--- slug: can-these-all-be-true passed: 2026-11-16 by: claude-wonderful-fermi-jahagn, two independent agents on oversight/claims-pass.md (one for the method page; one for the layer, the tool page, the catalogue entry, the callable face and the narration), with the reports kept in full in research/can-these-all-be-true/review/ checked: 213 confirmed: 169 wrong: 32 unverifiable: 0 observed: 12 stale: 0 --- # Claims pass: Can These All Be True? Live page: not deployed yet when the pass ran; both agents read the pages on a local copy and ran the reader check (the checker downloaded into an empty folder, its self-test, saved records checked) there. The agents' own tallies were 129 claims for the method page (105 confirmed, 2 wrong, 11 minor, 11 observed) and 100 for the rest (80 confirmed, 5 wrong, 14 minor, 1 observed). Their reports name most confirmed claims by their source-record id and some in words, and count a few they do not name. The lines below are every claim a report names, one line each: a claim named by id takes its wording and source from the source record the agents read (commit f57e8af425), one named in words takes the report's words, and every verdict and note is the report's. MINOR lines are counted as wrong here, because each led to a change. This file is written by `research/can-these-all-be-true/review/make-claims-record.py`. ## Claims - CONFIRMED | a distinct method determining the necessary conditions among their data, in order that they may represent a possible experience, and assigning the corresponding limits of the final solutions. | https://www.gutenberg.org/files/15114/15114-pdf.pdf (kept: research/can-these-all-be-true/sources/boole-laws-of-thought-pg15114.txt) | Laws of Thought ch. XVIII §1 and ch. XIX §1, word for word. - CONFIRMED | By the term statistical conditions, I mean those conditions which must connect the numerical data of a problem in order that those data may be consistent with each other, and therefore such as statistical observations might actually have furnished. | https://www.gutenberg.org/files/15114/15114-pdf.pdf (kept: research/can-these-all-be-true/sources/boole-laws-of-thought-pg15114.txt) | Laws of Thought ch. XVIII §1 and ch. XIX §1, word for word. - CONFIRMED | Eight years later, in a paper read to the Royal Society | https://jimclifford.ca/philosophical-transactions-ocr-1665-1869/articles/jstor-108830/ (kept: research/can-these-all-be-true/sources/boole-1862-phil-trans-ocr-jimclifford.txt) | "Received June 19,—Read June 19, 1862". - CONFIRMED | These, together with the former, may be termed the conditions of possible experience. When satisfied they indicate that the data may have, when not satisfied they indicate that the data cannot have, resulted from actual observation. | https://archive.org/download/jstor-108830/108830_djvu.txt (kept: research/can-these-all-be-true/sources/boole-1862-jstor-early-journal-content-djvu.txt) | word for word. The term is introduced with "may be termed". The 1857 Edinburgh memoir was not read, so an earlier use there cannot be ruled out. - CONFIRMED | Let the events in the data be resolved into the ultimate possible alternatives which they involve | https://jimclifford.ca/philosophical-transactions-ocr-1665-1869/articles/jstor-108830/ (kept: research/can-these-all-be-true/sources/boole-1862-phil-trans-ocr-jimclifford.txt) | word for word. - CONFIRMED | as the probability of each event in the data is equal to the sum of the probabilities of the alternatives which it involves, we shall have a system of equations | https://jimclifford.ca/philosophical-transactions-ocr-1665-1869/articles/jstor-108830/ (kept: research/can-these-all-be-true/sources/boole-1862-phil-trans-ocr-jimclifford.txt) | word for word. - CONFIRMED | the ideal events are regarded as independent | https://archive.org/download/jstor-108830/108830_djvu.txt (kept: research/can-these-all-be-true/sources/boole-1862-jstor-early-journal-content-djvu.txt) | word for word. - CONFIRMED | it calls it A problem dating back to Boole | https://arxiv.org/abs/2605.03556 (kept: research/can-these-all-be-true/sources/union-interval-np-hard-arxiv-abs.txt) | word for word. - CONFIRMED | says that Hailperin, in 1965, phrased the problem in the language of linear programming and generalized it to logical formulas of the events other than disjunction | https://arxiv.org/abs/2605.03556 (kept: research/can-these-all-be-true/sources/union-interval-np-hard-arxiv-abs.txt) | word for word. - CONFIRMED | The same paper shows that computing the optimal interval for the union probability is NP-hard | https://arxiv.org/abs/2605.03556 (kept: research/can-these-all-be-true/sources/union-interval-np-hard-arxiv-abs.txt) | word for word. - CONFIRMED | A standard reference traces the argument to Frank Ramsey (it traces to Ramsey’s work in “Truth and Probability”) | https://plato.stanford.edu/entries/dutch-book/ (kept: research/can-these-all-be-true/sources/sep-dutch-book.txt) | word for word. The page's split of credit (the argument traced to Ramsey, betting quotients and coherence to de Finetti) matches the Stanford Encyclopedia: "De Finetti identified degrees of belief with betting quotients and termed … incoherent". - CONFIRMED | Given a set of betting quotients that fails to satisfy the probability axioms, there is a set of bets with those quotients that guarantees a net loss to one side. | https://plato.stanford.edu/entries/dutch-book/ (kept: research/can-these-all-be-true/sources/sep-dutch-book.txt) | word for word. The page's split of credit (the argument traced to Ramsey, betting quotients and coherence to de Finetti) matches the Stanford Encyclopedia: "De Finetti identified degrees of belief with betting quotients and termed … incoherent". - CONFIRMED | a 2026 study cited in section 5 calls the result it builds on a theorem due to de Finetti | https://arxiv.org/abs/2609.02797 (kept: research/can-these-all-be-true/sources/andrews-sarkar-2026-arxiv-abs.txt) | word for word. - CONFIRMED | a probability assessment is coherent if and only if, in any finite combination of n bets, it may not happen that the values […] are all positive, or all negative (no Dutch Book). | https://arxiv.org/pdf/1304.4990 (kept: research/can-these-all-be-true/sources/gilio-sanfilippo-arxiv-1304.4990.txt) | word for word; the gloss follows Definition 1. - CONFIRMED | then you agree to pay an amount p, by receiving 1, or 0, or p, according to whether AB is true, or ABc is true, or Ac is true (bet called off). | https://arxiv.org/pdf/1304.4990 (kept: research/can-these-all-be-true/sources/gilio-sanfilippo-arxiv-1304.4990.txt) | word for word, in §2.1. - CONFIRMED | that paper’s