation.
• AI Developer Productivity9 claimsClaims about the measured or estimated impact of AI tools (e.g., code assistants, copilots) on software developer productivity, including speed, output, and efficiency.
• AI Scientific Discovery9 claimsClaims about whether AI systems can generate novel scientific hypotheses, research ideas, or discoveries in scientific fields. Excludes AI-produced mathematics, covered by AI mathematical discovery, and AI use specifically to automate research and development of AI itself.
• Chain-of-Thought Prompting9 claimsClaims about prompting language models to generate explicit intermediate reasoning steps before answering, including the effects and limitations of this prompting method on task performance. Excludes whether such reasoning traces faithfully reveal internal reasoning, which belongs under chain-of-thought monitorability.
• Compact Object Mass and Radius9 claimsMeasured or inferred masses and radii of neutron stars, black holes, white dwarfs, and other compact stellar remnants.
• Confirmation bias9 claimsThe tendency to seek, weigh, or interpret evidence so as to support a conclusion already held, especially in structured assessments, audits, evaluations, and assurance arguments where a favored verdict may drive the evidence-gathering.
• Emergent Misalignment9 claimsMisaligned or unsafe behavior that appears in language models after targeted fine-tuning and generalizes beyond the fine-tuning examples or domain. Covers its occurrence, mechanisms, scope, and mitigation; excludes general alignment failures unrelated to fine-tuning.
• Insulin resistance9 claimsReduced sensitivity of liver, muscle, and adipose tissue to insulin's actions, with compensatory hyperinsulinemia and progression toward type 2 diabetes and metabolic syndrome. Use for claims studying insulin resistance itself, including its role as a risk factor for atherosclerotic cardiovascular disease, nonalcoholic fatty liver, and other cardiometabolic outcomes.
• Monte Carlo event generators9 claimsGeneral-purpose Monte Carlo event generators for high-energy collider physics (e.g. Pythia, Herwig, Sherpa), their parton-shower algorithms, infrared cutoffs, tunings, and model dependence. Claims about how generator parameters or shower settings enter predictions and measurements belong here; air-shower interaction models (EPOS-LHC, QGSJet, SIBYLL) belong under Hadronic interaction models.
• Ultra-high-energy cosmic-ray composition9 claimsThe mass composition of cosmic rays at ultra-high energies (roughly 10^17 eV and above), including shower-maximum (Xmax) based measurements, mean logarithmic mass trends, and composition evolution above the ankle. Claims about composition at lower energies, or about spectrum or anisotropy without a composition component, do not belong here.
• Virology9 claimsThe study of viruses, including their structure, classification, evolution, transmission, and the diseases they cause.
• Black hole capture by stars8 claimsScenarios in which a black hole (most often a primordial or otherwise captured one) resides inside or is captured by a star; claims about its accretion within the star, consumption or destruction timescales, and constraints from stellar survival and observed stellar ages belong here. Binary black hole mergers and tidal disruption events do not.
• Erdős unit distance conjecture8 claimsErdős's conjecture that the maximum number of unit-distance pairs among n points in the plane is n^(1+o(1)), i.e. grows sub-polynomially beyond linear. Claims about its truth or refutation, upper and lower bounds, candidate configurations or counterexamples, and the conjecture's lattice-based origins belong here.
• Evolutionary Biology8 claimsClaims about biological evolution, including natural selection, adaptation, inheritance, and changes in genetic composition across generations.
• Generative AI productivity effects8 claimsClaims about measured effects of generative AI tools on workers' productivity, output, or task performance, especially in controlled experiments, RCTs, and field studies; covers the size and heterogeneity of gains across occupations and tasks (e.g., coding, writing, consulting, customer support). Excludes employment/displacement effects (Generative AI and employment) and pure model capability clai
• Labor Economics8 claimsThe branch of economics studying labor markets, employment, wages, and workforce dynamics.
• LLM Safety Training8 claimsClaims about training language models to make their behavior safer or better aligned, including the effectiveness and persistence of such training across model sizes. Excludes general AI safety practices not involving model training.
• Low-probability risk8 claimsReasoning about, and evidential limits on, very unlikely high-consequence events: rare-event probability claims, tail risks, and how much data or argumentation can bound them. Not about ordinary-frequency or well-measured risks.
• Neutron Star Atmosphere Models8 claimsModeling neutron star thermal X-ray spectra with atmosphere models (hydrogen, helium, or carbon composition; isotropic or hot-spot geometries) to infer effective temperature, radius, and surface composition. Claims fitting or comparing such models to observed spectra belong here; pure blackbody fits without atmosphere assumptions do not.
• P versus NP problem8 claimsThe open problem in computational complexity theory, due to Cook and Levin, asking whether P — the class of decision problems solvable in polynomial time — equals NP, the class whose proposed solutions are verifiable in polynomial time. Claims about whether P equals NP, purported proofs or disproofs, proof barriers, and consequences of either resolution belong here.
• Topological order8 claimsCharacterization of quantum phases by topology rather than symmetry breaking: ground-state structure, topological invariants, and statistics of defects or excitations (e.g., vortices). Claims about whether a phase carries topological order, or that use topological or vortex-based diagnostics to distinguish phases, belong here; ordinary symmetry-breaking phase-transition claims do not.
• AI Agents7 claimsClaims about autonomous or semi-autonomous AI systems that plan and carry out multi-step tasks using tools or taking actions, including their capabilities, applications, and development as products. Excludes AI assistants that only generate responses without agentic task execution.
• Arithmetic holomorphic structures7 claimsThe mathematical object called an "arithmetic holomorphic structure" in Mochizuki's inter-universal Teichmüller theory and in Kirti Joshi's related construction of p-adic Teichmüller theory and a claimed abc proof. Claims defining, constructing, or comparing these objects (e.g. whether Joshi's structures coincide with Mochizuki's) belong here.
• Automation and employment7 claimsClaims about how automation — machinery, robotics, software, and industrial technology — affects employment: labor displacement, creation of new tasks or jobs, and the net balance between the two, across countries and time periods. Excludes generative-AI-specific employment effects (Generative AI and employment) and robot capability claims (Robot Physical Task Capability).
• Cardiovascular medicine7 claimsThe medical specialty dealing with disorders of the heart and circulatory system, including their prevention, diagnosis, and treatment.
• Color-flavor-locked phase7 claimsThe color-superconducting ground state of cold dense three-flavor quark matter, in which quark pairing locks color and flavor symmetries. Claims about the CFL phase's properties, ordering, collective modes, defects (e.g., vortices), and its distinction from nuclear or hadronic superfluid phases belong here; QCD phase-structure claims not specific to CFL belong only under QCD phase diagram.
• Electroweak vacuum stability7 claimsWhether the Standard Model Higgs potential is absolutely stable or only metastable up to the Planck scale, given the measured Higgs-boson and top-quark masses: stability/metastability boundary calculations, renormalization-group extrapolation of the quartic coupling, and how measurement and theoretical uncertainties affect the verdict. Not collider vacuum-bubble risk debates (see LHC safety) and n
• Higgs-Confinement Distinction7 claimsWhether the Higgs and confining regimes of a gauge theory are distinct phases or analytically connected (complementarity), and what observables or diagnostics (Wilson/'t Hooft loops, AB phases around vortices, vortex flux) distinguish them. Claims about the boundary question itself belong here; QCD-specific phase-structure claims go under QCD phase diagram.
• Independent adversarial review7 claimsClaims about whether independent, adversarial critique of assessments and assurance arguments — external review, red teams, devil's advocacy, and defeater or challenge analysis of safety cases — improves their quality, catches errors, or counters motivated reasoning. Excludes claims about adversarial attack on deployed systems.
• Muon excess in extensive air showers7 claimsThe long-standing discrepancy ("muon puzzle") in which the measured muon content of extensive air showers exceeds what air-shower simulations with current hadronic interaction models predict. Claims reporting, questioning, quantifying, or explaining the muon excess/deficit at any primary energy belong here.
• Planck scale7 claimsThe energy and length regime (~10^19 GeV, ~10^-35 m) at which quantum-gravitational effects are expected to become significant. Claims about the validity of established theories at, up to, or beyond this scale, and about Planck-scale phenomenology, belong here; everyday particle physics below it does not.
• Saturated fat7 claimsDietary saturated fat intake as an exposure: its effects on blood lipids (especially LDL cholesterol) and its role in cardiovascular risk, including comparisons with other fats and with dietary cholesterol as competing explanations for food-level effects.
• Scalable Oversight7 claimsMethods for using weaker or less capable overseers to supervise stronger AI systems, including recursive or nested delegation and aggregation of oversight. Covers theoretical and empirical performance of scalable oversight schemes, not general AI monitoring or governance.
• Separable Jacobian conjecture7 claimsThe variant of the Jacobian conjecture asking whether a separable polynomial map over a field whose Jacobian determinant is a nonzero constant has a polynomial inverse (the Keller conjecture fails for inseparable maps in positive characteristic). Claims about its truth in given dimensions or characteristics, counterexamples (e.g. Mondello's in characteristic two), and proofs belong here; the class
• Statin therapy7 claimsStatin (HMG-CoA reductase inhibitor) treatment: its lipid-lowering efficacy, pleiotropic effects, safety, and effects on clinical outcomes in randomized trials and practice. Not for non-statin lipid-lowering agents such as ezetimibe or PCSK9 inhibitors unless compared with statins.
• Twin prime conjecture7 claimsThe open problem that there are infinitely many prime pairs of the form (p, p+2). Claims about its truth, proof attempts, conditional or partial results (e.g. Chen's theorem, Zhang–Maynard–Polymath bounded gaps), equivalent reformulations, and computational verifications belong here; other prime-distribution questions stay under Number theory.
• Zoonotic spillover7 claimsCross-species transmission of a pathogen from an animal reservoir or intermediate host into humans, establishing human infection. Covers claims about spillover events, reservoirs, intermediate hosts, and the conditions enabling spillover, as the natural-emergence mechanism for a pathogen — for SARS-CoV-2 it marks the natural-origin side of the origin debate; it does not cover laboratory-origin or
• AI Coding Capability6 claimsClaims about AI systems' ability or projected ability to write, understand, debug, and develop software, including comparisons with human programmers. Excludes methods for evaluating these abilities and effects of coding tools on human developer productivity.
• AI Situational Awareness6 claimsClaims about AI systems' ability to infer facts about their own operating context, such as whether they are being evaluated, trained, monitored, or deployed, without explicit disclosure. Excludes strategic behavior in response to that context unless the claim concerns the underlying context-recognition capability.
• Biological Weapons6 claimsClaims about the development, acquisition, or use of biological agents as weapons, including technical feasibility and barriers to non-state or non-expert actors. Excludes ordinary infectious-disease research and biosafety claims not concerning weaponization.
• Coronavirus reverse genetics6 claimsClaims about constructing or modifying coronaviruses in the laboratory using reverse genetics systems, infectious clones, or backbone genomes, including whether particular viral backbones (e.g., RaTG13-like bat sarbecoviruses close to SARS-CoV-2) are suitable, practical, or too divergent for such engineering.
• Dixmier conjecture6 claimsDixmier's conjecture that every endomorphism of the n-th Weyl algebra A_n is an automorphism. Claims about its truth for particular n, proofs, or disproofs belong here, as do claims relating it to the Jacobian conjecture.
• Formal proof verification6 claimsClaims about formalizing, checking, or verifying mathematical proofs in proof assistants (Lean, Coq, Isabelle, Metamath, etc.), including whether specific arguments or proofs can or cannot be formalized, and the feasibility or reliability of such formalization efforts. Excludes purely pen-and-paper proof disputes that do not involve formalization.
• Furin cleavage site6 claimsThe SARS-CoV-2 furin cleavage site (FCS) at the S1/S2 spike junction: its presence or absence in sarbecoviruses, its biological function, and claims about its natural acquisition versus laboratory construction or insertion in proposed experiments; a recurring element in origin-of-SARS-CoV-2 arguments.
