Lines of inquiry — A named line of the record’s own argument, so that a line can be set aside as a whole and what depends on it can be found. Retracting a line is a status change on its members, never a deletion (ADR-032).
Formal apparatus — results that hold whatever any substantive line concludes — gluing and contextuality, comparison of experiments and probes, identifiability, conservation, and thresholds for naming a unit; held as a second line by the general results other lines borrow, so that setting a substantive line aside does not retire them (ADR-040).
123 of 547 CLAIM documents. Back to the full index.
| # | Title | Summary | Status |
|---|---|---|---|
| CLAIM-011 | Intertwining with an uninformative correspondence is vacuous or lossy, so the source–target correspondence must be anchored to independently measured distinctions and evaluated on tasks other than those it was estimated on | The assistant’s own objections at A34 §3 and A40 §6.4, with the remedy of A40 §VIII. Recovered: the manuscript keeps the intertwining identity but states neither limit nor remedy, apparently by inadvertence. | Active |
| CLAIM-037 v4 | Formal contextuality is a failure of global extension, distinct from ordinary context dependence, and needs Contextuality-by-Default when marginals shift with context | The manuscript’s §3 and §9. It separates three things: context dependence, stochastic noncommutativity, and contextuality as an obstruction to a global section. Its note of 2026-10-29 names the version it should cite: CbD 2.0 (LIT-831). | Proposed |
| CLAIM-040 | Symmetry is a kind of invariance, and Noether’s theorem is the point: invariance under transformation yields conservation | The owner at U37, quoting A102’s “We do not require the structure to remain invariant”: “symmetry is a kind of invariance, no? isn’t that part of the point of Noether Theory?” Conceded at A105. Its Noether half was limited by the assistant in the same reply. | Active |
| CLAIM-041 v2 | Context-sensitive interpretation, order effects and noncommuting framing operations do not by themselves establish nonclassical probability or quantum contextuality | Built over A21–A30, against the assistant’s quantum move and then the owner’s U13. The manuscript §9 states it. Its strongest ground, the record’s THEORY-013, is not cited in the manuscript. | Active |
| CLAIM-043 | Equal entropy does not entail equal relational structure, and a collection of local entropies need not characterize global compatibility | A86 §4. The manuscript §5 keeps its example: perfect correlation and perfect anticorrelation of two fair binary observables have equal entropy and reverse a fundamental relationship. | Active |
| CLAIM-044 v2 | Ambiguity and contextuality are different phenomena: an ambiguous utterance admits several compatible global assignments, a contextual one admits none | A81 §3.4 and A82 §8.7, recovered. The manuscript keeps the contrast between context sensitivity and contextuality but not this one, between too many global readings and none. Dropped at C6 without critique. | Proposed |
| CLAIM-069 v3 | The 2016 Contextuality-by-Default criterion was being replaced by multimaximal couplings as it appeared, and the change matters for observables in more than two contexts | From the reading of Dzhafarov and Kujala 2016 (NOTE-600). The manuscript cites the 2016 paper; its general observational covers need the later version. | Superseded — by ARG-022 |
| CLAIM-070 | Transport induced by one global stochastic kernel carries a global extension of the source model to a global extension of the target model | Proposition II of A84, Proposition B2 of C6, the manuscript’s Proposition 2. Without a common global kernel preservation is not automatic, which leaves open whether transport can create or destroy contextuality. | Active |
| CLAIM-079 | Stability does not entail viability: a system can occupy a stable regime while losing access to trajectories that keep satisfying its constraints | Applied from viability theory (THEORY-134) and the stability/resilience distinction (THEORY-132) (Section 6.3). | Active |
| CLAIM-093 | Not every invariance of a stochastic or communicative process implies a Noether conservation law: the symmetry principle can be borrowed, a conservation analogue has to be derived | A105, limiting the owner’s U37 appeal to Noether in the same reply that conceded the rest. The manuscript §4 and §11: Noether-type results give “conservation criteria for particular stochastic processes, not a universal conservation law of meaning”. | Active |
| CLAIM-094 | A harmonic observable conserves only its mean, and Baez and Fong show that mean conservation for every initial state does not make an observable a symmetry | From the reading of Baez and Fong (NOTE-598). The manuscript’s Kf = f is the mean-only condition, which the cited theorem’s counterexample shows is weaker than commuting with the generator. | Active |
| CLAIM-107 v3 | Under Markov reconstruction kernels, an observable’s expectation is invariant when the observable is harmonic (K f = f), and drifts by at most Σ ε_i when ‖K_i f − f‖∞ ≤ ε_i; on a finite irreducible chain only constants are harmonic | The manuscript’s §4, citing Baez and Fong. CLAIM-094 says what the cited theorem needs that the discrete-time criterion leaves out. | Active |
| CLAIM-121 | A transport whose local kernels commute with restriction maps overlap-consistent source models to overlap-consistent target models | Proposition I of A84, Proposition B1 of C6, the manuscript’s Proposition 1. It concerns marginal consistency, not global noncontextuality. | Active |
| CLAIM-123 v6 | In conditional generation, adding scores composes two conditions exactly when the ratio p(y₁,y₂|x)/(p(y₁|x)p(y₂|x)) does not vary with the noisy state x, which conditional independence of the conditions at that state is sufficient for but does not require | The manuscript’s §7, citing Composable Diffusion, T2I-CompBench and Schrödinger bridges. The readings support it, and the manuscript is more careful than the paper it cites. | Active |
| CLAIM-131 | A compatible family of sections of the event sheaf always glues, so where a communicative object lacks a global realization the obstruction lies in its supports or distributions, not in the event sheaf | The assistant at A125 §2 and A129 §2, after the owner’s U41 “double click on this”. It is the standard sheaf property of E(U) = ∏ O_x (Abramsky and Brandenburger). Granted, because it is elementary; it is used against TERM-002’s wording, not against CLAIM-038’s thesis, which version 2 restates to agree with it. | Active |
| CLAIM-132 v2 | Characters classify finite-dimensional representations of a finite or compact group up to isomorphism only over a field of characteristic zero; they forget the isomorphisms, fix neither the group nor a permutation action, and classify kinds, not tokens or positions | Standard representation theory, which the exchange states correctly with its main hypotheses (A194 §1, A198 §§2 and 4, A203 §§5–6, A218 §3.2). This entry adds the limits the exchange left out, and the exact link to Yoneda: a character is a counts-only relational profile, complete because the category is semisimple. Granted, because it bounds the owner’s analogy: priority for classifying kinds up to isomorphism, in the most favourable setting, and no further. | Active |
| CLAIM-141 v2 | Causal-loop consistency and sheaf contextuality are both local-to-global problems, but passing to distributions affects them oppositely: a loop with no deterministic solution has a distributional fixed point, while a model with no global section in its supports has no global distribution either | A161, after the owner’s U46, sharpened. A161 saw that a causal loop and a contextual model are “not the same”, but not that they run in opposite directions under passage to distributions, or why. Its parity pattern is LIT-278’s identification of Liar cycles with contextual support tables, which the record holds and has read (NOTE-251). A173 and A178 kept the three kinds of global consistency apart; A218 Ch10.9 keeps the feedback case. | Active |
| CLAIM-144 | Inertial observers who disagree about temporal order differ in coordinates, not in a failure to glue: only spacelike-separated events change order between frames, and frames are global charts related by a group action | A164’s correction of the owner’s U47, which is standard special relativity, with two repairs. A164 left out the reading of the owner’s scenario whose answer is invariant (two passings of one marker). And it described frames as patches glued by a cocycle, which they are not: each inertial frame charts all of spacetime, and the frames are related by the Poincaré group. | Active |
| CLAIM-148 v4 | A matrix factorization is identified only up to an invertible change of latent coordinates, which weight decay narrows to rotations and nonnegativity to rescalings and permutations plus a data-dependent remainder; such a gauge is not a symmetry of the relational system, and automorphisms of the data act on the factors only when the factorization is identifiable | Drawn by the assistant at A207 §3, made explicit at A214 §6 and kept in A218 (§4.2, Ch5.4–5.9). Granted, because it is elementary, and the record already held its mathematics. The record adds the full list of invariants, the data-dependent freedom of nonnegative factorization, and why identifiability must come before automorphisms and characters. It does not say latent axes are meaningless, or that observed data have non-trivial automorphisms. | Active |
| CLAIM-152 v6 | Any individuation built on sufficiency is relative to the target it is sufficient for: fix the outcomes differently and a different partition wins, so no boundary drawn this way is universal | Dual-filed from MOM-POS-001 (ADR-037). A sufficient statistic is sufficient for a target Y, so every partition chosen as minimal sufficient inherits Y. MOM’s example: fix the outcomes differently and “organism” wins instead of “animal”. In this record it is the standing caution on every boundary criterion filed beside it (CLAIM-154, CLAIM-161). It does not say boundaries are arbitrary, only that each is relative to a stated target. | Active |
| CLAIM-154 v3 | A candidate structure is worth naming as a unit when its effect-to-noise ratio outruns the Occam cost of keeping it: reify when d’² exceeds the log prior-to-posterior variance ratio, where the model with the candidate stays regular at the parameters that switch it off | Dual-filed from MOM-CLM-160 (ADR-037). By the Savage–Dickey ratio, ln BF(keep) = ½(d’² − Occam), so a direction, block or sector is real enough to name exactly when δ/ε > τ with τ² = ln(s²/σ²), the information the data added about it. In this record it prices the boundary tests for a combined form: a near-decomposable block (LIT-905) earns a name when its gap beats its Occam factor. Re-checked here in the conjugate case. Narrowed to candidates whose model stays regular where they are switched off; where it is singular the rule overcharges (CLAIM-188). | Active |
| CLAIM-164 v4 | What a spectral gap makes definite is the subspace of the cluster it isolates, not a single direction: the gap picks the object, and the object is a vector only when its own gap is wide | Dual-filed from MOM-CLM-144 (ADR-037). Under a perturbation δT, the subspace spanned by a cluster of singular values separated by a gap δ rotates by at most ‖δT‖/δ (Davis–Kahan), whatever happens to the vectors inside it. So the well-defined unit is the gap-isolated projector. Re-measured here. With CLAIM-154 it is a criterion: the gap picks the object, and the Occam factor says whether the gap is big enough. | Active |
| CLAIM-188 v4 | Where the model with a candidate structure is singular at the parameters that switch the candidate off, with real log canonical threshold λ below d/2 and multiplicity m, the reification rule’s Gaussian Occam term overcharges keeping the candidate by (d/2 − λ) log n + (m − 1) log log n to leading order, so near the threshold the rule prunes candidates the Bayes factor keeps, and its derivation holds only where that model stays regular | A narrowing of CLAIM-154, filed on 2026-10-11 from a synthesis review that found the record already held the readings that limit it. CLAIM-154’s rule prices a candidate with a Gaussian Occam factor, which is exact in the conjugate case it was re-checked in. The candidates it is meant for (blocks, sectors, couplings with hidden structure) are often singular where they are switched off. There THEORY-087 puts the penalty at λ log n − (m − 1) log log n, not (d/2) log n. On CASE-152’s product model the rule prunes, near the threshold, candidates the exact Bayes factor keeps. CLAIM-154 is narrowed to models that stay regular at the switch-off. | Active |
| CLAIM-197 | Sections over contexts that do not overlap always glue, so a sense that differs between disjoint contexts is not a gluing failure: a failure needs an overlap carrying sections that disagree, and which families count as covers decides whether anything glues | Dual-filed from MOM-CLM-088 (ADR-037). The sheaf condition says a family of sections that agree on overlaps glues to one global section. Where two contexts do not overlap, agreement is vacuous and they glue. So “tall”, which takes one threshold for people and another for trees, is one global section with the comparison class part of the meaning, not a failure of gluing. A failure needs contexts that overlap and carry sections that disagree there. Which families count as covering is the piece that decides whether anything glues, and a sheaf picture of meaning that does not state it has decided nothing. | Active |