reference list points to earlier work on “coherence checking and propagation of lower probability bounds” (Biazzo, Gilio and Sanfilippo, 2003), which was not read for this page. | https://arxiv.org/pdf/1304.4990 (kept: research/can-these-all-be-true/sources/gilio-sanfilippo-arxiv-1304.4990.txt) | reference [3], word for word. - CONFIRMED | with P(B) = 0 and P(A / B) = 0.3, the answer to P(A / B) is 0.3, where a ratio gives 0/0. | research/can-these-all-be-true/verify.mjs | both examples worked by hand from Theorem 3 and agree. - CONFIRMED | With P(A and B) = 0, the answer to P(A / B) is 0 if B has any chance, and anything at all if it has none | research/can-these-all-be-true/verify.mjs | both examples worked by hand from Theorem 3 and agree. - CONFIRMED | they perform 245 trials testing the CHSH-Bell inequality and find S = 2.42 ± 0.20 | https://arxiv.org/pdf/1508.05949 (kept: research/can-these-all-be-true/sources/hensen-2015-arxiv-1508.05949.txt) | word for word or as stated. - CONFIRMED | they report a probability of p = 0.039 that a local-realist model for space-like separated sites produces data with a violation at least as large as observed | https://arxiv.org/pdf/1508.05949 (kept: research/can-these-all-be-true/sources/hensen-2015-arxiv-1508.05949.txt) | word for word or as stated. - CONFIRMED | In the experiment reported from Delft in 2015 | https://arxiv.org/pdf/1508.05949 (kept: research/can-these-all-be-true/sources/hensen-2015-arxiv-1508.05949.txt) | word for word or as stated. - CONFIRMED | The counts of agreeing and disagreeing answers are printed above the bars of the paper’s Figure 4a: 46 and 7, 63 and 16, 46 and 16, 10 and 41. | https://arxiv.org/pdf/1508.05949 (kept: research/can-these-all-be-true/sources/hensen-2015-arxiv-1508.05949.txt) | I read the image myself and got the same numbers: "+46, −7", "+63, −16", "+46, −16", "+10, −41" over a,b = (0,0), (0,1), (1,0), (1,1). The caption says "Numbers above bars represent the amount of correlated and anti-correlated outcomes respectively". (1,1) is the negative bar, so the anti-correlated pair is the fourth statement, P(a2 iff b2) = 10/51, as in examples.mjs. The mapping is right. The counts sum to 245; k = 196 (the paper's text has "we find k = 196"); S = 2.4224. - CONFIRMED | The theory’s own prediction for the ideal experiment, the paper says, is S = 2√2, about 2.83. | https://arxiv.org/pdf/1508.05949 (kept: research/can-these-all-be-true/sources/hensen-2015-arxiv-1508.05949.txt) | "quantum theory predicts a value of S = 2√2". "Ideal" is a fair gloss: the paper's dotted lines give lower values expected under real fidelities. - CONFIRMED | the CHSH inequality of Clauser, Horne, Shimony and Holt (1969) | https://api.crossref.org/works/10.1103/PhysRevLett.23.880 (kept: research/can-these-all-be-true/sources/crossref-chsh-1969.json) | Crossref record. - CONFIRMED | Some of Boole's conditions have been rediscovered in more recent years by physicists, including Bell inequalities, Clauser Horne inequalities, and many others. | https://api.openalex.org/works/doi:10.1093/bjps/45.1.95 (kept: research/can-these-all-be-true/sources/openalex-abstracts.txt) | word for word. - CONFIRMED | the abstract of his 1994 paper, George Boole’s ‘Conditions of Possible Experience’ and the Quantum Puzzle | https://api.openalex.org/works/doi:10.1093/bjps/45.1.95 (kept: research/can-these-all-be-true/sources/openalex-abstracts.txt) | word for word. - CONFIRMED | Arthur Fine had shown in 1982 that, for an experiment of this kind, the statement There is one joint distribution for all observables of the experiment, returning the experimental probabilities is equivalent to the statement The Bell inequalities hold | https://api.openalex.org/works/doi:10.1103/PhysRevLett.48.291 (kept: research/can-these-all-be-true/sources/openalex-abstracts.txt) | statements (3) and (5) of Fine's list of "mutually equivalent" statements. - CONFIRMED | finding the inequality by linear programming was described in 2009 by Matthew Elliott, whose program, when the correlations fit no such world, outputs a Bell inequality violated by the correlations | https://arxiv.org/abs/0905.2950 (kept: research/can-these-all-be-true/sources/elliott-arxiv-0905.2950-abs.txt) | word for word. - CONFIRMED | Pitowsky, on page 113 of the 1994 paper as Samson Abramsky quotes it: | https://arxiv.org/pdf/2010.13326 (kept: research/can-these-all-be-true/sources/abramsky-arxiv-2010.13326.txt) | Abramsky's [23] is the 1994 BJPS paper, so page 113 is that paper's; the quotation is exact. Abramsky then says Pitowsky puts it "too strongly", and the page's own note takes the same line, so the use is fair. - CONFIRMED | To find out the union, intersection, and other related probabilities of two independent events. | https://www.calculator.net/probability-calculator.html (kept: research/can-these-all-be-true/sources/calculator-net-probability.txt) | word for word, in context. - CONFIRMED | Your problem needs to be condensed into two independent events. | https://www.omnicalculator.com/statistics/probability (kept: research/can-these-all-be-true/sources/omnicalculator-probability.txt) | word for word, in context. - CONFIRMED | His first worked example in the chapter that illustrates the general method | https://www.gutenberg.org/files/15114/15114-pdf.pdf (kept: research/can-these-all-be-true/sources/boole-laws-of-thought-pg15114.txt) | Ex. 1; the solution is q + c(1−p). - CONFIRMED | The probability that it thunders upon a given day is p, the probability that it both thunders and hails is q, but of the connexion of the two phænomena of thunder and hail, nothing further is supposed to be known. Required the probability that it hails on the proposed day. | https://www.gutenberg.org/files/15114/15114-pdf.pdf (kept: research/can-these-all-be-true/sources/boole-laws-of-thought-pg15114.txt) | Ex. 1; the solution is q + c(1−p). - CONFIRMED | Inferior limit = q. Superior limit = q + 1 − p. | https://www.gutenberg.org/files/15114/15114-pdf.pdf (kept: research/can-these-all-be-true/sources/boole-laws-of-thought-pg15114.txt) | Ex. 1; the solution is q + c(1−p). - CONFIRMED | the verifier compares the instrument with those two formulas for every p and q in tenths. | research/can-these-all-be-true/verify.mjs | the grid is tenths. - CONFIRMED | A Tangled Tale (Macmillan, 1885), Knot X | https://www.gutenberg.org/files/29042/29042-h/29042-h.htm (kept: research/can-these-all-be-true/sources/carroll-tangled-tale-pg29042.txt) | word for word. - CONFIRMED | Say that 70 per cent. have lost an eye—75 per cent. an ear—80 per cent. an arm—85 per cent. a leg—that'll do it beautifully. Now, my dear, what