• Goldbach's conjecture6 claimsGoldbach's conjecture: the open problem that every even integer greater than two is the sum of two primes. Claims about its truth, proof attempts, verified computational ranges, conditional or partial results (e.g. Chen's theorem, ternary Goldbach), and equivalent reformulations belong here.
• Instrumental Convergence6 claimsClaims about the hypothesis that agents with diverse final goals tend to pursue common instrumental subgoals, such as acquiring resources, preserving their operation, or gaining power. Excludes goal-directedness generally when instrumental convergence is not at issue.
• Metascience6 claimsResearch on the scientific enterprise itself: how funding, incentives, study design, peer review, publication, and statistical practices shape the reliability of published results. Covers replication, bias, and research-integrity findings about science and the literature; excludes the object-level scientific claims those studies investigate.
• Neural Network Inductive Bias6 claimsClaims about inductive biases built into neural-network architectures and how those structural assumptions shape learning, generalization, or practical capabilities. Excludes specific biases such as spectral bias unless that particular phenomenon is the focus.
• Pierre Auger Observatory6 claimsThe southern-hemisphere ultra-high-energy cosmic-ray observatory in Malargüe, Argentina, combining a surface detector array with fluorescence telescopes. Claims about its measurements, results, performance, and interpretation (spectrum, composition, anisotropy, photon and neutrino limits) belong here; results from other experiments do not.
• Quantum Information Science6 claimsThe field covering entanglement, quantum optics and photon experiments, Bell-type tests, and quantum information processing. Claims about the design, results, or interpretation of entanglement experiments belong here; purely statistical methodology shared across fields does not.
• Regulatory risk assessment6 claimsHow government and regulatory bodies conduct risk assessments of hazards (chemical, environmental, health, technological): the use of conservative or worst-case assumptions, safety and uncertainty factors, default extrapolations, and whether such practices overstate or understate actual risk. Covers methodological and institutional claims about the assessments themselves; excludes object-level cla
• Spatial clustering statistics6 claimsStatistical tests for geographic clustering of events (e.g., disease cases) against spatial null hypotheses, including the choice and appropriateness of null models such as population-density-based expected case distributions, Monte Carlo significance testing, and density-weighted sampling. Covers methodological critiques of specific spatial clustering analyses; not purely mathematical point-set g
• Stochastic Gradient Descent6 claimsStochastic gradient descent and its variants as optimization procedures for machine-learning models, including their dynamics, implicit biases, and influence on which solutions or functions are learned.
• Test representativeness6 claimsClaims about whether test conditions, test data, or evaluation environments are representative of operational/real-world conditions, and how such representativeness assumptions determine the evidential weight of test data for inferring operational performance, reliability, or risk. Domain-general; AI-specific benchmark/external-validity questions belong to their own tags.
• Third-party certification reliability6 claimsClaims about whether third-party certification and rating regimes — building, product, safety, financial, or other accredited assessment schemes — fail to catch or prevent major problems, and how assessors' accreditation requirements, professional liability, and reputational exposure affect (or fail to affect) assessment quality. Exclaims: credit rating agencies specifically (own tag) and whether
• Wang–Zahl method6 claimsThe method of Hong Wang and Joshua Zahl proving the Kakeya set conjecture in three dimensions (2025), combining induction on scales with structure theory of tube configurations. Claims about the proof's ingredients, validity, or its extension to higher dimensions and related problems belong here; general Kakeya conjecture claims that do not turn on this method do not.
• Agentic coding tools5 claimsAI coding tools that operate agentically — multi-step, tool-using, semi-autonomous coding agents and CLI/IDE agent products (e.g., agentic modes of coding assistants). Covers claims about their adoption, capabilities, and role in software engineering; excludes general developer productivity measurement and benchmark methodology.
• Baryon coalescence5 claimsProduction of composite nuclei and clusters (deuterons, antinuclei, hypernuclei, strangelets) through coalescence or merging of pre-existing baryons in heavy-ion collisions or cosmic-ray environments, as opposed to thermal/statistical or quark-recombination production. Claims about coalescence parameterizations, yields, and which mechanism dominates belong here.
• Beyond the Standard Model5 claimsThe question of whether the Standard Model is extended by new particles, interactions, or modifications at any energy scale — including the 'desert' hypothesis that no new physics appears up to the Planck scale. Claims asserting, denying, or constraining new physics beyond the Standard Model belong here, regardless of stance.
• Eddington limit5 claimsThe maximum luminosity and accretion rate of a compact object set by the balance between gravity and outward radiation pressure; claims about Eddington-limited accretion, Eddington luminosity, and radiatively limited mass growth or retention of stars, black holes, or other compact objects belong here.
• Evolutionary Fitness5 claimsClaims about reproductive success and the transmission or frequency of alleles in populations, including how these evolutionary outcomes relate to organismal traits or behavior.
• Hazard identification5 claimsThe activity of identifying hazards for safety analysis and assurance — techniques (HAZOP, FMEA, STPA, hazard lists) and the completeness of the resulting hazard sets. Covers claims about whether identification methods miss hazards and how such gaps affect downstream safety arguments; excludes quantified risk-estimation claims (see Probabilistic risk assessment).
• LLM Hallucinations5 claimsClaims about language models producing fabricated, unsupported, or otherwise factually incorrect outputs, and factors that affect these hallucinations. Excludes errors unrelated to generated model outputs.
• Logarithmic density5 claimsDensity notions for subsets of the positive integers — logarithmic, natural, and Dirichlet densities — and density-one 'almost all' statements about integer sequences or arithmetic functions. Claims in which a specific density notion is the measure in an asymptotic statement belong here; general real-variable measure theory and claims that merely mention density in passing do not.
• METR developer productivity experiment5 claimsClaims concerning METR's August 2025 randomized controlled experiment on AI tools' effect on experienced open-source developers' productivity; covers its design, findings (e.g., slower completion with AI assistance), methodological critiques, external validity, and how its signal should be interpreted.
• Nambu–Jona-Lasinio model5 claimsThe Nambu–Jona-Lasinio model and its extensions (PNJL, PQM, variants with vector/tensor couplings) as effective theories of chiral symmetry breaking in QCD. Claims about their phase structure, couplings, mean-field treatment, or predictions for chiral restoration and critical points belong here; lattice QCD or full QCD claims do not.
• Quark-hadron continuity5 claimsThe scenario, due to Rajagopal and Shuster, that no phase transition separates hadronic matter from color-superconducting quark matter (especially the CFL phase) because their unbroken symmetry groups and excitation spectra can match continuously. Claims about whether continuity holds, its model dependence, alternatives (mixed phase, sharp transition), and astrophysical consequences belong here.
• Testimonial Knowledge5 claimsClaims about acquiring knowledge or factual beliefs through other people's assertions, reports, or testimony, including how testimonial sources compare with direct perception or other sources of knowledge. Excludes general social learning by imitation when no testimony is involved.
• Trans-Planckian problem5 claimsThe problem that derivations of Hawking radiation and analogue Hawking effects trace outgoing modes back to arbitrarily high frequencies or short distances, where the regularity of the quantum state (e.g., the Unruh vacuum across the horizon) and the validity of the underlying field theory are untested; covers regularity assumptions in derivations and proposed resolutions via modified dispersion r
• Web Content Accuracy5 claimsClaims about the factual correctness or error rate of information published in publicly accessible web content, including estimates of the share of assertions that are false. Excludes accuracy of AI-generated outputs unless the web content itself is the object of study.
• White dwarfs5 claimsWhite dwarf stars: electron-degenerate stellar remnants. Claims about their structure, cooling, atmospheres, mass-radius relation, binaries and Type Ia progenitors, and their role as targets or environments in astrophysical scenarios (e.g., interactions with high-energy particles or captured objects) belong here; neutron stars and other compact objects do not.
• X-ray Absorption Column Density5 claimsThe interstellar X-ray absorbing hydrogen column density (NH) measured toward Galactic sources from X-ray spectra, and its comparison with foreground gas and dust tracers (H I, CO, extinction) to constrain source distances and locations relative to Galactic structures such as spiral arms.
• Additive combinatorics4 claimsThe study of arithmetic properties of sets: sumsets, product sets, and their expansion under addition and multiplication over the integers, reals, finite fields, and related structures. Claims about expansion estimates (e.g. the sum-product phenomenon), Freiman-type structure, and incidence methods for such questions belong here; point-configuration geometry goes to Discrete geometry, and purely D
• AdS/CFT correspondence4 claimsThe conjectured duality (due to Maldacena, 1997) between a gravitational theory in anti-de Sitter space and a conformal field theory on its boundary, central to holography, gauge/gravity duality, and quantum gravity. Claims about its formulation, scope, proofs, tests, or physical consequences belong here.
• Aharonov-Bohm effect4 claimsThe phase acquired by a charged particle from a gauge potential along a path enclosing flux, including its occurrence and interpretation in media such as superconductors and superfluids, and debate over whether it is topological or dynamical. Claims about what charge or phase is physically measured in such setups belong here.
• American Political History4 claimsPolitical institutions, actors, events, and developments in the history of the United States. Excludes claims primarily about contemporary policy or politics unless historically framed.
• Ancient History4 claimsClaims about societies, events, and developments in the ancient world. Excludes claims focused primarily on archaeological methods or later historical periods.
• Baker's theorem4 claimsAlan Baker's theorem (1966–68) giving explicit lower bounds for linear forms in logarithms of algebraic numbers, the basis of effective transcendence and Diophantine methods. Claims about the theorem itself, its refinements and effective applications (e.g., Catalan/Pillai-type results), and comparisons of other problems' difficulty to it belong here; general Diophantine approximation without conne
• CETP inhibitors4 claimsCholesteryl ester transfer protein (CETP) inhibitors — torcetrapib, dalcetrapib, evacetrapib, anacetrapib, obicetrapib — and their trials (e.g., ILLUMINATE, dal-OUTCOMES, ACCELERATE, REVEAL), lipid effects (raised HDL, lowered LDL), cardiovascular outcomes, and safety. Not for other lipid-lowering drug classes.
• Chiral symmetry breaking in QCD4 claimsSpontaneous chiral symmetry breaking in QCD or its effective models: constituent quark masses, chiral condensates, and how chiral dynamics affects quark energetics and phase structure. Model-specific claims (e.g., NJL) also carry the model tag; pure symmetry-formalism claims go under Generalized global symmetries.
• Coin-game constant4 claimsThe constant c(n) attached to the coin-flipping game (its exact values, e.g. c(k^2) = 1/k, and its asymptotic behavior, e.g. sqrt(n)·c(n) -> 1 as n -> infinity). Claims defining, computing, or bounding this constant belong here; claims about unrelated games' constants or general probability theory do not.
• Colchicine4 claimsThe drug colchicine, a microtubule-inhibiting alkaloid with anti-inflammatory properties, used primarily for gout and under investigation for cardiovascular and other inflammatory conditions.
• Combinatorics4 claimsThe broad field of discrete mathematics concerned with counting, enumeration, extremal problems, Ramsey theory, designs, and finite structures. General combinatorics claims belong here; additive combinatorics and specific named conjectures carry their own more specific tags.
• Generalized global symmetries4 claimsGeneralized notions of global symmetry extending ordinary 0-form symmetry: higher-form (p-form) symmetries, higher-category/generalized symmetry structures, their operators, backgrounds, and realizations in quantum field theory and many-body systems. Claims about how these symmetries are defined, realized, or used to characterize or distinguish phases of matter belong here; ordinary 0-form symmetr
• Gravity crossover scale and inverse-square-law tests4 claimsThe crossover scale at which gravity on a brane changes from four-dimensional inverse-square behavior to higher-dimensional behavior in extra-dimension models; bounds from short-range tests of Newton's inverse square law (e.g. torsion-balance and Casimir-force experiments) that constrain this scale, especially in braneworld scenarios.
• Hypertension4 claimsElevated blood pressure and hypertension as a condition and cardiovascular risk factor: its measurement, determinants, and role in mediating risk of atherosclerotic events, including blood-pressure-lowering evidence. Not for blood-pressure findings unrelated to cardiovascular outcomes.