| CLAIM-209 | Probes and statistical models recover only the distinctions identifiable from the relational evidence they are given, and a signature sufficient for a stated family of communicative decisions outside the observed probes can be identified when the full latent factorization is not | The assistant’s second thesis at U62 (A240 T2), the positive answer to QUESTION-027. It has now been asserted four times: A214 §7, A218 (RQ3 and the priority theorem target), A232 (§3.2 and F2) and A240. Filed with QUESTION-027’s scope: the unrestricted claim is answered by standard results, so its content is in decisions outside the observed probes. No signature is yet named. A proposal; the owner argued none of it. | Proposed |
| CLAIM-245 | Under Contextuality-by-Default, a deterministic context-wise recoding can make a noncontextual signalling source maximally contextual, or turn a contextual source into pure signalling, so per-content signalling is not monotone under such maps and the quantity to bound is joint overlap discrepancy | Found in the record’s audit of the line on 2026-10-10 (formal lens); no turn of the exchange raises it. Two worked rank-2 systems answer NOTE-654’s open question negatively: THEORY-177’s per-content signalling can rise under deterministic recodings, once the source signals, and CbD contextuality can be created or removed by them. By data processing, the joint overlap discrepancy cannot rise, so that is the quantity CLAIM-125’s second part should name. Granted, because the examples are elementary. It does not say THEORY-174 is wrong, which is about no-signalling models. | Active |
| CLAIM-260 | Approximate transports compose into a category only if the distortion obeys a triangle inequality and later kernels are nonexpansive, so the semigroup claim’s “category or semigroup” holds for exact kernels, and for approximate ones only under those conditions | Found in the record’s audit of the line on 2026-10-10 (formal lens); no turn of the exchange raises it. TERM-008’s founding sense is a transport that gets “within some eps”, and CLAIM-106 makes fidelity graded. Kernels always compose, but the composite of an ε₁-transport and an ε₂-transport is an (ε₁ + ε₂)-transport only if the distortion obeys a triangle inequality and the second kernel is nonexpansive. In total variation both hold. QUESTION-017 holds the triangle inequality open and does not tie it to the category claim. Granted, because it is elementary. It does not say the category claim fails for exact transports or in total variation. | Active |
| CLAIM-264 v2 | Two objects can agree on every probe in a restricted family without being isomorphic, so equivalence under a probe family is coarser than identity unless the family is rich enough | The general content of CLAIM-142, filed alone under ADR-040 so that it does not rest on CLAIM-046 and CLAIM-011, which are about pragmatic identity. The restricted Yoneda functor is fully faithful exactly when the probe family is dense. A strong generator makes it conservative, which is weaker: a morphism the probes see as invertible is invertible, but objects with isomorphic profiles need not be isomorphic. Matching every probe’s marginal leaves the joint unconstrained. The evidential rule is CLAIM-322’s. | Active |
| CLAIM-265 v3 | Local descriptions that disagree are not thereby incompatible: whether they obstruct a joint description is decided only relative to a specified class of admissible correspondences between their observations, and a correspondence admitted only because it removes the disagreement decides nothing | The general content of CLAIM-143, filed alone under ADR-040 so that the order of tests does not rest on CLAIM-011, CLAIM-004 and CLAIM-035, which are about translation. Disagreement between local descriptions sorts into a coordinate difference, a structural transformation or a global obstruction, and the sort cannot be made until a class of admissible transition maps is given. Inertial frames (CASE-039) show disagreement with no obstruction once the maps are known. | Active |
| CLAIM-266 v2 | A family of probes that supplies an optimized system’s training signal cannot also certify what the system preserves: optimization pushes toward whatever the probes reward, including what they cannot see, so preservation must be shown under probes not used to train it | The general content of CLAIM-149, filed alone under ADR-040 so that it does not rest on CLAIM-011’s account of correspondences in translation. Training against a probe family pushes a system toward whatever the family rewards, including differences it cannot see, so the family is not a certificate of what was preserved. CASE-041 is a documented instance: a Projected FastGAN trained against ImageNet features matches StyleGAN2’s FID, also computed in ImageNet features, while doing worse in CLIP space and with human judges. | Active |
| CLAIM-267 v3 | Per-step discrepancies between a chain of maps and a reference chain compound through the sensitivity of the later steps, so the same local error can be damped, accumulated or amplified, and closeness at every step can coexist with large end-to-end divergence | The general content of CLAIM-056, filed alone under ADR-040 so that the line-neutral recursion does not rest on CLAIM-106’s account of fidelity. It is elementary: with e_(i+1) ≤ L_i e_i + ε_i, the end-to-end error is bounded by the discrepancies weighted by the later Lipschitz constants. Three regimes follow. For stochastic kernels in total variation, contraction coefficients do not exceed one, so amplification needs another metric or a nonlinear, adaptive state. | Active |
| CLAIM-268 v2 | Model stitching certifies an order of information only in its reconstruction form: a map fitted to reproduce the target layers shows the source representation is at least as informative for every decision on the label, task-loss stitching compares one decision problem through a fixed head, and a failure to stitch is no evidence against sufficiency | The general content of CLAIM-126, filed alone under ADR-040 so that the result about stitching does not rest on CLAIM-050’s account of fidelity. Reconstruction-fitted stitching makes the target a garbling of the source given the label, which certifies one direction of Blackwell sufficiency (THEORY-176). Task-loss stitching (LIT-850) compares one decision problem through a fixed head, orders neither representation, and is not invariant under reparametrisation. In both forms a failure says only that no map in the class works. | Active |
| CLAIM-269 v5 | A regularized linear readout can tell apart only representations that differ in normalized kernel, and kernels fix a representation only up to the symmetry of what is observed, so no such readout supplies a distinction beyond that symmetry | The general content of CLAIM-193, filed alone under ADR-040 so that the linear result does not rest on CLAIM-142, whose realist reading is about communicative identity. What a regularized linear readout decodes is a function of the normalized kernel (THEORY-008). Kernels fix a representation only up to the symmetry of what is observed (THEORY-004, THEORY-723). An inner product beyond that is a stipulation (THEORY-018). And spectral degeneracy does not identify the symmetry (THEORY-019). These are proved results for learned representations. | Active |
| CLAIM-270 v2 | Retaining information about where a Markov chain started and staying close to the starting distribution are different properties: a chain can forget its initial state while its distribution stays far from where it began, so convergence is not preservation and losing information about the origin is not degradation | The general content of CLAIM-081, filed alone under ADR-040 so that the distinction does not rest, through CLAIM-056, on CLAIM-106’s account of fidelity. I(U_0; U_n) and D[P(U_n), P(U_0)] answer different questions. A contractive chain loses information about its initial state while converging to a stationary distribution that can be far from the start. The data-processing inequality bounds the first and says nothing about the second. | Active |
| CLAIM-282 v2 | Forbidding self-endorsement does not stop an operator from amplifying its own authority: under damped recursive delegation, reciprocal and cyclic endorsements recycle authority, so an operator’s authority depends on its own endorsements, while one-hop allocation within an epoch, from budgets and baselines fixed independently of outgoing endorsements, makes it exactly independent of them | A2’s own analysis of the owner’s U2 design, conceded in the exchange and checked here. With damped recursive delegation (a PageRank-like fixed point), two operators that endorse each other pass authority back and forth, so each one’s authority depends on its own outgoing endorsements even though neither endorses itself. With one-hop allocation from budgets fixed independently of outgoing endorsements, that dependence is exactly zero. No accounting rule removes the slower route: an operator can change its authority by changing what others endorse. | Active |
| CLAIM-289 | Non-stochastic strings stay non-stochastic under total recursive models: their minimal randomness deficiency agrees with the finite-set one to within O(log n) in argument and value | The step that THEORY-260’s promotion (version 2) names as the record’s own: carrying LIT-239’s non-stochastic strings, proved for finite-set models, to total recursive function models through Propositions B.1(b) and B.2(b) with Remark B.3. Remark B.3 asserts wholesale that every result of the paper extends to total functions up to a logarithmic term. This writes the extension out for this one result. It holds at O(log n) precision only, and does not reach LIT-229’s sophistication, which is defined at constant precision. | Active |
| CLAIM-290 | A positive-definite kernel’s minimal realizations are unique up to a unitary, and two finite representation matrices with the same centred Gram matrix differ by a translation and an orthogonal map | The record’s own proofs of two steps that THEORY-004 rests on and that its sources state without proof. (a) Jorgensen & Tian’s Cor. 1.35: minimal realizations of a positive-definite kernel are unique up to a unitary. The book leaves this “as an exercise”, and NOTE-829 supplied the book’s own proof of Thm 4.35, transferred. (b) The finite case that LIT-250 §5.3 states as known: XXᵀ = YYᵀ implies Y = XQ with Q orthogonal, with no assumption that the rows span R^d. With centring this becomes a translation and an orthogonal map. It does not say Q is unique, which fails when the rank is below d. | Active |
| CLAIM-291 | The adjoint of a square-integrable integral operator is the integral operator with the conjugate-transposed kernel | The record’s own step, written down in THEORY-009 and marked there as not the Riesz article’s, which covers matrices and bounded operators only. For a kernel k in L²(μ⊗ν), the operator (Tf)(x) = ∫ k(x,y) f(y) dν(y) from L²(ν) to L²(μ) has adjoint (T*g)(y) = ∫ k̄(x,y) g(x) dμ(x), by Cauchy–Schwarz and Fubini. So T is self-adjoint, when μ = ν, exactly when k(x,y) = k̄(y,x) almost everywhere. It does not say where the symmetry of k comes from, which is THEORY-009’s question, and it is not claimed for kernels that are not square-integrable. | Active |
| CLAIM-293 | The topology of a support graph does not fix how a conclusion should be revised when one of its supports fails: two records with the same edges can carry evidence of different strength and call for different revisions | A6’s compensating-evidence case (CASE-089), granted against reading CLAIM-295 as saying a graph of justifications is enough. Two pieces of evidence support H in both worlds. Discount the first, and posterior odds are 20:1 in one world and 1.1:1 in the other. rests_on and grounds record that support exists, not how much. | Active |
| CLAIM-294 v2 | Two inquiry histories can agree on every current conclusion and still call for different revisions of it, so a current marginal belief is not a sufficient state for updating on evidence about what it depended on | A5’s central claim, made exact by A8’s construction (CASE-093) and checked here. Two joint distributions over a hypothesis H and a condition C share P(H) = 0.9; learning C = 1 sends one to 1.0 and the other to 0.8. Keeping only the marginal merges histories that need different updates. It does not say a full posterior falls short (CLAIM-300), or that a dependency graph is the fix (CLAIM-293). | Active |
| CLAIM-300 | A state sufficient for a family of decisions under a model makes the rest of an inquiry’s history worthless for those decisions, and it can stop being sufficient when a source is discredited, the model is revised or the decisions change | A10’s answer to the owner’s U10, with A8’s fourth case and A10’s normal constructions (CASE-082, CASE-087). Under a fixed, correctly specified model a sufficient posterior makes the history redundant for those decisions. The same sample mean that sufficed cannot say which reading remains when one instrument is discredited, or test a mixture. Sufficiency was always relative to the model and the family. CLAIM-152 says the same of individuation by target; this says it of a retained state over time. | Active |