percentage, at least, must have lost all four? | https://www.gutenberg.org/files/29042/29042-h/29042-h.htm (kept: research/can-these-all-be-true/sources/carroll-tangled-tale-pg29042.txt) | word for word. - CONFIRMED | The printed answer is Ten. | https://www.gutenberg.org/files/29042/29042-h/29042-h.htm (kept: research/can-these-all-be-true/sources/carroll-tangled-tale-pg29042.txt) | word for word. - CONFIRMED | with a solution Carroll credits to a reader signing as Polar Star: Adding the wounds together, we get 70 + 75 + 80 + 85 = 310, among 100 men; which gives 3 to each, and 4 to 10 men. | https://www.gutenberg.org/files/29042/29042-h/29042-h.htm (kept: research/can-these-all-be-true/sources/carroll-tangled-tale-pg29042.txt) | word for word. - CONFIRMED | The instrument’s proof of the lower end puts weight 1 on each of the four statements, which is the same sum. | research/can-these-all-be-true/verify.mjs | word for word. - CONFIRMED | Another reader, Janet, multiplied the four percentages together. | https://www.gutenberg.org/files/29042/29042-8.txt (kept: research/can-these-all-be-true/sources/carroll-tangled-tale-pg29042-plain.txt) | 7/10ths in the plain-text file, 2⁄10ths in the HTML; the product is 0.357. - CONFIRMED | Project Gutenberg’s plain-text file prints her answer as 35 and 7/10ths | https://www.gutenberg.org/files/29042/29042-8.txt (kept: research/can-these-all-be-true/sources/carroll-tangled-tale-pg29042-plain.txt) | 7/10ths in the plain-text file, 2⁄10ths in the HTML; the product is 0.357. - CONFIRMED | its HTML file prints 2/10ths | https://www.gutenberg.org/files/29042/29042-h/29042-h.htm (kept: research/can-these-all-be-true/sources/carroll-tangled-tale-pg29042.txt) | 7/10ths in the plain-text file, 2⁄10ths in the HTML; the product is 0.357. - CONFIRMED | Typed in on her reading, with each percentage given the ones before it, the instrument returns the single value 357/1000. | research/can-these-all-be-true/verify.mjs | 7/10ths in the plain-text file, 2⁄10ths in the HTML; the product is 0.357. - CONFIRMED | A study published in 1978 reports that its authors asked 20 house officers, 20 fourth-year medical students and 20 attending physicians, selected in 67 consecutive hallway encounters at four Harvard Medical School teaching hospitals | https://api.openalex.org/works/doi:10.1056/NEJM197811022991808 (kept: research/can-these-all-be-true/sources/openalex-abstracts.txt) | word for word. - CONFIRMED | a question that begins If a test to detect a disease whose prevalence is 1/1000 has a false positive rate of | https://api.openalex.org/works/doi:10.1056/NEJM197811022991808 (kept: research/can-these-all-be-true/sources/openalex-abstracts.txt) | word for word. - CONFIRMED | A 2014 repetition of the study quotes the question in full, with the rate at 5% and the thing asked as what is the chance that a person found to have a positive result actually has the disease | research/assay-line/queue-1014c/bayes-theorem/N1-sources.json | also checked online. https://pmc.ncbi.nlm.nih.gov/articles/PMC4955674/ (robots-policy said allowed; one request; HTTP 200) has both sentences word for word. - CONFIRMED | Assuming a perfectly sensitive test, we calculated that the correct answer is 1.96% | research/assay-line/queue-1014c/bayes-theorem/N1-sources.json | also checked online. https://pmc.ncbi.nlm.nih.gov/articles/PMC4955674/ (robots-policy said allowed; one request; HTTP 200) has both sentences word for word. - CONFIRMED | so the instrument returns a range, from 0 to 20/1019, whose top end is that 1.96…%; add the missing statement and it returns the single value. | research/can-these-all-be-true/verify.mjs | 20/1019 = 1.9627%. - CONFIRMED | Tversky and Kahneman’s 1983 paper, Extensional versus intuitive reasoning: The conjunction fallacy in probability judgment | https://api.openalex.org/works/doi:10.1037/0033-295X.90.4.293 (kept: research/can-these-all-be-true/sources/openalex-abstracts.txt) | word for word. - CONFIRMED | whose abstract states the rule it breaks: The probability of a conjunction, P(A&B), cannot exceed the probabilities of its constituents | https://api.openalex.org/works/doi:10.1037/0033-295X.90.4.293 (kept: research/can-these-all-be-true/sources/openalex-abstracts.txt) | word for word. - CONFIRMED | A 2026 study by Isaiah Andrews and Suproteem Sarkar elicited forecasts from language models about events generated from stock returns | https://arxiv.org/abs/2609.02797 (kept: research/can-these-all-be-true/sources/andrews-sarkar-2026-arxiv-abs.txt) | word for word. - CONFIRMED | used linear programs to compute the largest Dutch-book profit against them | https://arxiv.org/abs/2609.02797 (kept: research/can-these-all-be-true/sources/andrews-sarkar-2026-arxiv-abs.txt) | word for word. - CONFIRMED | its abstract reports substantial evidence of incoherence in language model forecasts | https://arxiv.org/abs/2609.02797 (kept: research/can-these-all-be-true/sources/andrews-sarkar-2026-arxiv-abs.txt) | word for word. - CONFIRMED | a 2015 paper by Glauber De Bona and Marcelo Finger is titled Measuring inconsistency in probabilistic logic: rationality postulates and Dutch book interpretation | https://api.crossref.org/works?query.bibliographic=Measuring+inconsistency+in+probabilistic+logic+rationality+postulates+and+Dutch+book+interpretation+De+Bona+Finger&rows=3&select=DOI,title,author,container-title,issued,volume,page (kept: research/can-these-all-be-true/sources/crossref-search-debona-finger-2015.json) | Crossref record. - CONFIRMED | He supposes medical statistics in which two symptoms x and y are seen together in two fifths of the cases of a disease, y and z in two thirds, and z and x in four fifths | https://jimclifford.ca/philosophical-transactions-ocr-1665-1869/articles/jstor-108830/ (kept: research/can-these-all-be-true/sources/boole-1862-phil-trans-ocr-jimclifford.txt) | the fractions and the condition check out (7/15 > 2/5), and so do the bets. - CONFIRMED | and concludes: The above formulae would show that the evidence was contradictory. | https://archive.org/download/jstor-108830/108830_djvu.txt (kept: research/can-these-all-be-true/sources/boole-1862-jstor-early-journal-content-djvu.txt) | the fractions and the condition check out (7/15 > 2/5), and so do the bets. - CONFIRMED | The condition that fails is that the first share be at least the sum of the other two less 1, and 2/5 is less than 2/3 + 4/5 − 1 = 7/15. | https://jimclifford.ca/philosophical-transactions-ocr-1665-1869/articles/jstor-108830/ (kept: research/can-these-all-be-true/sources/boole-1862-phil-trans-ocr-jimclifford.txt) | the fractions and the condition check out (7/15 > 2/5), and so do the bets. - CONFIRMED | each of the three majorities in a cycle can be as large as two thirds and no larger. The verifier checks the same for cycles of four and five (three quarters, four fifths). | research/can-these-all-be-true/verify.mjs | checked. - CONFIRMED | PrSAT 3.0, by Koissi Adjorlolo and Branden Fitelson: an open source web-based tool for testing the satisfiability of arbitrary sets of statements in the probability calculus. | https://fitelson.org/PrSAT/ (kept: research/can-these-all-be-true/sources/prsat-index.txt) | word for word. "Runs in a browser" is supported ("requires only a web browser"). The description is fair and not belittling. - CONFIRMED | its statements may multiply probabilities, so it takes independence in general | https://fitelson.org/PrSAT/ (kept: research/can-these-all-be-true/sources/prsat-index.txt) | word for word. "Runs in a browser" is supported ("requires only a web browser"). The description is fair and not belittling. - CONFIRMED | it describes itself as a decision procedure, so its solutions are exact | https://fitelson.org/PrSAT/ (kept: research/can-these-all-be-true/sources/prsat-index.txt) | word for word. "Runs in a browser" is supported ("requires only a web browser"). The description is fair and not belittling. - CONFIRMED | Of a set that cannot be satisfied, its page says the user sees Constraints are UNSATisfiable! with a truth table and the constraints in algebraic form. | https://fitelson.org/PrSAT/ (kept: research/can-these-all-be-true/sources/prsat-index.txt) | word for word. "Runs in a browser" is supported ("requires only a web browser"). The description is fair and not belittling. - CONFIRMED | Pinference, an R package by PierGianLuca Porta Mana: Implementation of T. Hailperin's procedure to calculate lower and upper bounds of the probability for a propositional-logic expression, given equality and inequality constraints on the probabilities for other expressions. | https://cran.r-project.org/web/packages/Pinference/index.html (kept: research/can-these-all-be-true/sources/cran-pinference.txt) | word for word. - CONFIRMED | its page says it Requires the 'lpSolve' package | https://cran.r-project.org/web/packages/Pinference/index.html (kept: research/can-these-all-be-true/sources/cran-pinference.txt) | word for word. - CONFIRMED | CkC, the Check Coherence package from the University of Perugia: Check-Coherence software have been designed to check the coherence (consistency) and to perform extensions (inferences) of probabilistic models based on partial conditional lower-upper probability assessments. | https://upkd-software.sites.dmi.unipg.it/paid/software.html (kept: research/can-these-all-be-true/sources/perugia-paid-software.txt) | word for word. - CONFIRMED | Its page gives 32-bit Windows as the platform. | https://upkd-software.sites.dmi.unipg.it/paid/software.html (kept: research/can-these-all-be-true/sources/perugia-paid-software.txt) | word for word. - CONFIRMED | dutchbook checks whether a language model's probabilities agree with each other. | https://raw.githubusercontent.com/PDiTO/dutchbook/main/README.md (kept: research/can-these-all-be-true/sources/pdito-dutchbook-README.md) | word for word. "Command-line" (`uv run dutchbook …`) and "builds its own bank" are both supported. - CONFIRMED | It builds its own bank of questions and reports, in its words, how many cents per dollar staked a bookie can lock in | https://raw.githubusercontent.com/PDiTO/dutchbook/main/README.md (kept: research/can-these-all-be-true/sources/pdito-dutchbook-README.md) | word for word. "Command-line" (`uv run dutchbook …`) and "builds its own bank" are both supported. - CONFIRMED | last pushed to on 3 November 2026 when this was written | research/can-these-all-be-true/github-search-2026-11-15.jsonl | pushed_at 2026-11-03 in the 15 November search reply. - CONFIRMED | The instrument solves the linear programs in exact fractions. | public/tools/can-these-all-be-true/lp.mjs | in the code as stated. - CONFIRMED | Nothing the scout computes is kept. | public/tools/can-these-all-be-true/engine.mjs | in the code as stated. - CONFIRMED | That proposal is then either proved, by two exact solves and a check in whole numbers of every condition that makes an answer the best one, or thrown away, and the exact solver finds its own way. | public/tools/can-these-all-be-true/lp.mjs | in the code as stated. - CONFIRMED | That set is not always the smallest there is. When more than one conflict hides among the statements, which one is reported can depend on the order of the lines. | public/tools/can-these-all-be-true/engine.mjs | in the code as stated. - CONFIRMED | checker.mjs shares no code with it, imports nothing, and contains no solver | public/tools/can-these-all-be-true/checker.mjs | in the code as stated. - CONFIRMED | the tables of cases on the page are computed by the checker’s code | public/tools/can-these-all-be-true/checker.mjs | in the code as stated. - CONFIRMED | Family F is the one that says the checker can fail | research/can-these-all-be-true/verify.mjs | in the code as stated. - CONFIRMED | That comparison now runs every time, as family L. | research/can-these-all-be-true/verify.mjs | in the code as stated. - CONFIRMED | In family K of the verifier, every end of every range is added back as a statement and must pass the instrument’s separate test for statements, and a value just outside must fail it. | research/can-these-all-be-true/verify.mjs | in the code as stated. - CONFIRMED | A problem that would need several times that work is refused at a fixed limit, the same on every device | public/tools/can-these-all-be-true/engine.mjs | in the code as stated. - CONFIRMED | node checker.mjs --selftest runs it against five known records and ten doctored ones | public/tools/can-these-all-be-true/checker.mjs | recreated, 15 of 15. - CONFIRMED | Boole then removes the unknowns by hand, playing one inequality off against another. | research/can-these-all-be-true/sources/boole-1862-phil-trans-ocr-jimclifford.txt | supported by Boole's elimination rule in the 1862 text - CONFIRMED | Shared evenly among the four pairs, that is agreement about 85.4% of the time on three of them and 14.6% on the fourth, and those rates, typed in, are incoherent too. | research/can-these-all-be-true/review/claims-method-report.md | recreated by the agent: 85.4% and 14.6% are (2±√2)/4, and those rates are incoherent (2.414 > 2) - CONFIRMED | The loss is at least 0.2112… in every case, or 0.0528… per unit staked. | research/can-these-all-be-true/review/claims-method-report.md | recreated by the agent with checker.mjs on a Bell record made