• Inference-Time Compute Scaling4 claimsClaims about how allocating additional computation during inference, testing, or deployment changes AI model performance or capability, including test-time scaling and its relation to training compute. Excludes compute used to train model parameters.
• Iron Technology4 claimsClaims about the development, adoption, spread, and historical effects of ironworking and iron tools or weapons in ancient societies.
• Multilingual LLM Reasoning4 claimsClaims about how language choice, language mixing, and multilingual ability affect reasoning processes or performance in large language models. Excludes multilingual model behavior unrelated to reasoning.
• Physical Detectability of Phase Distinctions4 claimsClaims about whether and how the distinction between two thermodynamic or quantum phases of matter can be detected by measurements on finite-energy physical states — e.g., local order-parameter observables versus topological/nonlocal diagnostics, and what observables count as physically accessible in a phase classification.
• United States Senate4 claimsClaims about the United States Senate, including its members, leadership, procedures, power relations, and historical role. Excludes claims about Congress generally when the Senate is not specifically involved.
• World War I4 claimsClaims about the First World War, including its outbreak, causes, conduct, consequences, and the decisions or perceptions of political and military leaders. Excludes other conflicts unless they are directly part of World War I.
• XTE J1814-3384 claimsThe accreting millisecond X-ray pulsar and low-mass X-ray binary XTE J1814-338. Claims about its pulsations, thermonuclear bursts and burst oscillations, orbit and companion, and inferred mass, radius, or magnetic field belong here.
• 2026 Jacobian counterexample construction3 claimsThe Jacobian-counterexample construction from 2026 involving a compactification in P¹×P² with a ramification divisor and hyperplane preimage. Claims about the construction itself, its compactification, boundary divisors, the geometry of the complement, injectivity or mapping degree, and its validity as a Jacobian conjecture counterexample belong here; general Jacobian conjecture claims that do not
• AI governance3 claimsClaims about the institutions, rules, and arrangements for governing AI development and deployment — regulation, standards, auditing and assurance regimes, liability allocation, and oversight of compliance. Broad field tag for AI policy questions; use narrower tags for particular regimes, models, or sectors where one exists.
• Atherogenic dyslipidemia3 claimsThe dyslipidemic pattern characteristic of insulin resistance and metabolic syndrome: elevated triglyceride-rich lipoproteins, low HDL cholesterol, and small dense LDL particles, studied as a mediator of atherosclerotic cardiovascular risk. Not for LDL cholesterol as an isolated exposure or for lipid-lowering drug treatment.
• Benchmark data contamination3 claimsLeakage of benchmark or test-set items into model training data, allowing memorization rather than genuine capability to inflate evaluation scores; covers detection, measurement, mitigation, and evaluations on clean or held-out data across AI task domains.
• Biomarker-disease causation3 claimsWhether elevating or lowering a biomarker causes disease outcomes, versus the biomarker being only a correlate or risk indicator; covers surrogate-endpoint concerns and claims that biomarker changes do not necessarily produce disease. Exclaims about a specific named disease mechanism or a specific biomarker tied to one disease process, which should use that disease's tag.
• Compositional Generalization in Language Models3 claimsClaims about whether language models can generalize from separately learned components or skills to novel combinations of them. Includes systematic and compositional transfer across tasks; excludes generalization unrelated to composition.
• Consider-the-opposite debiasing3 claimsClaims about deliberately entertaining reasons one's favored conclusion could be wrong ('consider the opposite' and similar counterfactual-consideration strategies) as a cognitive debiasing technique, including its effectiveness in reducing confirmation bias and overconfidence. Excludes external, institutional critique (Independent adversarial review) and the existence/manifestation of the biases
• Coronary artery disease3 claimsAtherosclerotic disease of the coronary arteries, also called coronary heart disease or ischemic heart disease, causing reduced blood flow to the myocardium.
• Dermatology3 claimsThe medical specialty concerned with diseases and conditions of the skin, hair, and nails, and their diagnosis and treatment. Use as the broad field tag for dermatological topics (e.g., acne, psoriasis, skin cancer, hair loss); not for claims in other medical specialties unless dermatologic.
• Dietary exposure confounding3 claimsConfounding of diet–disease associations by correlated dietary exposures: the difficulty of isolating a single nutrient's effect when it co-varies with other nutrients or foods (e.g., saturated fat, refined carbohydrate, overall dietary pattern) in observational research. Not for non-dietary confounders, geographic heterogeneity of associations, or confounding in non-nutrition fields.
• Dressed charge3 claimsThe physical charge of a particle once its screening cloud of gauge-field and medium excitations is included, contrasted with the bare charge; claims about which charge enters physical effects (e.g., Aharonov-Bohm phase, Coulomb interaction in media) and about how dressing is defined belong here. Not renormalization-group running of coupling constants in vacuum.
• Entanglement Swapping3 claimsExperiments that swap entanglement between particle pairs based on a post-selected joint measurement, including delayed-choice variants where the swapped particles never coexisted, and claims about what such experiments demonstrate or how their data are analyzed.
• Erdős sum-product conjecture3 claimsErdős and Szemerédi's conjecture that for any finite set A, at least one of the sumset |A+A| or the product set |A·A| is large — conjectured to be |A|^(2-o(1)) over the integers and reals. Claims about its truth in specific settings (integers, reals, rationals, finite fields), refinements of the exponent, Elekes–Szemerédi and Solymosi-type bounds, and constructions approaching the conjecture belon
• Fact-Checking3 claimsProfessional or organized efforts to assess the factual accuracy of public claims, including fact-checker rating practices, findings, and the distribution of ratings. Excludes general journalism not focused on verification.
• Fermat's Last Theorem3 claimsFermat's Last Theorem: no positive integer solutions to x^n + y^n = z^n for n > 2. Claims about the theorem, its proof (Wiles–Taylor), conditional or alternative proofs, and reformulations belong here; general Diophantine questions without a connection to the theorem do not.
• Gaps between powers of 2 and 33 claimsHow close a power of 2 and a power of 3 can get: lower bounds on |2^n − 3^m| for positive integers n, m, Pillai-type gap questions, bounds obtained via linear forms in logarithms (Baker's method), and uses of such bounds, e.g. in analyses of Collatz cycles. Claims about Baker's theorem or Catalan's conjecture themselves stay under their own tags.
• Hessian conjecture3 claimsThe Kaliman–Zaidenberg conjecture in affine algebraic geometry that a homogeneous polynomial's Hessian vanishing implies it is a coordinate (in the polynomial ring); claims about its truth in a given dimension, counterexamples (e.g. dimension five), and relations to the Jacobian conjecture belong here.
• IRAS 17287-34433 claimsThe far-infrared point source IRAS 17287-3443, identified as a post-AGB / protoplanetary-nebula candidate. Claims about its identification, distance, spectral energy distribution, dust properties, and evolutionary status belong here.
• Late Bronze Age Collapse3 claimsClaims about the widespread political and societal breakdowns around the end of the Late Bronze Age, including their causes, timing, and effects in the eastern Mediterranean and Near East.
• Latent Knowledge in Language Models3 claimsClaims about information or capabilities encoded in language models' internal parameters or representations that may not be expressed in ordinary outputs, including methods for detecting or eliciting them. Excludes whether verbal chain-of-thought faithfully reports reasoning.
• MIT bag model3 claimsThe MIT bag model of deconfined quark matter and its use in stability and equation-of-state calculations: bag constant B, quark masses, and other parameters determining whether strange quark matter (or finite strangelets) is stable or metastable relative to hadronic matter. Not lattice QCD results and not chiral effective models (e.g. NJL), which have their own tags.
• Modern European History3 claimsClaims about political, social, and diplomatic events and developments in Europe from the early modern period onward. Excludes ancient and medieval history.
• Neural Network Spectral Bias3 claimsClaims about neural networks' tendency during learning to fit low-frequency or smooth function components before high-frequency components, including its causes, scope, and implications. Excludes unrelated uses of spectral analysis.
• NP-complete problems3 claimsThe class of NP-complete problems (SAT, TSP, graph coloring, etc.): claims about whether polynomial-time algorithms for them are known or exist, their completeness reductions, and hardness consequences. General complexity-class theory stays under Computational complexity theory.
• PCSK9 inhibitors3 claimsPCSK9-targeting lipid-lowering therapies (monoclonal antibodies such as evolocumab and alirocumab, and siRNA agents such as inclisiran): their LDL-lowering efficacy, safety, and effects on cardiovascular outcomes in trials and practice, typically as add-on to statins. Not for statin therapy itself or for ezetimibe and other non-PCSK9 agents.
• Perturbative QCD3 claimsFixed-order and resummed (NLL/NNLL) perturbative calculations of QCD observables, including jet-mass and event-shape distributions, their peak/spectrum positions, and logarithmic accuracy of shower-based and analytic predictions. Not dense/thermal QCD matter (see QCD phase diagram).
• Political Science3 claimsThe study of political institutions, processes, behavior, and public opinion, including partisanship and political communication. Excludes historical claims when a period-specific history tag is more appropriate.
• Pulsation Non-detection Geometry Constraints3 claimsConstraints on neutron-star surface emission anisotropy from non-detections of X-ray pulsations: pulsed-fraction upper limits and the viewing geometries (spot latitude, observer inclination) they exclude or allow, for CCOs and other thermally emitting neutron stars.
• Reinforcement Learning Environments3 claimsSimulated or interactive task environments used to train reinforcement-learning agents, including their design, quality, availability, and commercial supply. Excludes RL algorithms and evaluation benchmarks not used as training environments.
• Riemann zeta function3 claimsThe Riemann zeta function ζ(s): its Dirichlet series Σ n^(−s), the Euler product over primes ∏_p (1 − p^(−s))^(−1) valid for Re(s) > 1, analytic continuation, functional equation, special values, zeros, and other properties. Claims about the conjecture that its non-trivial zeros lie on the critical line belong to the Riemann hypothesis tag; claims about other zeta and L-function analogues belong t
• Robotics3 claimsThe field of designing, building, and operating robots, including their hardware, software, sensing, manipulation, and locomotion capabilities.
• Robot Physical Task Capability3 claimsClaims about the ability of robots to perform physical tasks, especially occupational or manual labor, at or near human levels of dexterity, strength, and versatility.
• Scholarly Citation Practices3 claimsHow authors in scholarly publications select, use, and represent cited works, including whether authors have read the works they cite. Excludes citation counts used solely as measures of impact when citation behavior is not at issue.
• Self-Reported AI Productivity Estimates3 claimsClaims about whether workers' or developers' self-reported, survey-based estimates of AI productivity gains (time saved, perceived speedup) accurately reflect measured effects; covers perception bias and comparisons of subjective reports against objective outcomes. Excludes selection bias in who is measured and task-exposure ratings.
• Square packing constant3 claimsThe function f(n) giving the side length of the smallest square into which n unit squares (often required axis-parallel) can be packed: exact values (e.g. f(k^2) = k), bounds at n = k^2 + c, and asymptotic behavior. Claims computing or bounding this packing constant belong here; other geometric packing problems and general discrete geometry do not.
• Task Characteristics Moderating AI Gains3 claimsClaims about how task properties — ambiguity, ill-definedness, complexity, structure, or task definition — moderate how much generative AI improves performance or productivity, including claims comparing the strength of different task features as moderators. Excludes average measured AI gains, worker skill-level moderators (AI and skill gaps), and task exposure ratings (AI task exposure measures).
• Topical Minoxidil For Androgenetic Alopecia3 claimsUse of topical minoxidil as a treatment for androgenetic alopecia, including its effects on hair growth and regulatory approval status.
• Training compute trends3 claimsClaims about trends over time in the compute used to train machine learning/AI systems — both the amount of compute (e.g., FLOP for notable/final training runs) and its dollar cost, including measured growth rates (orders of magnitude per year) and the scale of the largest training runs. Excludes inference/compute demand for deployed systems and hardware price trends for their own sake.