| CLAIM-302 | Equivalence under feasible tests can be much coarser than statistical equivalence: there are ensembles almost disjoint from uniform that no polynomial-size test tells apart from it, and efficiently samplable ensembles of this kind exist if pseudorandom generators do | A15’s witness for “Complexity changes what we mean by an observable”, which the owner asked to have developed at U15, corrected by the reading of Goldreich (NOTE of 2026-10-15). CLAIM-264 says agreement on a restricted probe family is coarser than identity; feasibility gives a restriction that is forced, not chosen. The separation itself is unconditional: a counting argument gives ensembles supported on at most 2^(n/2) strings that no circuit of size 2^(n/8) tells from uniform. Only an efficiently samplable witness needs an assumption, and a secure pseudorandom generator supplies one. All of it is asymptotic, about ensembles indexed by n. | Active |
| CLAIM-304 | A rendering can keep a distinction recoverable in principle and make it infeasible to recover: if pseudorandom generators exist, a uniformly random bit can be encoded so that an unbounded receiver recovers it except with probability at most 2⁻⁽ⁿ⁺¹⁾, while no efficient receiver does non-negligibly better than chance | A14’s first failure mode, computational inaccessibility, and A18’s fourth candidate: “Recoverable in principle ≠ recoverable within budget.” CLAIM-199 shows a garbling can help a bounded receiver. This is the other direction: a nearly lossless rendering can defeat every efficient one. The construction is the record’s, from CLAIM-302’s witness; A18 asked for one and warned against “a vague appeal to an inverse being hard”. | Active |
| CLAIM-305 | Resolving a singularity by blow-up reparametrizes the states and adds no observation, so it cannot make indistinguishable states distinguishable: every point over a singular point gives the same distribution, and the Fisher information can stay degenerate there; what the resolution buys is a normal form for the divergence, and so the learning coefficient | The owner’s LEGO monograph says, in Part I (chapters 4, 6, 7, 9, the Overview and “Definitions and Theorems”), that blow-up “restores identifiability”, “introduces new observables” and makes the Fisher information non-degenerate, and maps it onto splitting a LEGO bin. The pulled-back model is p(x | π(u)), a function of u only through π(u), so every point of the exceptional divisor gives the distribution of the singular point. In p(x | a, b) = N(ab, 1), the chart a = u, b = uv gives mean u²v and a Fisher matrix of zero all along u = 0. Splitting a bin is refining the observable, which can separate states. MOM-CLM-099 records that the monograph uses “blow-up” in three senses; this records why the analogy cannot hold in any of them. | Active |
| CLAIM-307 | Retrieval pressure and evidence are different criteria for carving a category, and they move in opposite directions with copy count: evidence that a distinction exists grows with its copies, the search a bin saves shrinks with them, so the bins a deadline forces are often the ones the data least supports | From MOM-CLM-166, which reads the owner’s LEGO parable (CASE-094) against a formalism that cited it with the sign reversed. CLAIM-154 says when a distinction is real enough to name. Its evidence grows with the number of copies. The parable says when a distinction must be built to act in time: a single brick among similar ones. Expected search falls as copies grow. The two answer different questions, “is it there?” and “can it be reached in time?”, and come apart in both directions. MOM’s table was re-run here and reproduced, and a simulation of the search agrees. | Active |
| CLAIM-309 | How good a mixture is and how discriminable it is from its diluent come apart: under linear mixing a smoothie’s distance from milk only shrinks as milk is added, so a preference that rises and then falls with milk is over the ratio and cannot be a d′, and the signal-detection reading of the owner’s smoothie line reverses one of its directions | The smoothie note in the owner’s LEGO monograph reads the owner’s line (CASE-096) as signal detection. Milk is the noise, “Less Good” is d′ falling as the mixture approaches milk, and “More Good” is d′ rising as a little milk is added. Under linear mixing the mixture’s taste is (1 − f)s + f·m, and its distance from milk is (1 − f)|s − m|. That falls monotonically from the first drop, so discriminability from milk cannot rise and then fall. The note’s own first model is right: goodness is a single-peaked function of the ratio, which is a preference, not a discriminability. MOM-CLM-166 found the same kind of reversal in the formal reading of the LEGO parable. MOM’s record never mentions the smoothie note. | Active |
| CLAIM-310 v2 | Sorting schemes over objects with determinate properties never conflict as descriptions, since the product of any sortings refines them all; they conflict as arrangements, since a fixed number of bins realizes one partition when each item is stored once, and which costs least depends on what is searched for and on what stands out within a bin | What is directionally right in the owner’s LEGO monograph, as Mathematics of Meaning graded it. Its contextuality claim is empty: any sortings of bricks share a joint refinement (MOM-CLM-137), and the events they generate stay Boolean (MOM-ARG-047). Its intuition that builders cannot share a sorting is true of arrangements. One set of bins realizes one partition, the product may need more bins than there are, and the cheapest choice depends on what each builder searches for and on what stands out within a bin (CASE-095). The disagreement is a trade, not a failure of global truth. Sorting on one attribute also turns a search for a combination of features into a search for one feature within each bin, which Treisman and Gelade (LIT-1082, read) show is the difference between a serial scan and one found at a glance. | Active |
| CLAIM-313 v2 | A fixed number of bins realizes one partition only when each item is stored once: with copies, overlapping partitions coexist, as hashing-based similarity search stores each item in many partitions and pays for it in space, so sortings conflict as arrangements only under a one-copy constraint, which binds bricks but not index entries | Filed at the owner’s request, from the readings on locality-sensitive hashing (Indyk and Motwani 1998; Andoni, Laarhoven, Razenshteyn and Waingarten 2017) set against CLAIM-310’s premise that “a set of k bins realizes one partition of the collection at a time”. That holds when each item is in one place. An index can hold many copies, and hashing-based similarity indexes do exactly that, trading the space of the copies against query time on an exact curve. It narrows CLAIM-310 to collections whose items cannot be copied, which includes the LEGO case; it does not touch the claim about descriptions. | Active |
| CLAIM-319 | Over a scenario of n contexts in an odd cycle, changing the support of one context takes the contextual fraction from 1 to 0 while the other n − 1 contexts and every marginal are kept exactly, at a weighted observational distortion of 1/n, so within a scenario the fraction’s Lipschitz constant against that distortion is at least n; and any change of global extension needs a contextual model on one side | Split from CLAIM-201 under ADR-043: the derivation that claim called “decisive and checkable”, stated on its own. For odd n, an n-cycle of ±1 observables in which every context anticorrelates is strongly contextual; making one context correlate gives a model with a global distribution. The two agree on n − 1 contexts and on every marginal, differ by total variation 1 in the last, and so lie at weighted D_obs = 1/n with contextual fractions 1 and 0. For n = 3 it is CASE-037 with one observable relabelled. It answers how small the local change can be; it says nothing about whether any data are contextual. | Active |
| CLAIM-322 v2 | Agreement under a probe family is evidence that two objects share structure only to the degree that the family was chosen independently of the comparison and is evaluated on cases it was not chosen on, so a family selected to make two objects look alike shows only that it can be selected | Split from CLAIM-264 under ADR-043: its evidential half, stated on its own so that its warrant can be. CLAIM-264’s exact half shows that agreement on a restricted probe family does not fix identity; this claim draws the rule for reading such agreement as evidence. It is an argument, not a theorem, so it is interpretive. It does not say which probes to choose, or that held-out probes are neutral. | Active |
| CLAIM-333 | Measuring every relevant variable is necessary but not sufficient for a Granger-causal analysis to say anything about interventions: the intervention must also be an isolated exogenous change and no influence may act within one sampling step, and then Granger noncausality implies no intervention effect and the effect is identified from the observational predictor, but a Granger-causal relation is still only potentially causal | The record’s answer to NOTE-010’s C5, “if all relevant variables are measured, Granger- and Pearl-type analyses coincide”, which the ΦID review asserts without conditions. Eichler and Didelez (THEORY-704) give sufficient conditions, and complete measurement is one of them. A two-variable process with nothing hidden shows it is not enough: a acts on b within one step, a is Granger-noncausal for b, and setting a(t) changes the mean of b(t + 1) by ργ. Checked by hand and by simulation. | Active |
| CLAIM-334 | Granger causation from a macro variable to the micro variables it supervenes on, Seth’s downward Granger causation, is not evidence of a macro-to-micro intervention effect, even with every micro variable measured: supervenience leaves no intervention on the macro variable that holds its micro base fixed at the same time | Seth measures downward causation as Granger causation from the flock’s centre of mass to each boid, and calls it “ontologically innocent” (THEORY-703). The record files the boundary so that the stronger reading has an address. If M(t) = f(m(t)), setting M(t) means changing m(t), so the intervention Eichler and Didelez’s Assumption 2.3(i) requires does not exist (THEORY-704). Measuring every boid does not help, by CLAIM-333. | Active |
| CLAIM-335 v3 | Transfer entropy measures directed predictive improvement, not information flow: as a conditional mutual information it is the source’s unique information about the target’s next value plus its synergy with the target’s past, so it can credit a source with a bit the source carries only jointly and read zero from sources that jointly fix the target, and conditioning on more series does not remove the synergy | The record’s position on whether transfer entropy measures information flow. James, Barnett and Crutchfield show it does not, if flow is information attributable to the source alone (THEORY-707). Williams and Beer’s decomposition shows why, for any partial information decomposition satisfying the consistency equations (THEORY-712). Both papers’ examples were recomputed exactly. This is a different failure from CLAIM-333’s, prediction against intervention. It leaves transfer entropy standing as a predictive quantity. | Active |
| CLAIM-340 | Scoring operators on predictions committed before the evidence is not by itself Regime B: in the linearized operator model the coupling of interpretation to endorsement has the sign of κ′ − κ̄, so commitment removes capture only when the interpretation bends the reading of the evidence no more than it bends the predictions, and it keeps the learning that revising after the evidence would lose | The formal question QUESTION-121 asked, answered in CASE-099’s linearization (CASE-106): is committing predictions before the evidence equivalent to Regime B? No. The coupling γ is 4ηs(κ′ − κ̄), where κ̄ is how far the shared meaning bends the committed predictions and κ′ how far it bends the reading of the evidence. It is corrective, null or capturing as κ′ is below, at or above κ̄. Strong correction can fail by oscillation when |βγ| ≥ 1 − αδ. Revising predictions after the evidence sets γ = 0 but also stops authority from tracking who predicts well. Linear, at the symmetric state, for squared-error scoring. | Active |
| CLAIM-343 | For three or more parts, no ΦID can use a redundancy function that both satisfies the identity axiom and yields nonnegative atoms, because on the Williams–Beer lattice the two are inconsistent from three sources up; a ΦID verdict on three or more parts therefore rests on which axiom the analyst drops, and should name it | The record’s answer to the gap left by the PID filing: no held measure gave ΦID an identity-respecting redundancy for more than two sources. Rauh et al. show none can, on the lattice ΦID uses, without a negative atom (THEORY-713). The record’s measures split the difference: I_min keeps atoms nonnegative by calling independent copied bits redundant, and Ince’s I_ccs keeps an independent-sources identity and reports a −1-bit atom (THEORY-714). Two-part ΦID is untouched. | Active |