with the repository engine: holds over all 16 cases; loss 0.2112, or 349600/6619647 = 0.0528 per unit staked - CONFIRMED | In the verifier’s sweep of random agreement rates, every Dutch book the instrument returned against an incoherent set of four was one of the eight inequalities of this form. | research/can-these-all-be-true/verify.mjs | "Eight inequalities": verify.mjs line 253, which passed - CONFIRMED | The two notes of section 4: what the Bell example shows and does not, and that four rates from 245 trials are subject to chance. | research/can-these-all-be-true/sources/hensen-2015-arxiv-1508.05949.txt | The physics in the two notes is sound and not too strong - OBSERVED | A first search for this page, twelve queries on 15 November 2026 (eight on the web and four on GitHub), did not find PrSAT. | research/can-these-all-be-true/prior-art-2026-11-15.md | recreated or read in its record by the agent (report, OBSERVED) - OBSERVED | checks in twelve families: | research/can-these-all-be-true/verify-2026-11-15.txt | recreated or read in its record by the agent (report, OBSERVED) - OBSERVED | The verifier found that the conflicting set reported was a real conflict but could have a statement to spare when conditions of probability zero were involved | research/can-these-all-be-true/README.md | recreated or read in its record by the agent (report, OBSERVED) - OBSERVED | Its own solver, written separately, agreed with this one on 3,000 random problems | research/can-these-all-be-true/screen/REPORT.md | recreated or read in its record by the agent (report, OBSERVED) - OBSERVED | a decimal comma read as a thousands mark, so that “2,5%” became 25% | research/can-these-all-be-true/screen/REPORT.md | recreated or read in its record by the agent (report, OBSERVED) - OBSERVED | “more than” quietly read as “at least” | research/can-these-all-be-true/screen/REPORT.md | recreated or read in its record by the agent (report, OBSERVED) - OBSERVED | questions about a condition of probability zero answered by the ratio alone | research/can-these-all-be-true/screen/REPORT.md | recreated or read in its record by the agent (report, OBSERVED) - OBSERVED | a conflicting set called the smallest when it was only one with nothing spare | research/can-these-all-be-true/screen/REPORT.md | recreated or read in its record by the agent (report, OBSERVED) - OBSERVED | problems inside the stated limits that ran for minutes | research/can-these-all-be-true/screen/REPORT.md | recreated or read in its record by the agent (report, OBSERVED) - OBSERVED | a phrase this page had credited to Boole’s book of 1854 that is in his paper of 1862 | research/can-these-all-be-true/screen/REPORT.md | recreated or read in its record by the agent (report, OBSERVED) - OBSERVED | took about twenty seconds on the machine this was built on | research/can-these-all-be-true/timings-2026-11-15.txt | recreated or read in its record by the agent (report, OBSERVED) - WRONG | resting on a theorem they cite from earlier work (their Theorem 3) | https://arxiv.org/pdf/1304.4990 (kept: research/can-these-all-be-true/sources/gilio-sanfilippo-arxiv-1304.4990.txt) | Kept copy, section 2.3: "Then, we have ([4, Thm 3]) Theorem 5 … Based on Theorem 5 the following algorithm for coherence checking has been given in [4, see Remark 2]. Algorithm 1." Reference [4] is Biazzo, Gilio and Sanfilippo, IPMU’08 (2008); the paper’s own Theorem 3 is cited from [20], Gilio 1993 - MINOR | This is the procedure Gilio and Sanfilippo set out as their Algorithm 1 | https://arxiv.org/pdf/1304.4990 (kept: research/can-these-all-be-true/sources/gilio-sanfilippo-arxiv-1304.4990.txt) | It is their Algorithm 1, but they credit it to earlier work, [4] (Biazzo, Gilio and Sanfilippo, 2008), which the page did not say - WRONG | What this instrument adds is therefore narrower than it first seemed, and is these things: the least and the greatest value of anything asked, with a world at each end; … de Finetti’s reading of conditional statements and of questions; and a link | research/can-these-all-be-true/sources/cran-pinference.txt | Pinference: "to calculate lower and upper bounds of the probability for a propositional-logic expression"; CkC (perugia-paid-software.txt): "check the coherence (consistency) and to perform extensions (inferences) of probabilistic models based on partial conditional lower-upper probability assessments". The prior-art file itself says bounds "were not found in a browser" - MINOR | Lewis Carroll’s readers found exactly such a proof by hand in 1885 | https://www.gutenberg.org/files/29042/29042-h/29042-h.htm (kept: research/can-these-all-be-true/sources/carroll-tangled-tale-pg29042.txt) | The kept preface says "This Tale originally appeared as a serial in The Monthly Packet, beginning in April, 1880", so the readers' solutions came before the book of 1885 - MINOR | An independent agent, sent to attack the idea before it was built on, did. | research/can-these-all-be-true/screen/REPORT.md | screen/REPORT.md: "It ran against the engine as it stood at about 21:45 UTC (commit b43b6129ca…)", that is against the first build - MINOR | The same paper sets out the way of finding them that the instrument follows. | research/can-these-all-be-true/sources/boole-1862-phil-trans-ocr-jimclifford.txt | The 1862 paper says the method "is taken from the 'Memoir on the Combination of Testimonies and of Judgments'", and the symptoms example ends "(Edinburgh Memoir.)" - MINOR | The same paper shows that computing the optimal interval for the union probability is NP-hard, which is why the instrument stops at sixteen events and says so | research/can-these-all-be-true/sources/union-interval-np-hard-arxiv-abs.txt | The abstract also cites the older, more general probabilistic satisfiability result (1988); the stop at sixteen comes more directly from the checker going through all 2^n cases - MINOR | The statements are coherent exactly when no layer fails. | research/can-these-all-be-true/sources/gilio-sanfilippo-arxiv-1304.4990.txt | Theorems 3 and 5 in the kept copy are for precise assessments; the one-sided versions are in [3] (2003) and [4] (2008), not read. Missing support, not an error: families D, K and L test the extension - MINOR | The mathematics is old and public, and there is software for parts of it. What was found, with what each says of itself: | research/can-these-all-be-true/prior-art-2026-11-15.md | The first search also found research solvers ("NILS (Utah, MATLAB), the USP PSAT solvers (Finger and De Bona), Pronto, diff-SAT … not opened"), which the list leaves out - MINOR | The verifier, verify.mjs, runs 17,664 checks in twelve families | research/can-these-all-be-true/verify.mjs | The run is "17685 passed" over thirteen families: family J (21 checks) is outside the figure, and the page did not say so - MINOR | Sixteen events, and a fixed allowance of work. … took about twenty seconds on the machine this was built on. | public/tools/can-these-all-be-true/engine.mjs | LIMITS also refuses past 60 statements, 6 events declared independent at once and a cap on rounds, none of them named; and on a slow device an allowed problem may take minutes, which the research README says and the page did not - MINOR | (The 1978 abstract was read in a copy that is cut short where the first quotation ends. | research/can-these-all-be-true/sources/openalex-casscells-1978.json | The kept copy breaks off at "…false positive rate of . . .", the end of the second quotation on the page (the question), not the first - CONFIRMED | Type the probabilities you were given and it tells you, exactly, the lowest and highest that any other probability can be without assuming the events are independent, or proves that the numbers contradict each other by showing bets at your own odds that cannot win. | public/tools/can-these-all-be-true/engine.mjs | engine header; I ran the examples. - CONFIRMED | Every answer is checked by a second program before it is shown. | public/tools/can-these-all-be-true/app.mjs | app.mjs:334 withholds the answer unless `proof.ok === true` (but see F6). - CONFIRMED | George Boole asked what a handful of probabilities must satisfy “in order that they may represent a possible experience”, and what limits they then set on anything else. | https://www.gutenberg.org/files/15114/15114-pdf.pdf (kept: research/can-these-all-be-true/sources/boole-laws-of-thought-pg15114.txt) | Boole 1854, ch. XVIII §1 (kept copy, exact words). - CONFIRMED | Boole wrote of the conditions that data must meet “in order that they may represent a possible experience” | https://www.gutenberg.org/files/15114/15114-pdf.pdf (kept: research/can-these-all-be-true/sources/boole-laws-of-thought-pg15114.txt) | Boole 1854, ch. XVIII §1 (kept copy, exact words). - CONFIRMED | and of “the corresponding limits of the final solutions”. | https://www.gutenberg.org/files/15114/15114-pdf.pdf (kept: research/can-these-all-be-true/sources/boole-laws-of-thought-pg15114.txt) | Boole 1854, ch. XVIII §1 (kept copy, exact words). - CONFIRMED | It never assumes independence, and a second program with no solver in it proves each answer before the page shows it. | public/tools/can-these-all-be-true/checker.mjs | I ran the weekend example, 1/2 to 1; independence is used only when stated. - CONFIRMED | Boole's problem of 1854 as an exact linear program, checked by an independent program. | https://arxiv.org/abs/2605.03556 (kept: research/can-these-all-be-true/sources/union-interval-np-hard-arxiv-abs.txt) | arXiv 2605.03556 abstract (kept). Boole 1854 ch. XVII also says "1st, the conditions … 2ndly, the numerical limits". - CONFIRMED | Both questions have exact answers, and George Boole posed them both in 1854. | https://arxiv.org/abs/2605.03556 (kept: research/can-these-all-be-true/sources/union-interval-np-hard-arxiv-abs.txt) | arXiv 2605.03556 abstract (kept). Boole 1854 ch. XVII also says "1st, the conditions … 2ndly, the numerical limits". - CONFIRMED | the Bell game's three-quarters ceiling, found by trying all sixteen strategies | src/content/strata/a-game-you-shouldnt-win.md | the a-game-you-shouldnt-win dek, "enumerate all sixteen possible strategies … Three-quarters is a hard wall". Its sixteen strategies are the sixteen cases here. - CONFIRMED | four bets on agreement rates that lose in every one of the sixteen. | research/can-these-all-be-true/verify.mjs | all 16 rows are losses (page table, best case -0.2112…). - CONFIRMED | with the test's sensitivity set to 100%, which gives the usual answer | research/bayes-theorem/verify.mjs | `bayes-theorem/index.html` has `data-se="100" data-sp="95">Casscells 1978 (1-in-1000)` and "Casscells 1978 (≈ 2%)". - CONFIRMED | with the sensitivity left out, as the question was posed, which gives a range whose top end is that answer. | research/can-these-all-be-true/verify.mjs | the range comes back 0 to 20/1019 = 1.962…%; I redid 1000/50950 by hand. - CONFIRMED | A printed figure is a claim about an interval. | src/content/strata/the-room-inside-a-number.md | the room page's dek, "a claim about an interval". - CONFIRMED | Read that way, half its last digit either side, a poll whose shares add to 101% stops being a contradiction. | public/tools/can-these-all-be-true/engine.mjs | in the browser the poll shows "DEPENDS ON ROUNDING" (45.5 + 37.5 + 16.5 = 99.5). - CONFIRMED | Both ask whether reported numbers could have come from any world at all, and both answer with something a reader can check by hand. | src/content/strata/the-mean-twenty-people-cannot-make.md | the mean page, steps of 0.05. - CONFIRMED | In 1862 he gave the conditions the name this layer takes its title from | https://jimclifford.ca/philosophical-transactions-ocr-1665-1869/articles/jstor-108830/ (kept: research/can-these-all-be-true/sources/boole-1862-phil-trans-ocr-jimclifford.txt) | "may be termed the conditions of possible experience. When satisfied …" is verbatim in both transcriptions, which also carry "Read June 19, 1862". - CONFIRMED | and said what they tell you: “When satisfied they indicate that the data may have, when not satisfied they indicate that the data cannot have, resulted from actual observation.” | https://archive.org/download/jstor-108830/108830_djvu.txt (kept: research/can-these-all-be-true/sources/boole-1862-jstor-early-journal-content-djvu.txt) | "may be termed the conditions of possible experience. When satisfied …" is verbatim in both transcriptions, which also carry "Read June 19, 1862". - CONFIRMED | for the weekend above, multiplying gives 75%, and the honest answer is anywhere from 50% to 100%. | public/tools/can-these-all-be-true/examples.mjs | 1/2 to 1, and 1 - 1/4 = 3/4. - CONFIRMED | For the assistant's two forecasts: buy the bet on next year at 40%, sell the bet on the year after at 30%, and you are down at least a tenth whatever happens. | research/can-these-all-be-true/verify.mjs | rows -1/20, -11/20, -1/20 at stakes of 1/2, so at least a tenth lost at unit stakes. - CONFIRMED | The checker shares no code with the solver and has no solver in it. | public/tools/can-these-all-be-true/checker.mjs | `checker.mjs` read in full; no import of the tool's code, a loop over all 2^n cases, BigInt fractions. - CONFIRMED | It goes through every case and does nothing there but arithmetic on exact fractions | public/tools/can-these-all-be-true/checker.mjs | `checker.mjs` read in full; no import of the tool's code, a loop over all 2^n cases, BigInt fractions. - CONFIRMED | The record the page saves can be checked the same way without this site. | public/tools/can-these-all-be-true/checker.mjs | the reader check above. - CONFIRMED | One of Boole's conditions is Bell's inequality. | https://api.openalex.org/works/doi:10.1093/bjps/45.1.95 (kept: research/can-these-all-be-true/sources/openalex-abstracts.txt) | Pitowsky's abstract (OpenAlex, kept): "Some of Boole's conditions … including Bell inequalities". - CONFIRMED | Each rate was measured on its own share of 245 trials | https://arxiv.org/pdf/1508.05949 (kept: research/can-these-all-be-true/sources/hensen-2015-arxiv-1508.05949.txt) | Figure 4a, kept image viewed: +46/-7, +63/-16, +46/-16, +10/-41. These sum to 245, give k = 196 and give S = 2.4224; "+" means the answers agree. - CONFIRMED | whether chance alone could have produced the rates is a statistical question, which the experimenters answered separately. | https://arxiv.org/pdf/1508.05949 (kept: research/can-these-all-be-true/sources/hensen-2015-arxiv-1508.05949.txt) | "p = 0.039 that a local-realist model …". - CONFIRMED | The page also works Boole's own first example (thunder and hail, with the two limits he printed) | https://www.gutenberg.org/files/15114/15114-pdf.pdf (kept: research/can-these-all-be-true/sources/boole-laws-of-thought-pg15114.txt) | "Inferior limit = q. Superior limit = q + 1 − p." The run gives 1/10 to 4/5. - CONFIRMED | his 1862 example of medical statistics that contradict themselves | https://jimclifford.ca/philosophical-transactions-ocr-1665-1869/articles/jstor-108830/ (kept: research/can-these-all-be-true/sources/boole-1862-phil-trans-ocr-jimclifford.txt) | 2/5, 2/3, 4/5; "The above formulae would show that the evidence was contradictory." The run gives a loss of 1/45. - CONFIRMED | a problem of Lewis Carroll's whose printed answer is ten per cent | https://www.gutenberg.org/files/29042/29042-h/29042-h.htm (kept: research/can-these-all-be-true/sources/carroll-tangled-tale-pg29042.txt) | "Answer.—Ten." - CONFIRMED | the test question put to doctors and medical students at four Harvard teaching hospitals in a study published in 1978, which as posed has no single answer. | https://api.openalex.org/works/doi:10.1056/NEJM197811022991808 (kept: research/can-these-all-be-true/sources/openalex-abstracts.txt) | 20 house officers, 20 fourth-year students and 20 attending physicians at four Harvard Medical School teaching hospitals; NEJM 1978. - CONFIRMED | PrSAT, by Koissi Adjorlolo and Branden Fitelson, decides whether a set of them can be satisfied and returns a model when it can | https://fitelson.org/PrSAT/ (kept: research/can-these-all-be-true/sources/prsat-index.txt) | `prsat-index.txt`, the authors and "returns a probability model witnessing satisfiability". - CONFIRMED | it does not guess what a line meant, it does not take probability zero for never, it stops at sixteen events, and it says so when a condition that can never happen is doing the work. | research/can-these-all-be-true/verify.mjs | 17 events are refused ("stops at 16 events (65536 cases)"); probability-zero conditions get their notes. - CONFIRMED | The page’s content policy lets it contact only artwaste.land | public/tools/can-these-all-be-true/index.html | `connect-src 'self'`. - CONFIRMED | the site’s usual third-party analytics script is not added to this page | worker.js | worker.js:1065 and 2701. - CONFIRMED | nothing you type is sent anywhere. | public/tools/can-these-all-be-true/worker.mjs | network log. - CONFIRMED | Write them down here. The Artificial Wasteland’s checker works out, exactly, whether any world could make them all true at once, and what they force about anything else you ask. | public/tools/can-these-all-be-true/index.html | Tool lede (the sentences other than F4): world and range, with a world at each end for closed ends - CONFIRMED | independence never assumed | public/tools/can-these-all-be-true/engine.mjs | Chip: independence is used only when a line states it (expandIndependence) - CONFIRMED | what you type stays in your browser | public/tools/can-these-all-be-true/index.html | Chip: the agent’s network log of the page while typing showed no request - CONFIRMED | Assume independence. A calculator that multiplies two chances adds a statement you never made | research/can-these-all-be-true/sources/calculator-net-probability.txt | the calculator quotes are verbatim in the kept calculator.net and Omni copies - CONFIRMED | Guess what you meant. It shows how it read each line, and says each conditional back in words. | public/tools/can-these-all-be-true/index.html | every conditional is said back in words - CONFIRMED | Tell you the statements are true. “Can all be true” means some world fits them. | public/tools/can-these-all-be-true/index.html | the will-not-do item holds - CONFIRMED | Blur “more than” into “at least”. | public/tools/can-these-all-be-true/engine.mjs | an open end comes back marked open, and the closure note appears - CONFIRMED | Take probability zero for never. | public/tools/can-these-all-be-true/engine.mjs | in the browser, "If B has any chance at all: exactly 0 … full answer … 0 to 1" - CONFIRMED | Go past sixteen events, or run on without end. Sixteen events are 65,536 cases, which it goes through. | research/can-these-all-be-true/sources/union-interval-np-hard-arxiv-abs.txt | sixteen events, 65,536 cases, NP-hard (arXiv 2605.03556), the fixed work allowance - CONFIRMED | The problem is George Boole’s, set out in 1854 and 1862. Solving it as a linear program is Theodore Hailperin’s, from 1965 … Other software does parts of this, one of them in a browser. | research/can-these-all-be-true/sources/prsat-index.txt | Credit note: Hailperin 1965 is in the arXiv abstract; PrSAT "requires only a web browser" - CONFIRMED | Whoever opens it gets nothing from us but this page: their own browser reads the statements and works the answer out again, proof included. | public/tools/can-these-all-be-true/engine.mjs | the take-away text: linkFor puts the statements after the # - CONFIRMED | The checker is one file, imports nothing, and has no solver in it: it only evaluates what a record claims, in exact fractions, across every case. | public/tools/can-these-all-be-true/checker.mjs | The checker blurb holds - CONFIRMED | Verdict text for a Dutch book: at those prices somebody can bet against you so that you lose in every case in which any bet is on. | public/tools/can-these-all-be-true/checker.mjs | matches the checker's casesFor tables - CONFIRMED | Verdict text for bets that never