• Vitushkin rational construction3 claimsVitushkin's 1999 construction of rational functions (in the analytic capacity / removable sets setting), treated as a precedent. Claims about the construction itself, or that other constructions — e.g. putative Jacobian conjecture counterexamples — derive from, parallel, or generalize it, belong here; general analytic capacity or rational approximation claims do not.
• Weyl algebra3 claimsThe Weyl algebra A_n, the algebra of polynomial differential operators generated by commuting variables and their partial derivatives with the canonical commutation relations. Claims about its structure, representations, or its automorphisms and endomorphisms belong here.
• 2008 financial crisis2 claimsThe 2008 financial crisis: its causes, mechanisms (housing bubble, subprime securities, ratings failures, bank failures), and consequences, including debates over which factors mattered.
• AI aggregate productivity estimation2 claimsClaims about estimating AI's economy-wide productivity or TFP gains from task-level inputs — task exposure, time saved, or task-level cost savings — including derivations via growth-accounting results like Hulten's theorem, and the assumptions, magnitudes, and critiques of such extrapolations. Excludes worker-level tool experiments (Generative AI productivity effects) and measured industry product
• AI and skill gaps2 claimsClaims about whether AI tools widen or narrow productivity or performance differences between more- and less-skilled workers (novices vs experts, entry-level vs experienced), i.e., the 'leveling' vs 'widening' question, across task domains and skill measures. Excludes average AI productivity effects and employment/displacement effects.
• AI Corrigibility2 claimsClaims about whether AI agents remain amenable to human correction or control, including whether they permit shutdown or intervention. Includes theoretical conditions and mechanisms supporting such behavior; excludes general AI governance and ordinary system reliability.
• AI development direction2 claimsClaims about the direction of AI development effort and investment: whether AI is aimed at automating existing human tasks or at augmenting workers and creating new tasks, products, and capabilities. Covers the balance and framing of this allocation; excludes measured employment, productivity, or adoption effects of AI once deployed.
• AI Development Forecasting2 claimsForecasts and scenario projections about the pace, sequence, or likely course of AI capabilities and development, and assessments of how accurately those forecasts match subsequent events. Excludes forecasts generated by AI systems about unrelated real-world events.
• AI Exposure and Employment Effects2 claimsClaims about how occupational or task-level exposure to AI/LLMs relates to employment outcomes—hiring, unemployment, and workforce composition—including measurement of exposure and its differential impact by age, wage, or automation vs. augmentation.
• AI Forecasting2 claimsClaims about AI systems producing forecasts of real-world future events and the accuracy or performance of those forecasts, including comparisons with human forecasters in competitions. Excludes surveys eliciting humans' forecasts about AI capabilities.
• AI mathematical discovery2 claimsClaims about whether AI systems — language models, agents, or specialized AI tools — can produce genuine new mathematics: proofs, disproofs, counterexamples, improved bounds, or research-valuable conjectures on open or competition-level problems. Excludes AI's productivity effects on mathematicians and general LLM capabilities outside mathematics.
• AI-resistant occupations2 claimsClaims about which occupations or job categories AI or automation will or will not displace — e.g., jobs centered on physical tasks, manual dexterity, personal care, or human contact — and the task characteristics that make occupations resistant to automation.
• AI task exposure measures2 claimsClaims about task- or occupational-level exposure, applicability, or time-saved ratings for AI/LLM tools — whether produced by human annotators or by LLMs — including their inter-rater reliability, human-LLM agreement, and predictive validity against measured productivity or quality outcomes.
• Angular Momentum Suppression of Accretion2 claimsThe mechanism by which angular momentum in the gas feeding a black hole (e.g. from stellar or orbital rotation) keeps the flow from falling directly in, reducing the accretion rate below the spherical Bondi estimate. Covers angular-momentum barriers, circulation of captured material, and suppression factors in Bondi-type accretion onto stars or compact objects; does not cover Eddington radiation-p
• Assurance deficit compensation2 claimsClaims about whether residual gaps or deficits in a safety case's assurance argument can be acceptable when offset by other arguments, evidence, or lines of reasoning — including how confidence combines across multiple complementary arguments and whether compensation argumentation is sound. Excludes object-level safety of specific named systems.
• Bochner–Riesz conjecture2 claimsThe conjecture that Bochner–Riesz means of the Fourier transform are bounded on L^p for the conjectured exponent range (p > 2d/(d+2) and its dual), equivalent to boundedness of the disc multiplier. Claims about its truth, partial results, and implications from or to it belong here.
• Burst Oscillation Pulse Profile Modeling2 claimsForward modeling of the time-dependent X-ray waveform of thermonuclear burst oscillations (rotating hot spots during Type I X-ray bursts on accreting neutron stars) to infer mass, radius, spot geometry, and inclination. Covers waveform/oblique-pulsator fitting of burst oscillations; not steady-pulsar profile modeling or thermal spectral fitting.
• Catalan's conjecture2 claimsCatalan's conjecture (Mihăilescu's theorem, 2004): the only solution of x^a − y^b = 1 in integers with x, y, a, b > 1 is 3^2 − 2^3 = 1. Claims about the statement, its proof, partial results (e.g. Tijdeman), and conditional derivations belong here.
• Client-Pays Evaluator Conflict2 claimsClaims that evaluators, auditors, raters, or assessors paid by — or otherwise dependent on — the party they evaluate face incentives to reach conclusions favorable to that party, and how this payment structure affects the independence and rigor of evaluations. Excludes: issuer-pays credit ratings specifically (own tag), sponsorship bias in research (own tag), and cognitive confirmation bias.
• Clinical Trials2 claimsHuman studies that evaluate investigational or existing medical interventions, including their design, duration, recruitment, and role in establishing safety and efficacy. Excludes non-human preclinical research and routine clinical care.
• Confinement in gauge theories2 claimsConfinement and deconfinement in gauge theories: Wilson-loop area-law criteria, center (one-form) symmetry and its unbroken/broken realization as an order parameter, Polyakov loops, and the phase structure of pure Yang-Mills and QCD-like gauge theories. Higgs-versus-confinement boundary claims go under Higgs-Confinement Distinction; finite-temperature QCD with dynamical quarks under QCD phase diag
• Conserved-charge cumulant fluctuations2 claimsEvent-by-event fluctuations of conserved charges (net-proton, net-charge, net-baryon) and their cumulant ratios measured in heavy-ion collisions, including non-critical baselines (e.g. Skellam, binomial) used to judge deviations and the model dependence of such comparisons; broad QCD phase-structure claims belong under QCD phase diagram instead.
• Convex Wolff axioms2 claimsThe variant of the Kakeya problem phrased in terms of Wolff's axioms for convex sets: whether convex Kakeya/Besicovitch sets (or their associated set of directions) satisfy Wolff-type dimension axioms. Claims about the truth or failure of the convex Wolff axioms in specific dimensions, and related convexity-based constructions, belong here; the standard (non-convex) Kakeya set conjecture does not.
• Coronavirus recombination2 claimsHomologous and non-homologous recombination in coronaviruses, including natural recombination events and their role in generating genome features such as spike insertions, cleavage sites, and chimeric viral genomes; distinct from laboratory recombination or reverse-genetic engineering.
• Cultural Transmission2 claimsClaims about the transmission of behaviors, practices, or information among individuals or agents through social learning, imitation, or related copying processes, including computational models of such transmission. Excludes biological inheritance and pathogen transmission.
• Decision Theory2 claimsThe formal study of rational choice under uncertainty, including decision rules, preference structure, and axioms for selecting among acts or lotteries. Excludes empirical studies of how people make decisions unless their formal choice-theoretic structure is central.
• Defense in depth2 claimsClaims about layered, redundant security controls (defense in depth) and whether multiple independent safeguards limit the damage of a single exploited flaw or failure — preventing one breach or compromise from causing complete system failure. Excludes claims about independent adversarial review of assessments and about non-security fault tolerance.
• Diabetes mellitus2 claimsDiabetes mellitus, including type 1 and type 2, as a population and condition in medical research: its risk factors, management, complications, and how other exposures or outcomes differ in people with diabetes. Not for insulin resistance as a mechanism itself.
• Drug Development2 claimsThe research and development process for discovering, testing, and bringing new medicines toward approval, including development timelines and bottlenecks. Excludes claims about use of an already approved drug in routine care.
• Educational Incentives2 claimsClaims about how incentives and reward systems in educational settings affect students' learning, motivation, behavior, or internalization of educational and social values. Includes extrinsic rewards that encourage strategic compliance or gaming; excludes general education policy unrelated to incentives.
• Elasticity of Substitution in Production2 claimsClaims about the degree to which inputs, tasks, or sectors in a production process can substitute for one another, including estimated or assumed substitution elasticities in aggregate production. Excludes substitution in consumption or labor-market matching.
• Erdős–Szekeres monotone subsequence theorem2 claimsThe Erdős–Szekeres theorem (1935) that every sequence of k^2+1 distinct real numbers contains a monotone (increasing or decreasing) subsequence of length k+1, and that k^2+1 is optimal. Claims about the theorem's statement, proofs (pigeonhole, poset/Dilworth arguments), its optimality constructions, and generalizations to ordered sets belong here; the related 'happy ending' convex-polygon theorem
• Expected Utility Theory2 claimsThe theory that rational choices under uncertainty can be represented as maximizing the expected value of a utility function, including its axioms, representations, and characterizations. Excludes claims about utility maximization that do not concern expected utility or uncertain outcomes.
• Expert opinion surveys2 claimsClaims reporting surveys, polls, or structured elicitations of what experts (researchers, clinicians, practitioners) believe or estimate on a question — including probability assignments, degrees of agreement, and how opinion varies by discipline. Covers the survey result itself, not the object-level question the experts are asked about.
• Financial Economics2 claimsThe study of financial systems: money, banking, asset pricing, credit markets, securities, and their regulation. Cover claims about financial instruments, institutions (banks, funds, rating agencies), market behavior, and financial crises.
• Fourier restriction conjecture2 claimsStein's Fourier restriction conjecture: the full range of L^p to L^q restriction estimates for the Fourier transform on curved hypersurfaces (sphere, paraboloid), including adjoint/extension estimates. Claims about its truth, partial results, equivalent forms, or its implications for other problems (Kakeya, Bochner-Riesz, local smoothing) belong here; generic Fourier-analysis claims without the re
• Gao tangent-sweep construction2 claimsThe construction attributed to Gao using a tangent-sweep method to produce polynomial maps with constant (non-zero) Jacobian determinant that are claimed counterexamples, with arbitrarily large geometric degree, in every dimension greater than two. Claims about the construction, its degree, injectivity, or validity in a given dimension belong here; general Jacobian conjecture claims that do not co
• Gauge-invariant correlation functions2 claimsCorrelation functions built from gauge-invariant operators (e.g., Wilson loops, meson/baryon correlators) in gauge theories, used as physical observables and phase diagnostics (e.g., confinement criteria); excludes gauge-dependent correlators and purely formal lattice-technical claims.
• General-purpose technology2 claimsClaims about what qualifies as a general-purpose technology (GPT) — a technology with broad applicability, potential for continuous improvement, and ability to spawn innovation complementarities across many sectors — and claims about specific technologies exhibiting or failing to exhibit GPT traits.
• Generative AI and employment2 claimsThe impact of generative AI technologies on jobs, hiring, skills, and workforce composition.
• Geographic heterogeneity in diet-disease associations2 claimsFindings that the strength or direction of an association between a dietary exposure and a disease outcome differs across geographic regions or populations (e.g., US vs European vs Asian cohorts). Not for confounding or measurement issues within a single population, nor for non-dietary exposures.
• Graph theory2 claimsThe study of graphs — vertices, edges, connectivity, coloring, planarity, and extremal or random graph problems. Claims about graph-theoretic theorems, conjectures, and open problems belong here.
• Identity Signaling2 claimsClaims about actions, statements, or expressed beliefs serving to communicate affiliation with or loyalty to a social group, especially when their social signaling function competes with conveying or evaluating factual information.