| CLAIM-345 | Equal controllability rank does not fix the cost of control: for x_{t+1} = x_t + b·u the least quadratic effort for a displacement Δ is Δ²/b², unbounded as b → 0 while the system stays controllable, and in general the least effort is Δxᵀ W_T^† Δx, set by the Gramian and not by its rank | A9 §2 and A12’s Proposition 1 of the exchange of 2026-10-17, the step the owner’s U9 “controllability of a system isn’t binary” needs. For a scalar step x_{t+1} = x_t + b·u the cheapest input costs Δ²/b², so two systems that are both controllable can differ in cost by any factor. The general statement is the minimum-energy formula with the controllability Gramian. A9: “This is already a standard result from linear control theory.” Proved here because the record holds no source for it. The sharp edge: the cost is relative to a fixed input norm, and rescaling the input makes every b cost the same, so comparing costs across systems needs inputs in common units. | Active |
| CLAIM-346 | Feasibility, aggregation and epistemic coordination are independent obstructions to an organization’s realizing its demands: each can occur with the other two absent | Owner-seeded, U36 of the exchange of 2026-10-17: Arrow’s theorem, then cryptography and “the two generals problem”. A36 placed them: “Arrow’s theorem belongs in the theory of endorsement, not directly in the geometry of feasible updates”, and the two generals add “epistemic frustration”. Three elementary constructions, each with one obstruction and not the others, are given in the body. The sharp edge: the independence is relative to fixing the demands as the constituents’ own acceptance conditions; that no remedy for one removes another is A36’s further reading, argued in the body and not part of the claim. | Active |
| CLAIM-350 v2 | A coarse-graining preserves a deterministic judgment exactly when the judgment is constant on its fibres: the judgment then factors through the coarse description, uniquely on its image, and otherwise two states the judgment tells apart have been merged | A21 §3 and §18 (Candidate 1) of the exchange of 2026-10-17, and A22’s Construction A: for a judgment J and a coarse-graining C on the same states, J = j∘C for some j exactly when C(x₁) = C(x₂) implies J(x₁) = J(x₂). Granted, because it is elementary (the universal property of a quotient) and Ak’s; the proof is in the body. It is CLAIM-152’s judgment-valued special case with the converse added: CLAIM-152 says a partition sufficient for one target need not be for another, and this says which partitions serve a deterministic judgment, those whose fibres lie inside its level sets, so the coarsest is the judgment’s own partition. The sharp edge: it says when a description keeps a judgment, never whether the judgment is right. A21’s probabilistic and causal versions are recorded in the body, not claimed. | Active |
| CLAIM-351 | A zero aggregate gradient does not show that the demands it aggregates are reconciled: with positive weights it holds both when every demand’s own gradient vanishes and when nonzero gradients cancel, and in the second case no update improves them all to first order; even the first is only joint first-order stationarity, which a demand’s maximum or saddle also has | A33 §5 of the exchange of 2026-10-17: “Zero aggregate gradient does not imply absence of frustrated demands.” With L₁ = (θ − 1)² and L₂ = (θ + 1)² at θ = 0 (CASE-115), the equal-weight gradient is zero and the aggregate is at its global minimum, while each loss is 1 and wants the opposite move. In general a vanishing positive combination of gradients is either every gradient zero or a certificate that no update improves them all, and the scalar cannot tell which. The sharp edge: it is a fact about a weighted sum, not about what a monitor reading it represents. A29’s “the unity of the distress signal need not arise from the unity of the underlying causes” is the interpretive reading, kept apart in the body. | Active |
| CLAIM-365 v2 | On a norm-ball update set the local frustration residual Φ_r = min over ‖d‖ ≤ r of max_i(ε_i + g_iᵀd) equals the maximum over the simplex of Σλ_iε_i − r‖Σλ_ig_i‖_*, so infeasibility of the declared improvements has a certificate, and any admissible actual update exceeds Φ_r by a nonnegative governance regret | A34 §4 and A35 §3 of the exchange of 2026-10-17, checked. The residual is the smallest worst-case shortfall of the declared local improvements ε_i over updates of size at most r. Minimax for the bilinear problem on two compact convex sets gives the dual over the simplex, so Φ_r > 0 exactly when some λ certifies that no update meets all demands (A34’s form). Governance regret Γ is the gap between an actual admissible update’s shortfall and Φ_r, and is never negative. For L₁ = (θ − 1)², L₂ = (θ + 1)² at θ = 0, Φ_r = ε for every r. The sharp edge: per A37 it is a cardinal refinement of the ordinal fact (CLAIM-347), and its magnitude depends on units (A36 §2D). | Active |
| CLAIM-366 v3 | Whether several objectives share a strict first-order descent direction, 0 ∉ conv{g_i}, is invariant under smooth invertible reparameterization, while the sign of a Euclidean gradient inner product is not, and a negative inner product does not exclude common descent; a budgeted version is invariant only with the metric pulled back | A35 §2 of the exchange of 2026-10-17, at the owner’s U35 “double click” on whether functionally equivalent parameterizations give equivalent frustration diagnoses. Gradients transform as g′ = Jᵀg and directions as v′ = J⁻¹v, so directional derivatives, and with them the common-descent test 0 ∉ conv{g_i}, do not depend on coordinates. A Euclidean inner product does: g₁ = (1, 1), g₂ = (1, −½) have product ½, and under θ = diag(½, 2)ψ their transforms have −7/4, while d = (−1, 0) lowers both throughout. The sharp edge: it covers coordinate changes of one parameter space at first order, not function-preserving symmetries or training trajectories (QUESTION-127). The practice corollary, not to diagnose task conflict by gradient cosine, is the anthology’s. | Active |
| CLAIM-369 | In symmetric resource-mediated competition with interaction matrix A = UᵀU and growth rates r = Uᵀb − c, Lotka–Volterra population dynamics are multiplicative gradient flow on the convex feature-selection objective ½‖b − Un‖² + cᵀn over the nonnegative orthant: the invasion fitness of an absent component is the negative partial derivative of that objective, and the uninvadable equilibria are its KKT points | From a synthesis the owner relayed on 2026-10-18 (workbench entry of that date), checked here and given a proof. When competition between components is mediated by shared resources with feature or niche vectors u_i, so that A = UᵀU and r = Uᵀb − c, the Lotka–Volterra dynamics are ṅ_i = −n_i ∂_i L with L(n) = ½‖b − Un‖² + cᵀn: entropic mirror descent on the positive orthant. A rare component’s invasion fitness is −∂_i L, the value of adding that feature, and uninvadable equilibria are the KKT points of the convex program. The sharp edge: the identity needs a shared scalar objective, and it fails for asymmetric interactions or processes with competing objectives, which is where an organization stops being an optimizer. Mehta et al. (2019) and Marsland, Cui and Mehta (2020) give the general duality for consumer–resource models. | Active |
| CLAIM-370 | For a symmetric positive-pair distribution with marginal p, the conditional-expectation operator on L²(p) has the density ratio K⁺ as its kernel, and it is Hilbert–Schmidt exactly when the χ²-divergence of the pair from the product of its marginals is finite, in which case it is compact and self-adjoint with an orthonormal eigenbasis | THEORY-005’s derivation, stated as the record’s own and with its proof. Johnson et al. (LIT-258) write the operator on a finite space without an integrability condition. HaoChen et al. (LIT-249) reach the Hilbert– Schmidt property from a uniform bound, which is stronger. Here the condition is exact: Hilbert–Schmidt if and only if finite χ². It says nothing about what trained features span. | Active |
| CLAIM-371 | Under the componentwise order on distortion profiles, two profiles are incomparable exactly when they trade off, and then positive weightings can rank them either way, while every positive weighting agrees with a dominance | The order-theoretic part of CLAIM-128, on its own so that it can carry its own warrant. It says nothing about translation, what the components measure, or which weighting a task fixes. Only that, under the componentwise order of TERM-043, a trade-off is incomparability, that a trade-off can be resolved either way by positive weights, and that a dominance cannot. | Active |
| CLAIM-372 | Within a fixed measurement scenario, the contextual fraction is a piecewise-linear and Lipschitz-continuous function of the empirical model’s probability table, with a constant that depends only on the scenario | The record’s own derivation, from the linear programme of Abramsky, Barbosa and Mansfield (LIT-265, Eq. 3), which THEORY-165 holds. The non-contextual fraction is the minimum of the dual objective over finitely many vertices, so it is concave and piecewise linear in the probability table, and the contextual fraction is Lipschitz in it. The constant is the largest ℓ₁ norm of a dual vertex, which depends only on the scenario and is not bounded here. Nothing here crosses scenarios or applies to signalling data. | Active |
| CLAIM-373 | If a map approximately intertwines two operators, with defects E_A = ΦA − A′Φ and E_B = ΦB − B′Φ, then Φ[A,B] − [A′,B′]Φ = A′E_B + E_A B − B′E_A − E_B A, so the commutator is carried to within (‖A‖ + ‖A′‖)‖E_B‖ + (‖B‖ + ‖B′‖)‖E_A‖ | The commutator-distortion identity of the manuscript’s §9, on its own and proved. A map that nearly intertwines each of two operations carries their commutator to the commutator of the images, up to an error that is linear in the two intertwining defects. ARG-006 is the exact case. It says nothing about whether the defects can be identified from data (CLAIM-253), or what bounds an observed order effect (CLAIM-231). | Active |
| CLAIM-374 | On a consistently connected system, Contextuality-by-Default and the sheaf criterion agree: a maximally connected coupling exists exactly when a global law has the contexts’ laws as its marginals | The record’s derivation of what NOTE-600 marks as its own inference. In a consistently connected system, every observable has one distribution in all the contexts that contain it. Then each connection’s maximal coupling makes its copies equal with probability 1 (LIT-777, Theorem 3.3), so a maximally connected coupling is a joint law of one variable per observable, which is a global section in Abramsky and Brandenburger’s sense (LIT-016, THEORY-012). Conversely, a global section gives such a coupling. The two criteria come apart only when marginals shift with context. | Active |
| CLAIM-375 | Under one-hop allocation within an epoch, an operator that endorses an ally whose policy is nearer its own than the collective decision is raises how far its preferred policy is realized while its own authority stays fixed, so the two conditions of the endorsement scheme hold together | The demonstrated core of CLAIM-285. Under A3’s one-hop rule I_j = I_j⁰ + Σ_(k≠j) b_k E_kj, with budgets and baselines fixed within an epoch (CLAIM-282), an operator’s authority does not respond to its own endorsements. If the collective decision is the authority-weighted mean of the operators’ policies, then endorsing an ally whose policy is nearer one’s own than the decision is moves the decision strictly toward one’s policy. So the reward for realized policy, and not for held authority, pays for delegating. It does not say operators behave so, or that any rule beyond one hop keeps both conditions. | Active |
| CLAIM-376 | A linear system that is stable at every instant can have a positive Lyapunov exponent: shifting any operator with exponent λ > 0 by −δI lowers its exponent by exactly δ and makes every instantaneous operator stable, so for 0 < δ < λ the shifted system is stable at every instant and unstable | The case THEORY-724’s source asserts and does not show. If A(t) has first Lyapunov exponent λ > 0, the propagator of A(t) − δI is e^(−δt)Φ(t), so its exponent is exactly λ − δ, while every eigenvalue of every A(t) moves left by δ. Applied to the undamped oscillator switching ω between 0.5 and 3 every unit of time: λ = 0.1955, and with δ = 0.1 the system is stable at every instant, with eigenvalues −0.1 ± 3i and −0.1 ± 0.5i, and has exponent 0.0955. | Active |
| CLAIM-377 | For any differentiable model under squared loss with Gaussian noise of precision β, the model’s Fisher on n inputs is the Gauss–Newton matrix βJᵀJ, and its non-zero eigenvalues are exactly β times those of the empirical NTK Gram matrix JJᵀ, with the same multiplicities, its eigenvectors being Jᵀ applied to the Gram matrix’s, normalized | THEORY-084’s Step 1, stated as the record’s own and proved. Through the SVD J = UΣVᵀ, JᵀJ and JJᵀ share the non-zero eigenvalues σ_i² with equal multiplicities, and Jᵀu/σ is the eigenvector of JᵀJ for a unit eigenvector u of JJᵀ. Under y | x ~ N(f(x; θ), β⁻¹), the expected outer product of the score is β∇f∇fᵀ, which summed over inputs is βJᵀJ, the Gauss–Newton matrix. “Fisher” means the model’s Fisher, not the empirical Fisher of observed residuals. It holds at every width, in every regime and on any data. It says nothing about what the spectrum looks like, which is THEORY-084’s. | Active |