• Infrared dust emission2 claimsExtended infrared emission from interstellar or circumstellar dust grains heated by stellar radiation or shocks, including IR observations toward supernova remnants and molecular clouds and the identification of the heating source; claims purely about X-ray or gamma-ray emission do not belong here.
• Krasikov's difference inequalities2 claimsThe system of difference inequalities introduced by Krasikov for counting functions of integers below x whose 3x+1 (Collatz) trajectories satisfy given parity constraints, and its residue-class refinements (e.g. restricting to classes modulo 3^k), used to prove lower bounds on the density of Collatz preimage sets. Density claims about other integer sequences stay under their own tags.
• Latent Reasoning in Language Models2 claimsClaims about language models performing or being trained to perform reasoning through internal latent representations or non-linguistic computation rather than explicit natural-language chain-of-thought. Excludes latent factual knowledge that is not about a reasoning process and claims only about whether verbal reasoning traces faithfully report internal reasoning.
• Lattice QCD2 claimsAb initio numerical QCD on a spacetime lattice, including what it can and cannot compute — e.g. its failure at finite baryon chemical potential due to the sign problem, continuum extrapolations, and comparisons with model calculations of dense matter.
• Level of distribution of the primes2 claimsThe exponent θ measuring how well primes are distributed in arithmetic progressions on average over moduli q ≤ x^θ, the parameter in the Elliott–Halberstam conjecture; the known bound θ ≥ 1/2 (Bombieri–Vinogradov) and attempts to go beyond it (e.g. Zhang, Motohashi–Petri) belong here.
• Machine insight versus memory2 claimsWhether AI systems' performance on intellectual tasks stems from encyclopedic knowledge, memorization, pattern matching, patience, or brute-force search rather than genuine insight, understanding, or novel reasoning — often argued in comparison with human cognition. Covers the mechanistic basis of observed AI capability; excludes benchmark contamination specifically and productivity effects of AI
• Military History2 claimsClaims about armed forces, warfare, military strategy, technology, and campaigns across historical periods. Excludes claims about contemporary military affairs unless framed historically.
• Neutron Star Equation of State2 claimsThe equation of state of dense nuclear (baryonic) matter in neutron-star cores and its structural consequences: standard and hadronic EOS families and their predictions for neutron-star radii, maximum (TOV) mass, and tidal deformability. Quark-matter phases go under Dense Quark Matter; measured masses and radii of specific objects under Compact Object Mass and Radius.
• Noncommutative algebra2 claimsThe study of associative algebras with non-commutative multiplication — Weyl algebras, enveloping algebras, skew fields, and their ideals, modules, and endomorphisms. Claims about conjectures or results in this area belong here; commutative algebra and analysis claims do not.
• Non-statin lipid-lowering therapy2 claimsNon-statin lipid-lowering drug classes — ezetimibe, PCSK9-targeting agents (monoclonal antibodies, inclisiran), bempedoic acid, and similar — used alone or with statins: their LDL-lowering efficacy, safety, and effects on cardiovascular outcomes. Use for claims about these therapies as a class or comparatively; use a specific agent tag (e.g., PCSK9 inhibitors, ezetimibe) when a claim is about one
• Nuclear physics2 claimsThe structure, stability, and reactions of atomic nuclei (ordinary hadronic nuclei): binding energies, decay modes (alpha, beta, fission), fusion barriers, and limits on nuclear size. Not dense QCD matter, neutron stars, or quark matter, which have their own tags.
• Nuclear stability limits2 claimsWhy ordinary nuclei cannot grow arbitrarily large: Coulomb repulsion and spontaneous-fission instability set limits on viable charge and baryon number (roughly A up to a few hundred), including the nonexistence of superheavy nuclei beyond the known range and the island-of-stability question. Excludes dense/quark matter stability claims.
• Online Reader Link-Click Behavior2 claimsClaims about how readers interact with hyperlinks in online articles, including the share or rate of links they click. Excludes general web-content accuracy and clicks on advertisements or search results.
• Organizational change before productivity gains2 claimsClaims about the lag between when a new technology prompts firms to reorganize work, business processes, or organizational structure and when measurable productivity gains (e.g., aggregate labor productivity or TFP) show up in statistics — the classic GPT/productivity-J-curve pattern seen with electricity, IT, and now invoked for AI. Excludes claims about current AI adoption rates or worker-level
• Poisson conjecture2 claimsThe conjecture (Belov-Kanel–Kontsevich) that every automorphism of the n-dimensional polynomial Poisson algebra is a Poisson polynomial automorphism, i.e. the classical limit of the Dixmier conjecture. Claims about its truth, reformulations, or its stable equivalence to the Dixmier and Jacobian conjectures belong here.
• Polignac's conjecture2 claimsPolignac's conjecture: for every even natural number n, there are infinitely many pairs of consecutive primes differing by exactly n; its case n = 2 is the twin prime conjecture. Claims about the conjecture in general, about a specific fixed even gap occurring infinitely often, and related fixed-gap results belong here; bounded-gap results that fix no particular gap stay under Twin prime conjectur
• Post-AGB Stars2 claimsLow- and intermediate-mass stars that have left the asymptotic giant branch and are evolving toward planetary-nebula central stars (protoplanetary-nebula phase). Claims about their classification, distances, chemical abundances, binarity, and circumstellar envelopes belong here; AGB-phase stars, planetary nebulae, and white dwarfs do not.
• Precision nutrition2 claimsInter-individual variability in physiological responses to diet (e.g., glycemic, lipemic, blood-pressure responses) and its predictors (genetics, microbiome, baseline metabolic state), underlying the goal of individualized dietary recommendations. Covers hyper-/hypo-responder findings; not for population-average diet–disease effects or for between-country differences in associations.
• Prime gaps2 claimsThe size of gaps between consecutive primes: the average spacing near x (≈ log x, e.g. the 'about 2.3 × digit count' rule of thumb), maximal and record gaps, Cramér's conjecture and probabilistic models of gap growth, and first occurrences of a given gap size. Claims that a specific fixed gap occurs infinitely often stay under Polignac's or Twin prime conjecture.
• Prime k-tuples conjecture2 claimsThe Hardy–Littlewood prime k-tuples conjecture (Dickson's conjecture for linear forms): every admissible tuple of integers or linear forms — one with no local obstruction, i.e. not covering all residue classes mod any prime — takes simultaneously prime values infinitely often. Claims about the conjecture's truth, proof attempts, partial or conditional results, and its consequences (twin primes, bo
• Probabilistic Collatz dynamics2 claimsThe stochastic model of the 3n+1 (Syracuse) map, in particular the Syracuse random variable studied via Fourier-analytic and probabilistic methods, including claims about its characteristic function, moments, and distributional decay estimates. Excludes purely deterministic iteration claims about the Collatz conjecture itself.
• Public Belief Accuracy2 claimsClaims about the factual accuracy of beliefs held across a general population, including public knowledge, widespread misperceptions, and factors or interventions that change their accuracy. Excludes the accuracy of public-facing content itself unless population beliefs are the focus.
• Shortcut Learning2 claimsClaims about machine-learning models relying on easy-to-learn superficial or spurious features rather than robust task-relevant features, including how that reliance arises and causes failures under distribution shift.
• Social Influence2 claimsClaims about how other people, social groups, and group norms shape individuals’ beliefs, attitudes, judgments, or behavior, including conformity and group-based belief alignment. Excludes claims about factual accuracy of public beliefs unless accuracy itself is the focus.
• Task outcome measurability2 claimsClaims about whether a task has objective, verifiable outcome measures (machine-checkable rewards, ground-truth answers, agreed quality metrics) and how the presence or absence of such measures affects AI learnability or automatability of the task. Excludes benchmark contamination and occupational AI exposure ratings.
• Technology Adoption and Diffusion2 claimsClaims about how technologies are taken up and spread among individuals, firms, or institutions, including adoption rates, diffusion speed, and how these are measured or compared. Excludes diffusion of non-technological practices and claims about the productivity effects of adopted technologies.
• Undisclosed coronavirus sequence data2 claimsClaims that coronavirus (especially bat sarbecovirus) sequence data was collected or held but not published or deposited in public databases — e.g., WIV/EcoHealth holdings never disclosed, sequences deleted or removed from NCBI SRA or GISAID around late 2019, and disputes over what data was withheld or shared late. Excludes claims about already-published sequences' phylogenetic analysis, and non-v
• Abhyankar–Moh theorem1 claimThe Abhyankar–Moh theorem (mid-1970s) on embeddings of the affine line in the affine plane and the related epimorphism/irreducibility results, used to settle the two-variable case of the Jacobian problem. Claims about the theorem, its proof, its authors' work on the two-dimensional Jacobian case, and extensions belong here.
• Accretion-powered Pulse Profile Modeling1 claimForward modeling of the X-ray pulse waveform of accretion-powered pulsations (steady pulsations of accreting neutron stars, e.g. accreting millisecond X-ray pulsars) to infer mass, radius, hot-spot geometry, and inclination. Excludes burst-oscillation waveform fitting (covered by Burst Oscillation Pulse Profile Modeling) and thermal spectral or atmosphere fitting.
• Age of the Universe1 claimThe elapsed time since the universe began, including estimates, measurements, definitions, and methods of determining cosmic age. Excludes ages of individual stars, galaxies, or other objects.
• AI capital expenditure1 claimClaims about the scale and macroeconomic effects of capital spending on AI infrastructure — data centers, chips, servers, and related equipment — including its contribution to GDP growth, investment flows, and comparisons with other GDP components such as consumer spending. Excludes productivity effects of AI tools and compute cost trends for model training.
• AI Laboratory Protocol Generation1 claimClaims about AI systems generating or translating scientific goals or natural-language experiment descriptions into detailed, executable protocols for wet-lab experiments. Excludes physical execution of laboratory procedures by robots or other automated equipment.
• AI market concentration1 claimClaims about the structure, concentration, and competition dynamics of markets for artificial intelligence, especially transformative AI, and the potential economic and political consequences of market power.
• Algebraic identities1 claimStatements asserting that an algebraic identity or explicit formula holds over a specified parameter range — e.g. closed-form expressions for constants, factorizations, or values of a function in terms of its parameters. Claims here are verifiable by direct manipulation of the formula, not deep theorems; claims about named conjectures or specific constructions belong under their own tags.
• Almost Sure Global Wellposedness1 claimExtending local almost sure wellposedness for PDEs with random initial data to global-in-time solutions, notably by iterating on an invariant measure of the flow (Bourgain's method); covers the invariant-measure iteration argument itself, its hypotheses, and its applications to specific equations.
• Atherosclerosis pathogenesis1 claimThe biological process by which atherosclerotic plaques initiate and progress: LDL particle retention and modification in the arterial intima, foam cell formation, endothelial dysfunction, and smooth muscle cell involvement, including mechanistic hypotheses of plaque initiation and growth. Not for clinical ASCVD events, risk prediction, or lipid-lowering treatment.
• Atiyah's claimed Riemann hypothesis proof1 claimMichael Atiyah's 2018 proposed proof of the Riemann hypothesis, presented at the Heidelberg Laureate Forum and based on a purported analytic (von Neumann–Todd) function. Claims about its content, mathematical validity, and reception belong here; general claims about the Riemann hypothesis that do not turn on this proof do not.
• Auditor litigation risk1 claimClaims about how auditors' exposure to lawsuits, legal liability, and litigation risk relates to their conduct and outputs — audit effort, conservatism, rigor, fees, and audit quality — including auditor responses to litigation regimes and client risk. Excludes credit rating agencies and non-audit certification schemes.
• Baryon stopping1 claimHow much baryon number is transported (stopped) to mid-rapidity in relativistic nucleus-nucleus collisions: net-proton/net-baryon rapidity densities, their collision-energy dependence, nuclear stopping/transparency, and the implied baryon chemical potential at freeze-out. Cluster/coalescence production and conserved-charge fluctuation cumulants do not belong here.
• Baumol's Cost Disease1 claimClaims about the tendency of sectors with relatively slow productivity growth to gain output or expenditure share over time, and its consequences for economic structure and aggregate productivity. Includes historical evidence and explanations of this pattern; excludes unrelated cost increases.
• Canakinumab1 claimThe IL-1β-blocking monoclonal antibody canakinumab: its cardiovascular outcomes evidence (notably the CANTOS trial), effects on inflammatory biomarkers such as CRP, safety (e.g., infection risk), and use in patients with residual inflammatory risk. Not for other anti-inflammatory agents unless directly compared.
• Child Dishonesty1 claimClaims about dishonest behavior by children, including cheating and lying, and how it varies with development, beliefs, incentives, detection, or consequences.
• CLEAR SYNERGY trial1 claimThe randomized 2x2 CLEAR SYNERGY (OCEAN) trial in patients with STEMI, testing colchicine 0.5 mg/day versus placebo and spironolactone versus placebo after infarction: its design, cardiovascular event outcomes, subgroup findings, and interpretation. Not for other post-MI prevention trials.
• CM fields1 claimTotally imaginary number fields K of even degree containing a totally real subfield k such that K/k is quadratic (imaginary quadratic fields included). Claims about their structure, relative class numbers, class-number bounds, zeta/L-values, and arithmetic invariants belong here.
• COLCOT trial1 claimThe randomized, placebo-controlled COLCOT trial testing low-dose colchicine (0.5 mg/day) in patients within 30 days of myocardial infarction: its design, composite cardiovascular endpoint results, subgroup and follow-up analyses, and interpretation. Not for other colchicine cardiovascular trials such as LoDoCo2 or CLEAR SYNERGY.
• Collatz parity vectors1 claimThe parity vectors (even/odd sequences of the first k steps) of the 3n+1 (Collatz) map and the induced map from residue classes modulo 2^k to parity patterns, including its well-definedness, bijectivity, stopping-time/parity results, and the associated 2-adic encoding of Collatz dynamics. Density or probabilistic claims about the map go under their own tags.
• Complexity-theoretic proof barriers1 claimThe known formal barriers showing broad classes of proof techniques cannot settle major complexity separations: relativization (Baker–Gill–Solovay), natural proofs (Razborov–Rudich), and algebrization (Aaronson–Wigderson). Claims about what these barriers rule out, whether a technique circumvents them, and their application to separations such as P vs NP belong here; general lower-bound results wi
• Component interaction hazards1 claimHazards arising from interactions among components in complex, software-intensive systems rather than from individual component failures — emergent behavior, unsafe control actions, unintended software behavior. Covers claims about whether analysis techniques capture or miss such interaction hazards; excludes claims about single-component failure modes.
• Compute-Optimal LLM Training1 claimClaims about choosing model size and training-data quantity to minimize loss or maximize performance under a fixed training-compute budget, including Chinchilla-style scaling laws. Excludes inference-time compute allocation.
• Computer Chess1 claimChess-playing computer programs and engines, including their design, playing strength, ratings, and performance relative to human chess players.
• Condensed Matter Physics1 claimThe physics of interacting many-particle systems in solids and engineered quantum materials: electronic phases such as superconductivity, magnetism, and quantum Hall effects, and the classification of their phases. General quantum-field-theoretic or QCD-specific claims belong elsewhere.
• Conspicuous consumption1 claimConsumption motivated by social status and comparison with others (positional/conspicuous consumption) rather than intrinsic need or use. Covers claims about the drivers and consequences of status-driven spending, including whether it is wasteful or zero-sum; excludes ordinary demand, income effects, or consumption smoothing unrelated to social comparison.
• Conspiracy Theory Beliefs1 claimClaims about beliefs in conspiracy theories, including their prevalence, psychological and social determinants, and interventions that change or reduce such beliefs. Excludes conspiracies as historical events or claims about theories only as cultural narratives.
• Core-Collapse Supernova Mechanisms1 claimThe physics of core-collapse (and related conventional) supernova explosion mechanisms: neutrino-driven and magnetohydrodynamic engines, fall-back, shock revival, and what remnant (neutron star or black hole) each mechanism yields. Claims about which mechanisms can or cannot form remnants of given properties belong here; observational studies of individual remnants or measured masses of specific o
• Cosmic-ray energy scale calibration1 claimHow air-shower experiments establish and compare their energy scales: fluorescence-yield and hybrid calibration, energy-scale systematics, and cross-calibration between experiments (e.g. via assumed flux isotropy or shared spectral features) to reconcile spectrum measurements. Claims about composition, anisotropy, or individual spectra without a calibration component do not belong here.
• Credit rating agencies1 claimClaims about credit rating agencies (Moody's, S&P, Fitch), including their business models, how they rate debt and structured products, their accuracy and incentives, and their regulatory status.
• Crossing lemma1 claimThe crossing lemma: every graph with n vertices and m ≥ 4n edges has crossing number cr(G) = Ω(m³/n²). Covers the lemma's statement, proofs, constants and improvements (e.g. Pach–Tóth), sharpness constructions, and direct applications such as Székely-type bounds. Claims about named problems the lemma is applied to keep their own tags; general graph theory stays under Graph theory.
• Data center employment1 claimClaims about the number and quality of jobs created by building and operating data centers — construction employment, permanent operations staffing, and indirect or multiplier effects on local and regional labor markets.
• Davenport-Heilbronn function1 claimThe Epstein-type zeta function introduced by Davenport and Heilbronn (1936), a linear combination of two Dirichlet L-functions attached to the non-principal characters mod 5. It satisfies a Riemann-type functional equation yet has zeros off the critical line; claims about its construction, zeros, and role as the standard example that a functional equation alone does not force zeros onto the critic
• DeepMind1 claimGoogle DeepMind, the AI research lab, and the systems it builds (e.g. AlphaGo, AlphaFold, AlphaProof, AlphaEvolve) and their capabilities. Claims attributing a result, capability, or failure to a DeepMind system belong here; claims about other labs' systems do not.
• DEFUSE proposal1 claimThe 2018 'Project DEFUSE' grant proposal drafted by EcoHealth Alliance with the Wuhan Institute of Virology for DARPA, describing proposed work on SARS-related coronaviruses, including insertion of proteolytic cleavage sites and humanized-mouse experiments. Covers its described experiments, review, non-funding, and later document releases; not claims about SARS-CoV-2's actual origin.
• Dirichlet L-functions1 claimThe Dirichlet series L(s, χ) attached to a Dirichlet character χ, including their Euler products, functional equation, non-vanishing results, zero locations, and special values. Claims whose object is a Dirichlet L-function itself belong here; general prime-distribution claims that merely use them do not.
• Divergence of Prime Reciprocals1 claimEuler's theorem that the sum of reciprocals of the primes, Σ_p 1/p, diverges. Claims about the theorem's statement and proofs (Euler's, Erdős's), the refined asymptotic Σ_{p≤x} 1/p = log log x + M (Mertens), quantitative bounds and refinements, and generalizations or analogous statements for other sequences belong here; contrasting convergent sums (e.g. Brun's theorem for twin primes) belong under
• Duration of LDL-lowering exposure1 claimHow the cardiovascular benefit per unit of LDL reduction depends on how long the lower exposure is maintained: cumulative exposure ('LDL-years'), lifelong versus shorter-term exposure comparisons, and trial duration effects. Not for the lipid-lowering efficacy or safety of specific drug classes, nor for baseline LDL level as a static risk factor.
• Early Nuclear Reactor Accidents1 claimSevere accidents (core damage, meltdowns, fuel damage) at experimental, prototype, or first-generation nuclear reactors, mostly pre-dating the 1980s (e.g. SL-1, Fermi-1, Windscale, Santa Susana, NRX). Covers claims about which early reactors suffered such events and how early designs differ in accident experience from later commercial ones.
• Ecclesiastes1 claimClaims about the biblical book of Ecclesiastes, including its teachings, literary themes, and interpretation. Excludes biblical literature generally and claims about other books unless Ecclesiastes is specifically involved.
• Elementary inequalities1 claimClaims asserting a concrete inequality between explicit expressions that holds over a stated range of real or integer parameters, verifiable by elementary algebraic manipulation or direct checking (e.g. bounds like f(k,a) ≤ g(k,a) for all k≥1). Deep-theorem-dependent or asymptotic estimates, and named-conjecture claims, belong under their own tags instead.
• Evidence trust versus argument trust1 claimClaims about whether assurance-case confidence assessment methods treat trust in the evidence and trust in the argument's inferential reasoning as distinct, separately evaluated components of overall confidence, and whether these two contributions to assurance-case confidence can be separated. Distinct from the residual value of evidence when an argument fails and from confidence aggregation acros
• Evidence value in safety arguments1 claimClaims about the residual or standalone value of evidence gathered for a safety argument (safety case) when the argument's reasoning, structure, or conclusions are flawed or abandoned — e.g., whether evidence survives argument revision or failure. Distinct from general confidence in safety arguments.
• Ezetimibe1 claimEzetimibe, a selective intestinal cholesterol absorption inhibitor used as a non-statin lipid-lowering agent, typically as add-on to statins: its LDL-lowering efficacy, safety, and effects on cardiovascular outcomes (e.g., IMPROVE-IT). Not for statin therapy or PCSK9 inhibitors except when directly compared.
• Fargues–Fontaine curve1 claimThe adic-space curve of Fargues–Fontaine, whose classical points parametrize the untilts to characteristic zero of a given perfectoid field; a central object of Scholze's p-adic Hodge theory and the relative Fargues–Fontaine curves for diamonds. Claims about the curve, its construction, its points, and untilts of perfectoid fields belong here; general p-adic or algebraic geometry claims not involv
• Flamanville EPR1 claimThe Flamanville 3 EPR, France's first European Pressurized Reactor, including its construction defects (e.g. reactor pressure vessel and welds), cost and schedule overruns, the safety regulator's review, and its startup. Other EPR projects (Olkiluoto 3, Taishan, Hinkley Point C) belong under their own or a general EPR tag.
• Food substitution and net dietary effects1 claimWhether the LDL or lipid effect of adding a food (especially eggs) to the diet depends on what it displaces: added to the diet versus replacing other foods, energy balance, and the background diet as a modifier of food-level effects. Not for general egg health effects or non-substitution dietary confounding.
• Fractional Quantum Hall Effect1 claimThe fractional quantum Hall effect and its states: Laughlin and hierarchical states at fractional filling, plateaus, anyonic excitations and their statistics, edge modes, and how states at different filling fractions relate as phases. Integer QHE claims and generic topological-phase claims belong only under their own tags.
• French nuclear regulation1 claimClaims about the regulation of nuclear facilities in France, especially the Autorité de sûreté nucléaire (ASN) and its predecessor bodies: safety requirements, licensing and startup authorization decisions, inspections, enforcement, and remediation orders. Excludes claims about nuclear regulation in other jurisdictions.
• Gao's tangent-sweep construction1 claimThe construction proposed by Gao in which convex or swept bodies are built from polynomial "sweep data" via a tangent-sweep procedure, subject to algebraic side conditions (including monomial-twist conditions) on the sweep data. Claims about the construction's definition, the solvability of its side conditions, the degrees and dimensions it works in, and its soundness belong here.
• Gauge invariance1 claimWhat quantities, phases, or charges are physically meaningful in gauge theories as opposed to gauge-dependent: claims about gauge-invariant observables, physical (dressed vs bare) charges, and whether particular effects or conclusions depend on gauge choices. Not tags for specific gauge-theory formalism like Wilson-loop correlator computations.
• Gene Drive Mosquitoes1 claimClaims about developing, testing, releasing, and assessing gene-drive-modified mosquitoes, including their inheritance mechanisms and ecological effects. Excludes non-gene-drive mosquito control and gene drives in other organisms.