| CLAIM-378 | A representational similarity index invariant to every invertible linear map depends on a representation only through its column space, so it takes one value on all column-centred representations of rank n − 1, as it does on all representations of rank n | The generalization of THEORY-236’s theorem. If col(X) = col(Y), write X = BC and Y = BD with B a basis of the common column space and C, D of full row rank, extend C and D to invertible matrices, and the quotient maps X onto Y. Since XA always has the column space of X, the GL(p) orbits are exactly the sets of n × p matrices sharing a column space. Theorem 1 is the case of column space ℝⁿ. Column-centred matrices of rank n − 1 all have column space 1^⊥, so an invariant index cannot tell them apart either. | Active |
| CLAIM-379 | On a finite set of stimuli, regularized linear readouts with ridge b > 0 cannot tell apart two centred representations of the same width exactly when one is an orthogonal transformation of the other, and the unregularized CCA member cannot tell apart representations with the same column space, which differ by an invertible linear map | The proved core of CLAIM-269. What a regularized linear readout decodes is a function of K_X = X G⁻¹ Xᵀ/M, with G = aC_X + bI (THEORY-008). For b > 0, K_X and the Gram matrix XXᵀ determine each other, by a monotone map on the shared eigenvalues; and equal Gram matrices mean an orthogonal transformation (CLAIM-290, THEORY-004). For b = 0, K_X is the projection onto the column space, which fixes X only up to an invertible linear map, as for a softmax layer (THEORY-723). | Active |
| CLAIM-380 v2 | With finitely many states and finitely many outcomes, the positive part of the Bayes-risk increase on the same decision problem is directed, is zero from a garbling back to its source, vanishes for every finite problem and prior exactly when the rendering is at least as informative as the source, and has A86’s D_Q as its symmetrization | The finite-outcome part of CLAIM-326. Define D⁺_Q(U → V; π) = sup over q in Q of [r_q,π(V) − r_q,π(U)]⁺, with losses in [0, 1] and a stated prior. A garbling never gains. The term is directed. A86’s D_Q is its symmetrization. And with every finite problem and prior, it vanishes exactly when V is at least as informative as U, by Blackwell’s Theorem 6 (THEORY-717). It does not cover outcome spaces that are not finite, or say this is the right decision term. | Active |
| CLAIM-381 | Over a finite class of items with finite judgement outcomes, a constant outcome kernel fits every item only when the target readers’ averaged responses do not vary with the item, an exact fit is a garbling with the item as common parameter, and Le Cam’s deficiency and its restriction to finitely many decision problems are finite computations | The derivational core of CLAIM-228. Over a finite class of items, the target term is the readers’ response averaged over a procedure’s output, and one outcome kernel family is fixed across the items. Then a constant kernel fits only target responses that carry no information about the item. An exact fit is a garbling, with the item as the common parameter. And Le Cam’s deficiency between the two finite tables is one linear programme. Its restriction to finitely many decision problems is a finite computation. It does not say the measure is informative, which is CLAIM-228’s empirical question. | Active |
| CLAIM-383 | In plane Couette and plane Poiseuille flow, measured in the energy norm, the linearized operator about the laminar profile has transient growth at least cR and resolvent norm at least cR² with c independent of R, so wherever it is eigenvalue-stable a perturbation of norm at most CR⁻² makes it eigenvalue-unstable | THEORY-304’s scalings as lower bounds, derived. At streamwise wavenumber zero, the cross-stream velocities obey an R-independent Stokes problem in the time t/R. They force the streamwise velocity through −U′v, so the streamwise response is R times an R-independent field. In the energy norm this gives transient growth at least cR, and in the steady forced problem a resolvent norm at least cR². Through pseudospectra, a perturbation of norm O(R⁻²) then moves an eigenvalue into the unstable half-plane wherever the flow is eigenvalue-stable. The sharp constants, and that the growth is no more than order R, come from the computation and are not here. | Active |
| CLAIM-384 | For a network linearised at its initial parameters, with an isotropic Gaussian prior of precision α on the parameter displacement and Gaussian noise of precision β, the log-determinant part of the evidence is −½ Σ log(1 + βλ/α) over the non-zero eigenvalues λ of the empirical neural tangent kernel’s Gram matrix on the training inputs, so at most n directions are priced however many parameters the network has | Active | |
| CLAIM-385 v2 | If a symmetric operator on ℝⁿ commutes with an orthogonal representation of a compact group, each eigenspace is a sum of real irreducibles and each multiplicity a sum of multiples of their dimensions, so SO(2) and ℤ/n for n ≥ 3 force paired eigenvalues; conversely the spectrum fixes neither the group nor a basis inside an eigenspace | Active | |
| CLAIM-386 | For x(t+1) = (A* + εC)x with A* block-diagonal, the Simon–Ando aggregate  = QAV has one-step closure defect exactly QA − ÂQ = QA(I − VQ); the part from the blocks’ own dynamics vanishes iff every block has equal column sums, and then the whole defect is ε·QC(I − VQ) | Derived in the reading of Shpak, Stadler, Wagner and Altenberg (2004), NOTE-977, which states the Simon–Ando aggregate without an error term, and re-checked here. It splits the closure defect of block aggregates into a part from coupling between blocks, which is exactly first order in ε, and a part from the blocks’ own dynamics, which vanishes exactly in the equitable case of THEORY-755. Deterministic dynamics, one step; the post-transient orders reported below are numerical, not part of the claim. | Active |
| CLAIM-387 | Optimal deterministic information-bottleneck partitions at two budgets need not be nested: with X uniform on four states and binary Y with P(Y=1|x) = (0, 0, 1/8, 3/8), the optimum is {0,1,2 | 3} for 6.91 < β < 9.26 and {0,1 | 2,3} for 9.26 < β < 16.06, two partitions neither of which refines the other | Found by brute force in the reading of Equitz and Cover (NOTE-988) and re-run here from scratch. It answers the deterministic half of QUESTION-220 in the negative: raising the budget can reorganize the optimal partition rather than split it. The two optima are not even laminar, so no fixed tree holds both as prunings, and NOTE-676’s nesting is bought by its tree restriction. It says nothing about stochastic encoders, where finite-β optima generally live (THEORY-743). | Active |
| CLAIM-388 v3 | Exact dynamical closure of a collective variable does not require a nearly decomposable block: in a linear Gaussian VAR(1) with uniform coupling among n units, the sum of all units is exactly dynamically independent at every stable coupling strength, while every partition of the units into two or more groups couples as strongly across its edges as within them | From A5 of the exchange of 2026-10-19 on coarse-to-fine processes, which gave the two-unit continuous-time case (ż = (b − a)z for z = x₁ + x₂) and the boxed slogan “Exact aggregation ⇏ near-decomposability”. Made exact here for n units with THEORY-733’s sufficiency condition, and checked numerically. It answers the converse half of QUESTION-221. It does not say the closed sum is one system (QUESTION-218), and it is about partitions of the units: in the eigenbasis of A the same system is exactly decomposable, which is the point. | Active |
| CLAIM-389 v2 | In a stable linear Gaussian VAR(1), exact dynamical independence of a linear coarse-graining implies that its Mori–Zwanzig memory vanishes, but not conversely: memory needs coupling out of the coarse variable and back, and dependence needs only the way back, so a coarse variable can be Markov in its own law with zero memory and still be dynamically dependent | Derived in the reading of Chorin, Hald and Kupferman (2000), NOTE-989, and re-checked here. It settles, for the linear Gaussian VAR(1) class, the relation QUESTION-222 left open between memory in the reduced dynamics and THEORY-257’s dynamical independence: memory is the round trip Y → rest → Y, dependence the inward leg alone. Not about VAR(p ≥ 2), where the reading reports that an independent variable can carry memory, nor about nonlinear systems. | Active |
| CLAIM-390 | For a block of coordinates in a stable linear Gaussian VAR(1), dynamical dependence is at most ½ log det(I + Σ_YY⁻¹ A_YR C A_YRᵀ), with A_YR the coupling into the block and C the covariance of the rest given the block’s present state, so it is of second order in the coupling into the block, whatever the coupling out of it | First found in the reading of Chorin, Hald and Kupferman (2000), NOTE-989, as a by-product; derived and checked here. It answers the forward half of QUESTION-221 for a block’s full state: weak coupling into a block makes its dynamical dependence small, to second order. It is about the block’s full state, not its aggregate. For the aggregate, the block’s own internal dynamics add a defect of their own (CLAIM-386). | Active |
| CLAIM-391 | For any classes with weights π_c and per-class factor matrices H_c and Δ_c, the class-wise Kronecker product Σ_c π_c H_c ⊗ Δ_c exceeds the product of the weighted averages H̄ ⊗ Δ̄ by exactly the weighted between-class covariance Σ_c π_c (H_c − H̄) ⊗ (Δ_c − Δ̄), so the two agree when either factor is the same in every class | Active | |
| CLAIM-392 v2 | Stochastic information-bottleneck optima at two budgets need not be nested: for X uniform on four states and binary Y with P(Y=1|x) = (0, 0, 1/8, 3/8), the unique optima at β = 8 and β = 12 admit no Markov chain X → T₂ → T₁, so raising the budget reorganizes the optimum even with noisy encoders | Computed in the reading of Charvin, Catenacci Volpi and Polani (2023), NOTE-981, on the source of CLAIM-387, and re-run here. With THEORY-753’s criterion (refinable iff some pair of optima forms the chain, iff the coarse decoder points lie in the fine decoder hull when that hull is non-degenerate), and optima certified unique, the hull test fails. It answers the stochastic half of QUESTION-220 in the negative for this source. The certificate is numerical, not interval arithmetic, and the loss is small: 0.49% of the larger budget, in unique information. | Active |
| CLAIM-394 | In a stationary linear Gaussian VAR(1) that is time-reversible (AΠ = ΠAᵀ, Π the stationary covariance), every slow-feature direction is a left eigenvector of A, so any span of slow features is an exactly dynamically independent coarse-graining; off reversibility a slow feature is in general neither independent nor free of memory | Derived in the reading of Schüler and Wiskott (2025), NOTE-1015, and re-checked here. It says when slowness, the criterion of slow feature analysis (THEORY-746), picks out a closed coarse-graining in THEORY-733’s sense: always in the reversible linear Gaussian VAR(1), which is the class of noise-free square mixtures of independent AR(1) processes, and not in general otherwise. Linear features, linear Gaussian dynamics, lag one. | Active |
| CLAIM-395 | Under Montévil and Mossio’s definition of closure, a closed set of constraints has at least two members, and if each constraint is produced at the time scale of a constraint in the set that generates it and not at the scale at which it is itself conserved, the set includes at least one slow and one fast dependence, so a system with a single constraint of its own cannot realise closure | Active | |
| CLAIM-397 v2 | For a scalar stationary process with autocovariance C, the Mori–Zwanzig memory kernels for the projection onto the current value are fixed by C alone, through C(m+1) = Σ_{k≤m} c_k C(m−k); the memory vanishes iff C is geometric, and when C has finite Hankel rank d the memory has Hankel rank exactly d − 1 | Derived in the reading of Lin and Lu (2019), NOTE-1029, and re-checked here. It settles the minimal-order question CLAIM-398 left open, for every scalar stationary process with rational autocovariance, and gives the structural reason for CLAIM-389’s “not conversely”: the memory is a functional of the coarse variable’s own law, so it cannot register dependence on the rest of the system. | Active |
| CLAIM-398 | In a stable linear Gaussian VAR(1) on ℝⁿ, a scalar linear coarse variable Y = LX is, in its own law, an ARMA process of orders at most (n, n − 1); so its Mori–Zwanzig memory, though non-zero at every lag in general, is carried exactly by finitely many coordinates that are functions of Y’s own past and innovations | Derived in the reading of Chorin and Lu (2015), NOTE-1028, and re-checked here. Their NARMAX closures are fitted, with no error bound; in the linear Gaussian class the same idea is exact. A coarse variable with memory admits a finite Markovian augmentation built only from its own history, which is the augmentation THEORY-257’s criterion conditions on. Scalar Y, VAR(1), Gaussian noise. | Active |