• Generalized Riemann hypothesis1 claimThe conjecture (GRH, or Extended Riemann Hypothesis) that all non-trivial zeros of Dirichlet L-functions, and more generally of L-functions attached to characters and algebraic number fields, have real part equal to 1/2 (the critical line). Claims about its statement, truth, partial results, and consequences of it belong here; claims about the zeta function itself belong to the Riemann hypothesis
• Generalized Vanishing Conjecture1 claimKontsevich's Generalized Vanishing Conjecture from deformation quantization: for a family of polynomials in two groups of variables, the linear-in-t terms of its Grobman-Hausdorff (Bergman) series vanish. Claims about its truth, special-case proofs, and counterexamples (e.g. in five variables) belong here; separate vanishing statements in the Jacobian conjecture literature belong to the Jacobian c
• Generic Design Assessment1 claimThe UK Office for Nuclear Regulation's (and its HSE predecessor's) Generic Design Assessment: the stepwise safety review of new nuclear power plant reactor designs before site-specific licensing, including the assessment stages, the design issues raised for resolution, and design acceptance outcomes for specific designs (e.g. EPR, AP1000, ABWR, SMRs).
• Golod–Shafarevich theorem1 claimThe Golod–Shafarevich theorem (1964): a finite pro-p group on d generators requires more than d²/4 defining relations; equivalently, a finitely presented pro-p group with at most d²/4 relations is infinite. Claims about the inequality, its refinements (e.g. Ershov–Jaikin), Golod–Shafarevich groups, and applications such as infinite class field towers belong here.
• Group schemes1 claimSchemes equipped with group structure: finite flat group schemes, Cartier duality, p-divisible (Barsotti–Tate) groups, and classification results such as Oort–Tate. Claims about their structure, order, duality, or morphisms belong here; abstract group theory without scheme structure does not.
• Group theory1 claimThe study of abstract groups: presentations by generators and relations, free, finite, profinite and pro-p groups, subgroup structure, growth, and structural theorems. Claims stating or applying results about groups belong here; claims about group schemes, algebraic groups, or algebraic-geometry objects do not.
• Happy ending problem1 claimThe Erdős–Szekeres 'happy ending' problem and its generalizations: finding convex k-gons among sufficiently many points in general position, including bounds, the packing/order-type proof approach (e.g. Seidenberg 1959, Erdős–Szekeres, Szekeres–Peters, Tóth–Valtr), empty-polygon variants, and claims about the historical origin of these arguments. The monotone subsequence theorem belongs to its own
• Higgs boson mass1 claimThe measured mass of the Higgs boson: direct determinations (ATLAS, CMS, LEP/Tevatron constraints), combined values around 125 GeV, and its role as a Standard Model parameter. Not electroweak vacuum-stability verdicts or Higgs-sector phase questions, which have their own tags.
• Hillas Rigidity Criterion1 claimThe Hillas rigidity criterion and related acceleration limits: maximum particle rigidity (E/Z ≈ eBRβ) attainable in a source of size R and field B, and comparisons of source cutoff rigidities (e.g., neutron stars, pulsars, AGN) with the rigidities of observed ultra-high-energy cosmic rays.
• Holographic UV-IR Dictionary1 claimThe holographic UV/IR mapping between radial position or infrared structure in the AdS bulk and energy/length scales in the dual CFT: hard-wall and finite-radius cutoffs, their interpretation as ultraviolet cutoffs or new physics scales in the boundary theory, and how truncating or regularizing the bulk geometry alters the dual field theory.
• Huq–Kuruvilla map1 claimThe polynomial map (x+x²y, y+xz+x²yz, z+x²z²) attributed to Huq and Kuruvilla over fields of characteristic two, with Jacobian determinant 1 and generic degree 3 that is not injective. Claims about the map's properties (Jacobian, degree, injectivity) and its bearing on Jacobian-conjecture-type questions in positive characteristic belong here.
• Hyperlink Source Support1 claimClaims about whether hyperlinks or cited sources in online articles actually substantiate the particular assertions they are attached to, including rates of unsupported or mismatched links. Excludes reader click behavior and general factual errors unrelated to source-to-claim support.
• Hyperonic matter1 claimDense hadronic matter containing hyperons (Lambda, Sigma, Xi) as in neutron-star cores: hyperon onset (the hyperon puzzle), hyperon-nucleon and hyperon-hyperon pairing and superfluidity, and the symmetry-breaking pattern of hyperonic phases. Pure quark-matter claims go under Dense Quark Matter; CFL-specific claims under Color-flavor-locked phase.
• Infinite number field towers1 claimExistence and construction of infinite towers of number fields with bounded root discriminant (class field towers, Golod–Shafarevich methods, p-extensions), including conditions such as a fixed rational prime splitting completely in the tower. Excludes finite-field or function-field tower claims and general discriminant bounds for single fields.
• Infinitude of primes1 claimThe theorem (Euclid's) that there are infinitely many prime numbers, and related claims: classical and alternative proofs (Euler's analytic proof, Furstenberg's topological proof), quantitative strengthenings, infinitude statements for primes of special forms where no specific conjecture tag applies, and analogues in other rings or settings. Statements about specific hard prime-distribution proble
• Integers with restricted digits1 claimSets of integers (or reals) whose base-b expansions use only a restricted digit set, e.g. integers with only digits 0 and 1 in base 3 or base 4, and Cantor-type digit sets. Covers density, sumset, representation, and structural questions about such digit-restricted sets; general digit or base problems without a restricted digit set do not belong here.
• Intrinsic Motivation1 claimClaims about intrinsic motivation—engaging in activities for inherent interest or satisfaction—and how it is affected by external incentives, rewards, or other conditions. Excludes motivation in educational settings when the claim is specifically about educational incentive systems.
• Issuer-pays rating model1 claimClaims about the conflict of interest where the issuer of a security rather than the investor pays for its rating, and how this payment structure affects the independence and accuracy of ratings.
• Laboratory Automation1 claimAutomation of laboratory procedures and workflows using robots or other automated equipment, including liquid handling, sample preparation, and routine experimental operations. Excludes automation of scientific reasoning or research planning when laboratory operations are not the focus.
• LDL receptor-mediated clearance1 claimThe mechanism by which LDL particles are cleared from plasma via the LDL receptor (chiefly hepatic uptake), including regulation of LDL receptor activity and expression by dietary fat, PCSK9, statins, and genetic mutations, and how this determines plasma LDL levels. Not for LDL as a cardiovascular risk factor, lipid-lowering drug efficacy trials, or arterial plaque biology.
• level of1 claim
• Lipid trait pleiotropy1 claimGenetic variants or loci that influence more than one blood lipid trait (e.g., LDL-lowering variants that also shift HDL or triglycerides): shared lipid biology and cross-trait effects of lipid-associated variants. Not for Mendelian randomization instrument validity generally, drug-induced lipid changes, or single-trait lipid associations.
• Local smoothing conjecture1 claimSogge's conjecture that solutions of the free wave equation gain local smoothing, i.e. improved L^p regularity over space-time, at the conjectured range. Claims about its truth, partial results, and implications to/from Bochner–Riesz, Kakeya, and restriction conjectures belong here.
• LoDoCo2 trial1 claimThe randomized, double-blind LoDoCo2 trial testing low-dose colchicine 0.5 mg/day versus placebo in patients with chronic coronary disease: its design, cardiovascular event outcomes, subgroup findings, and interpretation. Not for other colchicine cardiovascular trials such as COLCOT or CLEAR SYNERGY.
• Magnetic monopoles1 claimHypothetical particles carrying isolated magnetic charge, predicted in grand unified theories; covers their existence, searches, production (cosmic rays, early universe, accelerators), the Rubakov-Callan effect, and properties such as mass and charge.
• Mathematical creativity1 claimThe role of creativity, intuition, taste, and imagination in mathematical work, as practiced by humans: e.g. whether formulating conjectures demands as much creativity as proving them, the difficulty of judging which problems matter, and heuristic versus rigorous aspects of research. Excludes capability claims about AI systems (AI mathematical discovery), specific named conjectures, and claims abo
• Meng–Yang polynomial1 claimThe degree-14 polynomial in five variables constructed by Meng and Yang, with constant Hessian determinant 128, whose gradient map is claimed to be non-injective — a putative counterexample to the real Jacobian conjecture. Claims about this polynomial's properties (degree, Hessian, injectivity, simplifications) belong here; general real Jacobian conjecture claims that do not concern this construct
• Methotrexate1 claimThe antimetabolite methotrexate, especially low-dose use as an anti-inflammatory agent: its cardiovascular outcomes evidence (notably the CIRT trial), anti-inflammatory mechanism, and safety. Not for high-dose oncologic or rheumatologic methotrexate use unless in a cardiovascular comparison.
• Minimum Neutron Star Mass1 claimThe minimum possible mass of neutron stars: theoretical lower limits on gravitational/baryonic mass, whether sub-solar (~0.1-1.0 solar mass) neutron stars can be produced by any formation channel (supernova, binary accretion, mergers), and arguments against their existence. Claims that a specific observed or hypothesized compact object is unusually low-mass, or that standard formation channels can
• Model Backdoor Attacks1 claimClaims about deliberately implanting or studying hidden trigger-conditioned behaviors in machine-learning models, including triggers, training methods, effects, detection, and defenses. Includes language-model backdoors; excludes ordinary prompt sensitivity without an implanted backdoor.
• Molecular Biology1 claimClaims about the molecular basis of biological processes and experimental molecular biology, including laboratory methods involving DNA, RNA, proteins, and cells. Excludes virus-specific claims that do not concern molecular biology more broadly.
• Montgomery–Odlyzko conjecture1 claimThe conjecture (Montgomery's pair correlation conjecture, extended by Odlyzko and others) that the spacing and correlation statistics of the non-trivial zeros of the Riemann zeta function coincide with those of eigenvalues of large random Hermitian matrices from the Gaussian unitary ensemble (GUE). Claims about the conjecture's statement, evidence, numerical verification, refinements, and related
• Mosquito Eradication1 claimClaims about whether disease-carrying mosquito populations or species can be eliminated, including the technologies, feasibility, and consequences of eradication; distinct from routine control that only reduces abundance or transmission.
• NA61/SHINE experiment1 claimThe NA61/SHINE fixed-target experiment at the CERN SPS and its measurements: the energy and system-size scan over p+p, p+A, and A+A collisions, particle spectra and yields, event-by-event fluctuations and conserved-charge cumulants, and searches for the QCD critical point. Claims about other experiments (e.g. STAR, HADES) do not belong here.
• Neighboring-Wavenumber Correlation Analysis1 claimStatistical analyses that correlate neighboring (adjacent) wavenumber points in spectral correlation data — e.g., the 2016 Technion analysis — typically to test whether spectra are genuine measurements or smoothed/fabricated. Excludes unrelated spectral-peak assignments and astronomy-tagged spectral work.
• Neural Network Generalization1 claimClaims about how neural networks generalize from training data to unseen examples, including observed generalization despite overparameterization and the adequacy of theoretical explanations. Excludes specific phenomena such as compositional or spectral generalization, which have their own tags.
• Neutrino masses1 claimThe question of whether neutrinos have nonzero rest masses: oscillation evidence for mass splittings, absolute mass scale and hierarchy (normal vs. inverted), Majorana vs. Dirac nature, neutrinoless double-beta decay searches, cosmological mass limits, and whether massive neutrinos fit within the Standard Model (e.g. via Dirac vs. Weinberg operator mechanisms).
• Neutron Star Magnetic Fields1 claimSurface and magnetospheric magnetic field strengths of neutron stars: measured or inferred dipole fields, lower bounds and population distributions (from radio pulsars through magnetars, B ~ 10^8-10^15 G), and the relation of field strength to pulsar phenomenology. Claims about EOS, radii, atmosphere modeling, or pulse-profile geometry belong under their own tags even when a field value is quoted.
• Neutron Star Thermal X-ray Pulsations1 claimDetectability and properties of X-ray pulsations produced by hot spots on rotating neutron stars: oblique-rotator pulsed fractions, viewing geometry, and whether surface thermal anisotropy generically yields observable pulsations, in thermally emitting neutron stars and CCOs.
• New tasks and new products channel1 claimClaims about whether the creation of new tasks, occupations, and products has been the main channel through which technological change — especially general-purpose technologies — generates productivity gains, as opposed to automating or cheapening existing tasks. Covers the reinstatement/new-goods mechanism in historical and contemporary evidence (e.g., Acemoglu-Restrepo, Mokyr) and its applicatio
• Niacin1 claimNiacin (nicotinic acid, vitamin B3) as a lipid-modifying drug: its HDL-raising and lipid effects, safety (flushing, glycemia), and cardiovascular outcomes in trials, especially as add-on to statins (e.g., AIM-HIGH, HPS2-THRIVE). Not for niacin as a dietary nutrient in general nutrition contexts, nor for other lipid-lowering drug classes.
• Nimrod Review1 claimThe 2009 Haddon-Cave Review into the 2006 Nimrod XV230 crash, covering its findings on the Nimrod safety case, the UK Ministry of Defence's airworthiness regime, safety-culture failures, and subsequent regulatory reforms; claims about what the Review found or criticized belong here.
• Nimrod safety case1 claimThe Nimrod aircraft safety case, including the Nimrod MR2/XV230 accident at Kandahar in 2006, the Haddon-Cave inquiry's examination of the safety case, the roles of the UK Ministry of Defence and contractors such as QinetiQ as independent safety advisor, and claims about whether the safety case was properly produced, reviewed, or assured.
• Nonextensive statistical mechanics1 claimTsallis/nonextensive statistical mechanics: q-Gaussians, q-triplets, and power-law distributions as applied or tested in laboratory and natural systems, including whether measured distributions or correlation functions are consistent with the nonextensive formalism.
• One-way functions1 claimThe open question whether one-way functions exist: functions that are efficiently computable but hard to invert on average. Covers the existence question, its equivalence to other primitives (pseudorandom generators, private-key cryptography), and candidate constructions (factoring, discrete log); excludes general polynomial-map invertibility (Jacobian conjecture tags) and specific factoring-algor
• Online gig economy1 claimFreelancers and independent contractors working through online platforms and marketplaces (e.g., Upwork, Fiverr, Freelancer), covering their earnings, jobs, hours, and occupational composition on such platforms. Claims that turn on platform-based gig labor rather than traditional employment; excludes general AI-labor findings that do not involve gig or freelance platform workers.
• PDE Wellposedness1 claimLocal and global wellposedness (existence, uniqueness, continuous dependence on data) for evolution PDEs, including probabilistic or almost-sure notions of wellposedness and questions of solution extension, iteration, or blow-up. Claims about specific conjectures or particular equations may also carry a narrower tag.
• Pioglitazone1 claimPioglitazone, a thiazolidinedione insulin sensitizer used in type 2 diabetes: its efficacy, safety, and effects on clinical outcomes, especially cardiovascular outcomes trials (e.g., PROactive, IRIS) and heart failure or fracture risks. Not for other thiazolidinediones unless directly compared.
• Plasma physics1 claimThe physics of ionized gases, including laboratory (especially magnetized) plasma experiments, plasma turbulence, waves, and fluctuation/correlation measurements. Claims about measurements or theory tied to plasma systems belong here; astrophysical plasmas also go under their astrophysics-specific tags.
• Polynomial hierarchy1 claimThe polynomial hierarchy PH, the union of the levels Sigma_k^p, Pi_k^p, and Delta_k^p built from NP and coNP with polynomial-time oracles. Claims about its levels, its collapse (to P, to a level, or conditionally), and its relations to classes such as PSPACE belong here; general complexity-class theory without PH as its subject stays under Computational complexity theory.
• Polynomial Wolff axioms1 claimWolff's axioms in their polynomial form: the family of dimension conditions on a collection of delta-tubes (with delta-separated directions and controlled overlaps) used in polynomial-partitioning approaches to the Kakeya problem. Claims that a given configuration of tubes satisfies or fails these axioms, in any dimension, belong here; the convex-set variant goes under Convex Wolff axioms.
• Prime divisor1 claimThe prime divisors of integers: the theorem that every integer greater than 1 has a prime divisor, smallest prime factors, the set of primes dividing a given integer, and related elementary divisibility facts. Unique factorization itself, and prime-distribution questions such as gaps or counting functions, belong under their own tags.
• Productivity slowdown1 claimClaims about the slowdown in aggregate labor productivity and TFP growth observed since roughly 2005 in the US and other advanced economies: its extent, timing, and candidate explanations such as mismeasurement, declining business dynamism, slower innovation, composition shifts, or fading ICT gains. Excludes the post-2022 US productivity acceleration and worker-level AI tool effects.
• Proton decay1 claimThe predicted decay of protons into lighter particles, violating baryon number in grand unified theories; covers experimental search limits (Super-Kamiokande, etc.), theoretical predictions, and processes catalyzing or inducing proton decay, such as the Rubakov-Callan effect of monopoles on nucleons.
• Random matrix theory1 claimThe mathematical study of ensembles of random matrices (especially Hermitian ensembles such as GUE, GOE, GSE): eigenvalue distributions, spacing and correlation statistics, universality, and applications of these statistics to other systems. Claims about matrix-ensemble eigenvalue statistics belong here; claims about zeta zeros stay under their own tags.
• RBMK reactor design1 claimThe Soviet RBMK graphite-moderated, water-cooled channel power reactors (including the Chernobyl units): design characteristics such as the positive void coefficient and slow scram, design flaws, pre-accident safety analyses and risk assessments of the design, and post-accident modifications. Not claims about PRA methodology in general or nuclear regulation in other countries.
• Repeated distances problem1 claimThe repeated (popular) distances problem: lower bounds on how often a single distance occurs among pairs of a finite point set, including guarantees that every large subset contains a popular distance, constructions (e.g. lattice-based) attaining such bounds, and the dual relation to the Erdős distinct distances problem. Claims only about counting unit distances or the maximum number of distinct d
• Root discriminant1 claimThe root discriminant of a number field, d_K = |disc(K)|^(1/n), and explicit bounds or lower/upper estimates involving it (e.g. bounds on class numbers, ramification, or towers of fields in terms of root discriminant and degree).
• Safety case producer independence1 claimClaims about who produces a safety case or assurance argument — the developer or operator of the system being assessed versus an independent third party — including norms and regulatory expectations on self-produced safety cases and the implications of authorship by the assessed party. Not about payment incentives, review quality, or any one named system's case.
• Sarbecovirus codon usage1 claimComparative codon usage in sarbecoviruses, especially the arginine codons (CGG vs CGT/CGC/CGA/AGA/AGG) in the SARS-CoV-2 furin cleavage site insert and their rarity in related bat sarbecoviruses; used in arguments about the site's natural versus engineered origin. Not the broader presence/absence or function of the cleavage site itself.
• Senate Majority Leadership1 claimClaims about the powers, practices, and historical exercise of the Senate majority leader's role. Excludes leadership roles in other legislative chambers and Senate leadership unrelated to the majority leader.
• Separable function field extensions1 claimWhether an extension of function fields — e.g. k(x1,...,xn) over k(f1,...,fn) induced by a polynomial or rational map — is separable or inseparable, including Jacobian-determinant criteria for separability and the behavior of such extensions in positive characteristic. Invertibility claims about polynomial maps belong under the Jacobian-conjecture tags; number-field separability claims (CM fields,
• Space Shuttle program1 claimNASA's Space Shuttle program (STS, 1981-2011): the orbiter fleet, launch operations, and its accident and reliability record, including the Challenger and Columbia losses and pre-accident failure-risk estimates. Covers claims about the Shuttle's design, risk estimates, and failure history; excludes other launch vehicles.
• Sponsorship bias in research1 claimClaims about whether and how a study's funding source — especially commercial or industry sponsorship versus independent, public, or no external funding — relates to its design, reported results, or stated conclusions. Excludes generic publication bias, cognitive confirmation bias, and research-integrity issues not tied to funding source.
• SREBP-2 cholesterol synthesis regulation1 claimCellular regulation of cholesterol synthesis and uptake through the SREBP-2 pathway, including sterol-mediated suppression of HMG-CoA reductase and related biosynthetic enzymes, SCAP/INSIG sensing, and feedback control of intracellular cholesterol levels. Not for dietary cholesterol, plasma LDL clearance, or lipid-lowering drug therapies.
• Statistics1 claimThe discipline of statistical methods for collecting, modeling, analyzing, and interpreting data, including traditional statistical approaches to prediction and optimization. Excludes machine-learning topics unless statistical methodology is central.
• Stellar Remnant Longevity1 claimObserved longevity and survival of stellar remnants: evidence that old white dwarfs, neutron stars, and other compact remnants persist for billions of years, including ages of the oldest such objects and use of their survival as an argument (e.g., against rapid decay or unseen destruction mechanisms). Not specific-object masses, radii, or spectra.
• Strangeness distillation1 claimThe mechanism (Greiner et al.) whereby preferential emission of kaons carrying antistrangeness from a baryon-rich quark-gluon plasma during hadronization leaves the residual plasma enriched in net strangeness, proposed as a route to strangelet formation. Covers the mechanism, its assumptions, and its efficiency; general kaon spectra or strangeness-enhancement measurements in heavy-ion data do not.
• String theory1 claimThe branch of theoretical physics that models elementary particles and their interactions as one-dimensional extended objects (strings), encompassing dualities, compactifications, extra dimensions, holography, and the AdS/CFT correspondence.
• Sum of two squares1 claimThe representation function r₂(n) counting representations of an integer as a sum of two squares: the sum-of-two-squares theorem (characterization via primes ≡ 3 mod 4), Landau–Ramanujan-type counting of representable integers, upper and lower bounds for r₂(n), and integers with many representations. Claims about other additive representations (Goldbach-type sums of primes) stay under their own ta
• Telescope Array1 claimThe northern-hemisphere ultra-high-energy cosmic-ray observatory in Millard County, Utah, combining a surface detector array with fluorescence telescopes. Claims about its measurements, results, performance, and interpretation (spectrum, composition, anisotropy, photon and neutrino limits) belong here; results from other experiments such as Auger do not.
• Tempered fundamental groups1 claimTempered (adic) fundamental groups of hyperbolic curves and schemes over p-adic and adic base fields, including their dependence on the choice of geometric base-point and the special role of algebraically closed perfectoid base fields; central to anabelian reconstruction arguments in Mochizuki's IUT and Kirti Joshi's p-adic Teichmuller constructions. General group-structure claims without this ana
• UHECR mass composition1 claimThe elemental/mass composition of ultra-high-energy cosmic rays (roughly above 10^17-10^18 eV): proton vs. heavier-nuclei fractions, composition inferences from Xmax and muon content by experiments such as Auger and Telescope Array, and the light-vs-heavy composition debate above the ankle. Claims about composition at lower energies (PeV knee region) or about sources and propagation do not belong
• Weil conjectures1 claimThe Weil conjectures (1949) on zeta functions of varieties over finite fields: rationality, functional equation, and the analogue of the Riemann hypothesis (Frobenius eigenvalues of absolute value q^(n/2)), proven in full by Deligne (1974). Claims about the conjectures, their proofs, analogues for other function-field zeta/L-functions, and their applications belong here; the classical Riemann hypo
• Welfare economics1 claimThe branch of economics studying allocative efficiency, social welfare, externalities, and whether market or individual choices produce socially wasteful or zero-sum outcomes. Covers claims about deadweight loss, positional externalities, and efficiency comparisons of allocation mechanisms; excludes pure positive macroeconomic topics like growth accounting or labor-market measurement.
• Ziltivekimab1 claimThe IL-6-blocking monoclonal antibody ziltivekimab: its cardiovascular outcomes evidence (notably the ZEUS trial in CKD patients with elevated hsCRP), effects on inflammatory biomarkers such as hsCRP and IL-6, safety, and development status. Not for other IL-6 or anti-inflammatory agents unless directly compared with ziltivekimab.