| CLAIM-399 | Causal abstraction in the senses of Rubenstein et al., Beckers and Halpern, and Beckers, Eberhardt and Halpern does not secure governance-response sufficiency: exact transformation, τ-abstraction and zero approximate-abstraction distance all accept the hidden-bit coarse-graining, which loses ρ(h = 1) bits of the response to h, because each lets high-level noise carry what the coarse state drops; for one outcome and the identity correspondence, sufficiency is their compatibility condition with the high-level context fixed to the coarse state, and strong abstraction rejects the case only through interventions on the hidden parameter | The record’s join, on 2026-10-21, of its readings of the causal-abstraction papers (NOTE-1037, NOTE-1032, NOTE-1045) with CLAIM-407. A25 of the exchange of 2026-10-20: “We should not present governance-preserving coarse-graining as a new general theory of causal abstraction.” That stands. But it is not an instance of these papers’ notions either: their criteria compare interventional laws and let the high-level context be any map that works, so they accept the hidden governance bit, whose coarse state keeps nothing of the response to h. Re-checked step by step against the three THEORYs. It corrects TERM-067’s first wording, which read the profile as TERM-043’s causal component with the identity correspondence. | Active |
| CLAIM-400 v3 | In the meaning–authority model linearized at symmetry, a third, maintained legibility state that authority moves (ε) and that moves interpretation (ν) or endorsement (μ), with persistence λ < 1 and nonnegative couplings, destabilizes the symmetric state exactly when the two-variable condition holds with β + νε/(1 − λ) in place of β and δ + με/(1 − λ) in place of δ: at the threshold the regime is indistinguishable from stronger two-variable couplings, and the slower it is the more it adds, while away from the threshold it adds a mode that no two-variable model has, so a lean in the regime can outlast the advantages it was built from | QUESTION-253 asked whether authority’s effect on what becomes legible as evidence is a third channel with its own instability condition, or CLAIM-340’s κ′ renamed. Both, in different respects. With a maintained legibility state r, det(I − J₃) = (1 − λ)[(1 − α)(1 − δ) − βγ] − ε[νγ + μ(1 − α)], which is (1 − λ) times CLAIM-330’s expression with β_eff = β + νε/(1 − λ) and δ_eff = δ + με/(1 − λ). So the instability threshold is a two-variable one with stronger couplings, and a slower regime (λ near 1) adds more. Away from the threshold, r is a third mode: a lean in the regime can hold interpretation and authority off symmetry after they would have relaxed, which no two-variable parameters reproduce. | Active |
| CLAIM-401 v3 | The conditions an effective description is commonly asked to meet are not of one kind: predictive sufficiency and approximation fidelity are adequacy relative to a declared target, metric and horizon; tractability restricts the operations under which adequacy is assessed; dynamical independence and memory closure are relations of the representation to the system, memory relative to a chosen projection; so a matrix over them as independent binary axes conflates kinds | A4 and A5 of the exchange of 2026-10-20, on the owner’s U4 (“it sounds like considering it from this angle has potentially already altered our characterization of those ‘virtues’”). A4: “(P) and (A) look increasingly like evaluation criteria, while (D) and (M) describe different aspects of system–representation structure. (C) describes resource accessibility.” A5: approximation fidelity “may specify the degree to which a prediction or dynamical relation is preserved”, and tractability “can instead restrict the operations over which preservation is evaluated”. The record adds what A5 missed: memory is relative to the projection too (QUESTION-222, THEORY-752, CLAIM-397). It is CLAIM-348’s “adequacy is fixed by the evaluative target” applied to the axes of a comparison. It does not say the five reduce to three. | Active |
| CLAIM-407 v3 | A coarse-graining exactly informationally closed under a process’s ordinary dynamics need not preserve its responses to an intervention on the mechanism: for a declared intervention distribution ρ, independent of the state, the deficiency Γ = I(S; Y^(I) | T(S), I) can be positive, as in the hidden governance bit, where it is ρ(h = 1) bits; and Γ = 0 exactly when, for every intervention ρ charges, the response kernels are constant on T’s fibres over the states of positive probability | A16 §3–4 of the exchange of 2026-10-20 and A17 §4.2–4.3, its paper’s Proposition 1, reused at A18 and in A23’s audit. A16: “a representation can be perfectly informationally closed under the ordinary dynamics while failing to preserve a governance-relevant counterfactual distinction.” The witness is CASE-125; the proof is in the body, checked exactly on 2026-10-21. The qualitative point was on file: CLAIM-350’s “What it does not say” (A22’s Construction B of 2026-10-17, CLAIM-127), and CLAIM-122’s supervisory test, whose note of 2026-10-18 has it without parameters. New here are the parameters, the deficiency Γ and the exact condition for Γ = 0. It is an application of ordinary causal sufficiency, not new information theory, and it does not say the system governs itself. | Active |
| CLAIM-408 v3 | In ḋ = (α − κ)d − βd³, with d the difference between two parties’ shares of authority, α their self-reinforcement and κ a constraint that bears on both alike, the equal allocation is the only equilibrium and is stable for κ > α, and for κ < α it is unstable and the unequal allocations ±√((α − κ)/β) are stable: a constraint that favours neither party can, by weakening, destabilize the equal allocation and make persistent unequal allocations stable, of any size near the threshold (|d*| → 0 as κ → α), without determining which party gains the larger share, and in this model restoring κ > α restores the equal allocation | A19 of the exchange of 2026-10-20, on the owner’s U19: does the account “capture ‘deregulation’ as a disguised effort to induce organizational symmetry breaking?” A19’s model is a supercritical pitchfork, CASE-112’s normal form with no tilt and μ = α − κ. The stability analysis is checked here, algebraically and numerically. A19: “A policy can be neutral with respect to the identities of participants while being non-neutral with respect to the kinds of power distributions its dynamics produce.” The claim is about the model. It says nothing about any reform, or about anyone’s intent, and it does not show that a tipped allocation outlasts the constraint’s return: in this model it does not. | Active |
| CLAIM-409 v2 | For one state law, one intervention distribution and one outcome, a deficiency of sufficiency cannot increase under refinement: if T₁ = g∘T₂ then Γ(T₂) ≤ Γ(T₁); informational closure, by contrast, is monotone in neither direction, since in the three-bit chain of Rosas et al. (2024) A is closed, its refinement (A, B) is not and the full state refining that is closed again, while A⊕B, a coarsening of (A, B), is not closed either; closure fails to be monotone because refining a description changes the future it must predict as well as the evidence, so Blackwell dominance, which every deterministic refinement has over its coarsening, does not order closure | Two readers’ drafts of one point, from A6, A7, A16 and A17 of the exchange of 2026-10-20. A16 §5 and A17’s Proposition 2: a fixed-target deficiency is refinement-monotone; “Adding information cannot hurt exact sufficiency for an unchanged question, but changing what the description is expected to explain can expose new dependencies.” A6 §6 and A7’s Proposition 3: “Increasing the information retained by a representation can destroy a desirable dynamical property”, and “There is no general data-processing inequality ordering these quantities, because both the predicted variable and the conditioning information change.” This claim joins them, and corrects A7’s slogan “When More Information Reveals Less Autonomy”: closure is not reversed by refinement, it is unordered by it. Checked exactly on 2026-10-21. | Active |
| CLAIM-414 v4 | Matched ordinary trajectories can conceal differences in an interpretation’s efficacy, in the allocation of that efficacy, and in the maintenance of its enabling constraints: in two finite constructions, systems that agree under ordinary operation differ in each, and a specified intervention exposes the difference that passive observation of the matched trajectories does not | A10 of the exchange on experience and interpretation, re-run here. Three one-bit systems behave alike when the interpretation is 0. Under an intervention that sets it to 1, the first never acts on it, the second acts at once, and the third acts one step late, because its own interpretation opens the gate. Two two-constraint systems have the same trajectory, (1,1) throughout. A single perturbation runs one into (0,0) and leaves the other restored to (1,1): one maintains its constraints by CLAIM-395’s minimal slow–fast closure, the other is reset from outside. These are CLAIM-407’s lesson about governance and CLAIM-194’s about boundaries, applied to interpretation. | Active |
| CLAIM-434 | On a finite set of outcomes, some function rises with every move that wins one of a negotiation’s conflicts exactly when no chain of winning moves returns to where it began, and the length of the longest such chain ending at an outcome is one such function; a perfectly transitive record of which side won each pair, in a complete tournament contesting every pair once, leaves a cyclic share of (n − 2)/3n under the Hodge decomposition, a baseline that does not carry to sparse records; the Hodge gradient can reverse a comparison the wins settle, netting a record erases alternation, a record fixes a class of potentials rather than one, a coherent record is not yet one aligned with any single perspective’s ordering, a best-fitting ordering of a cycle is no Lyapunov function, the sign a class of potentials gives can be set-valued, and an ordering fitted to a record overstates its agreement there | The formal core of the rename of 2026-10-23 from “ascendancy” to “potential” (TERM-076). Part 1 is the finite case of Monderer and Shapley’s equivalence between the finite improvement property and a generalized ordinal potential, proved here because the record has not read their paper (LIT-1462). Part 2 is a calibration of the Hodge decomposition (Candogan et al.; Jiang et al.) for binary conflict records, computed here: even a perfect ranking leaves a cyclic residual. Part 3 separates coherence from alignment. Parts 9 to 11 were added after a second review. Parts 4 to 8 were added when a review of the pull request found that the measures built on parts 2 and 3 confused ordinal and cardinal structure; each is checked in scratch/formal/valence.py. | Active |
| CLAIM-436 v4 | An instantaneous ordering is a common currency on a finite set of options, where every complete preorder has a real-valued representation, but not in general: the lexicographic order on the unit square has no real-valued representation, while a continuous complete preorder on a connected separable space has a continuous one; so admitting an ordering is strictly weaker than admitting a currency, a graded lexicographic override lies in the gap while an exact threshold override does not, a threshold override with a discrimination threshold is a lexicographic semiorder whose choices can cycle and so is no ordering at all, and even where each instant has a currency, successive currencies need not share exchange rates | The formal half of the owner’s reframing of 2026-10-23, that a “common currency” is better read as a negotiation admitting an (at least instantaneous) ordering. On finite option sets the two coincide at each instant. On a continuum they do not: a graded lexicographic order, one form ALARM’s “100% weight” override could take, admits no numerical representation, while a continuous ordering, the form of Veit’s proportional trade-offs, admits a continuous one (Eilenberg; Debreu). A threshold override admits a discontinuous one, and one with a discrimination threshold, a lexicographic semiorder, can cycle and admits no ordering at all (Luce; Tversky). Across instants, representations need not be related, so a sequence of currencies need not be one currency. | Active |
| CLAIM-452 | For an organization whose response in each probe context is the authority-weighted mixture of its processes’ known response profiles, the authority shares are identifiable from a set of contexts exactly when the processes’ stacked profiles over those contexts are affinely independent; because the response is linear in the shares, local and global identifiability coincide, the condition is that the Fisher information of any design supported on those contexts is positive definite on the simplex’s tangent space, and with binary outcomes at least one context fewer than there are processes is needed | When an organization’s response in context c is p_π(·|c) = Σᵢ πᵢ pᵢ(·|c), the mixture CLAIM-444 reduces the replicator model to, which probe contexts reveal its authority shares π? Exactly those over which the profiles are affinely independent. The response is linear in π, so there is no gap between local and global identifiability, and the condition is the Fisher information of CASE-159’s pullback being positive definite on the tangent space of the simplex. Each binary context adds at most one rank, so k processes need at least k − 1 contexts. It is a condition checkable before any data, which is what QUESTION-031 asks for; none of the three works read on 2026-10-25 states it. | Active |
| CLAIM-455 | For one bounded observer, a transformation of a representation can make one task’s distinction legible and another’s illegible at once, so legibility regimes are ordered only relative to a task, and a regime that improves one perspective’s competence can disable another’s | Stated in CASE-154’s title and its “What it can show” and held by no claim, found in the audit of 2026-10-27 for propositions cases state that no claim holds. With two fair bits and an observer that reads one coordinate and cannot compute XOR, the raw pair gives 50% error on the XOR task and 0% on the first bit, and the XOR gives the reverse. CLAIM-199 says a garbling can raise usable information for a receiver; this adds that the same garbling lowers it for another task of the same receiver. The edge: one observer, two tasks; nothing about which task matters. | Active |
| CLAIM-459 | An input to a bistable balance of authority can be sincere and causally efficacious without changing which regime operates: below a threshold the structure of the competition sets, it moves the balance within the current basin, so a higher-order volition that changes nothing visible need not be idle | Stated in CASE-112’s title and its first “What it can show” bullet and held by no claim, found in the audit of 2026-10-27. In ẇ = μw − w³ + h with μ > 0, the operative regime changes only when |h| passes 2μ^{3/2}/(3√3); for 0 < h < h_c the input moves w and leaves the regime. A8 of the exchange of 2026-10-17: “An endorsement can be sincere and causally efficacious while remaining too weak to shift the system out of its current attractor basin.” The edge: efficacy on the balance, not on behaviour; the claim says a null in behaviour is no evidence of a null input. | Active |
| CLAIM-463 | Restoring the conditions under which a capacity was lost need not restore the capacity: where the activity that maintains it reinforces itself, the same conditions can sustain both the capacity and its absence, so a capacity lost past a threshold returns only under conditions better than those it was lost under, or by rebuilding the activity directly, and restoring it can cost more than preserving it would have | A4’s claim candidate from the exchange “About the Chagossians” of 2026-10-26, “Resource restoration does not imply organizational restoration”, which the first filing declined because CASE-111 already derives it; the owner asked for it to be filed as a claim with its connection to the case. Demonstrated by CASE-111’s normal form: in the bistable range, returning the parameter to its pre-collapse value leaves maintenance at zero. The edge: a possibility shown in a model, not a finding about any grant program, people or institute, and a slow recovery is not evidence of the threshold. | Active |
| CLAIM-480 v2 | Composing interactions can turn an interior into a constraint, and sometimes into a constraint of a new kind: in the chain E = ½k₁(x₁ − x₂)² + ½k₂(x₂ − x₃)², with x₂ internal and equilibrated, the boundary sees E_eff = ½·k₁k₂/(k₁ + k₂)·(x₁ − x₃)², a coupling between x₁ and x₃ that neither part has; if the interior relaxes slowly, eliminating it leaves a history-dependent constraint, a memory kernel, so interconnection fixes an effective constraint on the boundary’s trajectories and separation of time scales decides only whether that constraint is approximately memoryless; and for cellular sheaves the reduced constraint can relate three boundary values where every relation among the parts is pairwise | From A22 §4 and A23 §IV of the exchange of 2026-10-27, which called composition “the most important missing mechanism” after the owner’s U20. Eliminating an equilibrated interior leaves an effective boundary relation that no part has; an outside system interacting only with the boundary meets it as a constraint, and “The interior has effectively become apparatus.” Baez and Fong’s black-boxing functor is the general theorem for passive linear networks. The memory case ties it to Mori–Zwanzig and to CLAIM-478: interconnection yields an effective constraint on the boundary’s trajectories in every case, and separation of time scales decides only whether it is approximately memoryless. For cellular sheaves the reduction can need a richer relational vocabulary than the parts used (THEORY-1090; QUESTION-283). It is independent evidence for CLAIM-477, that constraints are realized by organization. | Active |
| CLAIM-481 | Coupling does not by itself separate time scales: two identical undamped oscillators coupled with stiffness k have normal modes at ω₀ and √(ω₀² + 2k), nearly equal when k ≪ ω₀², and neither decays; separation comes where the coupling dissipates relative differences faster than collective ones, as in ẋ = −ax − k(x − y), ẏ = −ay − k(y − x), whose collective and relative modes relax at rates a and a + 2k, and in a Laplacian flow ẋ = −κLx, where the spectrum of L sets the rates | The answer A20, A21 and A23 gave to the owner’s conjecture at U20 that coupling necessarily induces scale separation (CLAIM-473). Two small computations: conservative coupling splits modes without separating or damping them, and dissipative coupling damps the relative mode at a rate that grows with the coupling. For a Laplacian flow the kernel is exact compatibility and the nonzero spectrum is the hierarchy of relaxation rates, so a spectral gap is where the separation comes from. Conceded, so Active. It gives a mechanism for CLAIM-478’s second half and is not a test of individuality. | Active |
| CLAIM-495 | For an O(d)-valued connection on a connected graph, the assignments that meet every edge relation xᵢ = Tᵢⱼxⱼ form a space isomorphic to the vectors at one vertex fixed by every holonomy based there, which is the kernel of the connection Laplacian; for d = 1, and for U(1) phases, that space is all or nothing, but for d > 1 a connection whose loops are not trivial can still keep a subspace on which every loop is, so compatibility among local frames can be partial and graded by direction; and edge maps that do not commute can compose to a flat loop, so non-commutativity is not holonomy | From A2 and A3 of the exchange “Paper assessment” (2026-10-29), which drew it out of LIT-1820’s O(d) setting against THEORY-1082’s U(1) one. The proof and the example are the record’s, checked numerically. It is the zeroth cohomology of the connection sheaf, so standard, and it says nothing about which organizations have such transports. | Active |
| CLAIM-496 | Agreement between routes of translation is a measure of coherence, not of fidelity, and the two are independent: a rendering that sends every source utterance to one output makes direct and relay transport agree exactly while a receiver who could tell a promise from a prediction with no error is left at chance, and two routes that keep the act and differ in register disagree completely while losing nothing the act-identification task needs | From A4 and A5 of the exchange “Paper assessment” (2026-10-29), where the assistant carried LIT-1820’s comparison of alternative paths over to translation and then warned that “path agreement is a coherence criterion, not a fidelity criterion”. The constructions are the record’s. It does not say path discrepancy is uninformative: it says it measures something CLAIM-259’s directed loss does not, and conversely. | Active |
| CLAIM-497 v2 | A system’s causes are found at the coarsest description whose interventions are unambiguous: a macro description is causally proportionate where it merges only micro states that intervention cannot tell apart, and no coarse-graining carries more causal information than the micro description it supervenes on | The record’s position on causal emergence after THEORY-787 was rejected. A coarse-graining can be the right causal description, but not by beating the micro. Under one intervention distribution at both levels, no coarse-graining of a transition carries more effective information than its micro (THEORY-1167). The ones that lose nothing are exactly causal feature learning’s partitions, which merge only micro states with the same interventional effects (THEORY-1169). The step from there to “the causes are found at the coarsest such description” is the record’s own, in the sense of proportionality. It does not say the macro adds causal information. | Active |
| CLAIM-502 | Multi-hop delegation can keep an operator’s authority exactly independent of its own endorsements: under leave-one-out damped delegation, which counts the authority reaching each operator only along paths that have not already passed through it, an operator’s endorsements raise the authority of every operator the endorsed one reaches and leave its own unchanged | A construction that shows CLAIM-282’s old title false. Compute each operator’s authority from A2’s damped recursive rule with that operator’s own outgoing endorsements removed. Then authority reaches it along paths of any length, but only along paths that have not passed through it, so nothing it endorses can return to it. Its authority is exactly independent of its own endorsements, and an endorsement moves authority two hops and more. It does not say the rule conserves authority, that it closes the social route of moving others’ endorsements, or that CLAIM-375’s guarantee carries over. | Active |
| CLAIM-503 | Over finitely many states and finite outcome sets, one experiment is ε-deficient relative to another in every decision problem with losses in [0, 1] exactly when a single Markov kernel brings it within ε in total variation at every state, so the deficiency computed as a linear programme is the least such ε | Le Cam’s randomization criterion for finite experiments, uniform in the state, with losses in [0, 1]. The least ε for which some kernel brings M within ε of N in total variation at every state, which CLAIM-381 computes as a linear programme, equals the least ε for which M is ε-deficient relative to N in every finite decision problem. One direction composes N’s rule with the kernel; the other is one minimax step over a bilinear objective. At ε = 0 it is the finite Blackwell order (THEORY-717). | Active |
| CLAIM-515 | Whether an organization has a cognitive property is four questions, not one: what information about the property can be recovered from its behaviour, what capabilities it can actually exercise, what mechanisms realize them, and whether it can maintain or transform those mechanisms; the four can come apart, and none should be silently substituted for another | The general principle the owner drew at the second review of the filing of the exchange “Analyze nucleation concepts” (2026-10-31): “When asking whether an organization ‘has’ a cognitive property, distinguish the information recoverable from its behavior, the capabilities it can actually exercise, the mechanisms realizing those capabilities, and its ability to maintain or transform those mechanisms.” It was drawn from the world-model case (CLAIM-514) and judged “sufficiently general—and sufficiently testable—that it deserves an explicit place in Nucleation’s research program”. Proposed and interpretive: its instances are demonstrated case by case, and the generalization is not. | Proposed |
| CLAIM-516 | Where every constraint capacity turns over at a positive rate and is produced only by others, with no basal production and production vanishing as its producers do, capacities whose production graph is acyclic all decay to zero: a capacity that persists needs a productive cycle among its ancestors, and removing a feedback edge set extinguishes every capacity; a cycle is necessary and not sufficient, and a one-member cycle is not closure | From A10 §1 and §5A of the exchange “Analyze nucleation concepts” (2026-10-31), after the owner’s “try to progress the theoretical program”. The proof was rechecked by the reader of U7–U10. It is known in substance from autocatalytic-set theory, as A10 said. What it adds to the record is the boundary with closure: the necessary cycle may be a self-loop, and a self-loop is not closure in TERM-070’s sense or CLAIM-395’s, which need two or more constraints. | Active |
| CLAIM-517 | A successor regulator keeps a viable set positively invariant if, at every boundary context, its response distribution is within γ/L in total variation of a regulator that certifies inward margin γ, the dynamics being L-Lipschitz in that distribution; an average distillation loss certifies this only through coverage of the boundary contexts, so fidelity measured in ordinary operation certifies nothing about viability | From A16 §4 and §5 of the exchange “Analyze nucleation concepts” (2026-10-31), the owner having quoted its conclusion at U17: “preserving a regulator’s response distribution sufficiently accurately can guarantee preservation of a physical organization’s viability”. The algebra was rechecked; the regularity conditions A16 left implicit are stated here. A16’s application to its own construction (γ = 0.16) was notional, since that model has no regulator distribution and L is undefined there. | Active |
| CLAIM-521 | An organization’s present coupling strength need not reveal the stability of the process maintaining it: in a constructed system whose maintained connector and capacities set a three-unit coupling, the recovery rate of the maintenance goes to zero at its saddle-node while the coupling’s spectral gap there stays near 0.989 | From A12 §6 of the exchange “Analyze nucleation concepts” (2026-10-31), which called it “a surprising result”. It is generic for any saddle-node whose fold sits away from zero with no feedback from the coupled units to the maintaining core, so it is built in rather than surprising. It is filed because it is a construction QUESTION-125 asks for: one in which the regime changes while the block’s gap does not. All numbers were recomputed. | Active |
| CLAIM-522 v2 | Recovery under interventions on state cannot show where maintenance is realized: an outside controller that measures the constraint capacities and supplies the internal production law as state feedback yields the same trajectory under every perturbation of state, damage and recovery included, even with every capacity observed exactly; only interventions on the production mechanisms, relative to a declared boundary, separate the two | From A10 §4 (its Proposition 5) of the exchange “Analyze nucleation concepts” (2026-10-31). Close to definitional, and checked: substituting the controller gives the same differential equation. It extends CLAIM-414 from passive observation to every intervention on state. As the reader noted, the causal dependence among the capacities is the same in both systems; only where the mechanism sits relative to the boundary differs, which is CLAIM-194’s point about boundaries. | Active |
| CLAIM-524 | Whether functionally equivalent realizations are connected depends on which paths are admissible: for f = u₁v₁ + u₂v₂ with nonnegative weights, (1, 1, 0, 0) and (0, 0, 1, 1) are joined by a curve along which f stays 1 though the straight line between them falls to ½, and under any resource budget below √2 times the endpoints’ cost no path with f = 1 joins them | From the owner’s remark at U17 of the exchange “Analyze nucleation concepts” (2026-10-31), that the reconstruction fibres “reminds me of loss landscape connected mode stuff (which might just be a consequence of high dimensional geometry? … probably already discussed on the nucleation record)”. It is: see below. A17 §9 gave the curve and A18 §7 the budget for a tied-weight family; the general bound here is the reader’s, rechecked. “Viable mode connectivity” is A18’s name and is not filed as a term. | Active |
| CLAIM-528 | That a mechanism realizes a capacity can mean that the capacity is produced through it over time or that the capacity consists in it at a time, and interventions show only the first: a pathway whose disabling changes the capacity after a delay is a cause, can be intervened on surgically, and can be located relative to a declared boundary, whereas every intervention on a capacity that consists in its parts also changes some part along another path, so that relation is underdetermined by intervention data and is inferred by abduction or from simultaneity and settled parthood; and which of the two a given dependence is depends on the time resolution at which a description treats the parts’ response as instantaneous | Filed on reading the fat-handedness debate after Craver: Baumgartner & Gebharter 2016, Baumgartner & Casini 2017 and Baumgartner, Casini & Krickel 2020. It says what their result does and does not reach in the record. CLAIM-522’s conclusion is causal and is left standing; CLAIM-515’s “realized”, read synchronically, is not shown by discriminating interventions; and the debate’s own line between causation and constitution, simultaneity, moves with the time scale, as CLAIM-480’s chain shows. It does not say that constitution is unknowable, or that the record’s organizations have constituents in the mechanists’ sense. | Proposed |
| CLAIM-530 | Where maintenance is realized is a question of constitutive relevance only through its parthood condition: an internal producer and an outside controller supplying the same production law as state feedback meet mutual manipulability’s two manipulability conditions identically, since disabling or stimulating either changes the maintained capacities and intervening on the capacities changes the activity of either, so what puts one inside and the other outside is parthood, a boundary the experiments presuppose and do not find | The record’s join of CLAIM-522 with the mechanist account of constitution, from Craver and Bechtel’s admission that interlevel experiments are symmetrical and show componency only prima facie (THEORY-1287). In CLAIM-522’s pair the bottom-up and top-down interventions come out alike for producer and controller. So the criterion separates them only by its parthood clause, which is CLAIM-522’s declared boundary. Proposed: the criterion’s statement is taken from secondary sources. It does not say no criterion can draw the boundary. Closure, a producer that is itself maintained by what it produces, is a candidate that CLAIM-522’s one-variable model cannot express. | Proposed |
| CLAIM-535 | An organization that acts on what it observes has a policy acceptable in every circumstance exactly when, for each observation, the circumstances it cannot tell apart share an acceptable action; so every circumstance having an acceptable action does not give a policy, a finer observation never removes one, and where acceptable actions overlap the coarsest observations that suffice are several and incomparable, so which distinctions must be made legible depends on which acceptable action the organization commits to | A33 §4 of the exchange “Assess SDT Coverage” (2026-11-02), at the owner’s U33 request for positive accounts, with A30 §2’s example and A32’s Claim B. Its second appearance in the exchange, so it gets a code. The criterion is elementary, and its monotonicity is Blackwell’s easy direction (THEORY-1096). What it adds to CLAIM-350, its single-valued case, is found by the reader of U32–U33: with overlapping acceptable sets, legibility is relative to policy. A positive characterization of what legibility must supply for action. | Active |
| CLAIM-536 | A categorical model of perspectives and an LDA-style model differ in where the mixture lives: the first draws every act’s perspective from one fixed categorical distribution, the second gives each episode its own mixture, drawn from a Dirichlet, and draws each act from it; the Dirichlet is a distribution over mixtures, not the perspectives themselves, and with symmetric concentration a over K perspectives two acts in one episode share a perspective with probability (a + 1)/(Ka + 1) against 1/K across episodes, so the concentration runs from one perspective per episode, as in persona selection (a → 0), to a fresh independent draw for every act (a → ∞) | The owner’s U12 of the exchange “Assess SDT Coverage” (2026-11-02), stated precisely: “H1 isn’t LDA like though. that’s my point. it’s a categorical distribution. the topics in H@ are the dirichlet distribution which parameterizes the categorical distribution, which is what distinguishes H2 from H1 and why H2 is an LDA.” The structure is right; one word is not. The Dirichlet is a distribution over each episode’s mixture, and the topics, here the perspectives, are the categories that mixture weights. The agreement formula, derived here and checked by simulation, shows that both readings of H1 are limits of H2. | Active |
| CLAIM-537 | In tolerance-bounded compatibility on a graph with disturbed offsets, |Dx − a₀ − w| ≤ δ for |w_e| ≤ η_e, a compatible configuration exists for every disturbance exactly when |cᵀa₀| + Σ η_e|c_e| ≤ Σ δ_e|c_e| on every cycle, and one configuration serves every disturbance exactly when η ≤ δ and |cᵀa₀| ≤ Σ(δ_e − η_e)|c_e|; an agent that chooses its configuration from what it observes needs the second condition within each class of disturbances it cannot tell apart, so whether a compatible state can be selected lies between the two and depends on what the agent can read | A30 §§2–3 of the exchange “Assess SDT Coverage” (2026-11-02), at the owner’s “keep going”, with the second and third conditions derived by the reader of U30–U31 and checked by simulation (500 random instances, no mismatch). It is the first concrete instance of the branch’s note of 2026-11-02 on CLAIM-354, that a reachable set should range over the policies an organization can carry out on what it can read. The framing is adjustable robust optimization (LIT-1975). Static selection only, scalar overlaps only. | Active |
| CLAIM-538 | On a finite graph of contexts whose overlaps carry preferred real offsets a and tolerances δ_e ≥ 0, an assignment within every tolerance, |Dx − a| ≤ δ, exists exactly when no cycle’s holonomy exceeds the sum of its tolerances; with every δ_e > 0 the least uniform enlargement of the tolerances that admits one is the largest ratio, over simple cycles, of holonomy to summed tolerance, so the obstruction is certified by a cycle; the feasible allocations of tolerance depend on the offsets only through their class in H¹, so local adjustment can move a mismatch among the overlaps of a cycle and never remove it; and an evaluation that omits one overlap of every witnessing cycle reports a compatibility that does not hold | The owner’s U27 of the exchange “Assess SDT Coverage” (2026-11-02), “compatibility of tolerance bounds”, with U28’s pointer to the record’s sheaf tools, worked out at A28 §3 and A29 §§3–4, 7 and A32’s Claim A. The feasibility criterion is the classical feasible-differential theorem of network flows (Rockafellar 1984; no checked identifier, not registered), and the ratio is LP duality, so the result is granted, not the record’s discovery. What the record adds is the reading: the frontier is an invariant of the obstruction class, governance chooses where on it the mismatch falls, and omission can fake compatibility. Checked by linear programming against brute-force cycle enumeration on random graphs. | Active |
| CLAIM-540 | The persistence a Dirichlet–categorical hierarchy gives a run of selections is an observer’s learning about a fixed mixture, not a change in what selects: under a uniform prior an observed first choice raises the predicted chance of repeating it from 1/2 to 2/3, while the same choice forced leaves it at 1/2, so the hierarchy’s reinforcement is not evidence of self-formation or of a maintained will-organization, which need the mixture itself to change under the choice | A13 of the exchange “Assess SDT Coverage” (2026-11-02), answering the owner’s U13, what the Dirichlet–categorical reading implies for Jennings’s attention as the source of a free will. A13: “Self-formation requires a causal update, not merely a posterior update in the observer.” Correct, short and not held. It limits the record’s own H1/H2 reading (CLAIM-536) and A11’s suggestion that LDA already has “a primitive analogue of a will-organization”. Rechecked here by simulation. | Active |
| CLAIM-542 | Sensitivity does not fix the allocation or the authorship of control: linearized, every input reaches the final state through the same transition operator, ∂x_T/∂w_t = Φ(T, t)B_w(t), so systems with the same Lyapunov spectrum can differ arbitrarily in what an agent’s own channel, an outside party’s channel and noise each reach and at what least energy; and along one unstable direction a larger exponent λ lowers the least effort to reach a displacement, since W_T = b²(e^{2λT} − 1)/2λ grows with λ, while a constant open-loop input against white noise entering the same direction has terminal signal-to-noise ratio tending to (bu/σ)√(2/λ), which falls as λ grows, so only feedback that observes the state turns cheaper control into reliable direction | The demonstrated part of the owner’s U16 and U20 of the exchange “Assess SDT Coverage” (2026-11-02), “controllability vs sensitivity”. A20 §2 stated the impulse case (“the relative contribution is not determined by e^{λT}”) and A22 §5 the channel case. The scalar signal-to-noise result is the reader’s derivation, checked by simulation. It sharpens the owner’s point: sensitivity makes control cheaper and, without feedback, less decisive against noise. CLAIM-345 holds the cost formula and not the allocation across channels. | Active |
| CLAIM-546 | Negative frequency dependence that keeps two types in coexistence does not make their interaction cooperative: in Hawk–Dove with cost C above value V each type increases when rare and they coexist at the interior equilibrium of V/C Hawks, with Hawks still taking the whole resource from Doves, and on the Black Queen account producers of a leaked good and the cells that use it without paying coexist at a frequency set by the community’s need; so whether types persist together and whether they cooperate are separate questions, and a coexistence criterion answers only the first | From §3 of the synthesis “The Evolution of Cooperation: From Coexistence to Symbiosis to Morality”, which the owner supplied on 2026-11-04: “stable coexistence of strategies is not necessarily the same as stable cooperation: cooperators and defectors can coexist without eliminating exploitation”. The record held the ecological half, invasion growth when rare (CASE-118), and a case of the point (the Black Queen, THEORY-1048). It did not hold the point itself. Hawk–Dove makes it exact. The wording is the reader’s; adopting it is the owner’s call. | Proposed |