win: together they can never pay you anything. | public/tools/can-these-all-be-true/checker.mjs | matches the checker's casesFor tables - CONFIRMED | Verdict text for statements that can all be true: at least one way of sharing out the chances fits every statement. | public/tools/can-these-all-be-true/checker.mjs | matches the checker's casesFor tables - CONFIRMED | Verdict text for a conditional answered below the first level: the levels and the statements that still hold the price. | public/tools/can-these-all-be-true/checker.mjs | matches the checker's casesFor tables - CONFIRMED | A decimal comma (0,3) is refused, because it could be read two ways | public/tools/can-these-all-be-true/rational.mjs | "0,3" and "2,5%" are refused with the stated reason - CONFIRMED | a probability, as a decimal, a percentage (30%), a fraction (3/10) or “1 in 8,543”. | public/tools/can-these-all-be-true/rational.mjs | "1 in 8,543" reads as 1/8543 - CONFIRMED | These are rounded figures (“= 70%” then means anything that rounds to 70%: 69.5% to 70.5%) | public/tools/can-these-all-be-true/engine.mjs | Rounded 70% reads as 139/200 to 141/200 - CONFIRMED | The example notes: the weekend forecast | public/tools/can-these-all-be-true/examples.mjs | checked against the run - CONFIRMED | The example notes: Boole’s thunder and hail | public/tools/can-these-all-be-true/examples.mjs | checked against the run - CONFIRMED | The example notes: two things more likely than not | public/tools/can-these-all-be-true/examples.mjs | checked against the run - CONFIRMED | The example notes: a forecast against itself | public/tools/can-these-all-be-true/examples.mjs | checked against the run - CONFIRMED | The example notes: a poll that adds to 101% | public/tools/can-these-all-be-true/examples.mjs | checked against the run - CONFIRMED | The example notes: three majorities | public/tools/can-these-all-be-true/examples.mjs | 2/3 by hand - CONFIRMED | The example notes: given something impossible | public/tools/can-these-all-be-true/examples.mjs | by hand - CONFIRMED | The example notes: Carroll’s pensioners | research/can-these-all-be-true/sources/carroll-tangled-tale-pg29042-plain.txt | Janet’s case and the 1885 date - CONFIRMED | The example notes: the Harvard question | research/can-these-all-be-true/sources/openalex-casscells-1978.json | 5%, 1/1000, no sensitivity given - CONFIRMED | The example notes: a conjunction ranked too high | research/can-these-all-be-true/sources/openalex-tversky-kahneman-1983.json | title kept - CONFIRMED | The example notes: Boole’s three symptoms | research/can-these-all-be-true/sources/boole-1862-phil-trans-ocr-jimclifford.txt | both 1862 transcriptions - CONFIRMED | The example notes: a Bell experiment | research/can-these-all-be-true/sources/hensen-2015-arxiv-1508.05949.txt | counts as read from Figure 4a - CONFIRMED | Exact bounds on any probability from what you were told, with no independence assumed, or the bets that prove the numbers cannot all be true. | src/data/tools.mjs | Catalogue oneLine - CONFIRMED | The link carries the statements in the address after the #, one per semicolon … opens on the answer, worked out and proved again in the reader’s own browser. Add r=1& before s= for rounded figures. | src/data/tools.mjs | Catalogue faceNote: the example link runs to 1/2..1; r=1& handled - CONFIRMED | Can These All Be True?: two statements typed in, rain on Saturday at 50% and rain on Sunday at 50% … anywhere from 1/2 to 1, drawn as a bar | src/data/tools.mjs | Catalogue alt: screenshot viewed - CONFIRMED | The callable face’s input schema | mcp.js | MCP input schema - CONFIRMED | The callable face’s `declines` | mcp.js | holds - CONFIRMED | The callable face’s `privacy` | mcp.js | mcp.js records only method, tool and outcome; Cloudflare-side logs not visible - CONFIRMED | The callable face’s `how_to_check_a_range` | public/tools/can-these-all-be-true/checker.mjs | matches checker.mjs:235-238 term for term - CONFIRMED | The callable face’s `how_it_was_read` | mcp.js | holds - CONFIRMED | Up to 10 events per call; the page takes 16. | mcp.js | 11 are refused - CONFIRMED | The callable face’s `meaning` for bets that never win | mcp.js | holds - CONFIRMED | The callable face’s `checked` | mcp.js | holds - CONFIRMED | Narration: Somebody hands you a few probabilities, and each one sounds reasonable on its own. | public/tv/lib/narration/batch-120.js | supported by the layer, lines 28-69 - CONFIRMED | Narration: The first question is not whether they are true. It is whether they could all be true at once. | public/tv/lib/narration/batch-120.js | supported by the layer - CONFIRMED | Narration: George Boole asked exactly that, and asked what such numbers then settle about everything else. | public/tv/lib/narration/batch-120.js | supported by the layer - CONFIRMED | Narration: Multiply two chances together and you have quietly added a claim that nobody made. | public/tv/lib/narration/batch-120.js | supported by the layer - CONFIRMED | Narration: When the numbers cannot all hold, the proof is a set of bets, at your own prices, that cannot win. | public/tv/lib/narration/batch-120.js | supported by the layer - CONFIRMED | Narration: Put the agreement rates of a famous physics experiment through the same test, and no single world of ready answers fits them. | public/tv/lib/narration/batch-120.js | supported by the layer; "famous" is not in the layer, and harmless - CONFIRMED | Narration: Nothing is shown until a second program, with no solver in it, has checked the first. | public/tv/lib/narration/batch-120.js | supported by the layer (but see F6) - OBSERVED | it was found by an independent screen of this idea after a first search had missed it. | research/can-these-all-be-true/prior-art-2026-11-15.md | prior-art-2026-11-15.md:57, "The first search's \"none found\" was wrong about (a)"; screen/REPORT.md:22 lists PrSAT - WRONG | exact fractions, nothing rounded … every figure in an answer is an exact fraction; decimals shown beside it are cut short and marked with an ellipsis | public/tools/can-these-all-be-true/index.html | F1: "P(A) = 1/1234567891" was answered "Exactly 0.000000… (0.000000081000…%)", with no fraction anywhere a reader could see; tidy() in app.mjs put a cut decimal where the fraction belonged - WRONG | Round, or call a rounding slip a contradiction … When figures conflict as written but would fit as rounded figures, the page says that first. | public/tools/can-these-all-be-true/index.html | F2: "P(A) = 0.5 / P(A) > 0.5" came back plain INCOHERENT although the rounded reading fits; the never-wins branch of drawIncoherent returned before the rounding test - WRONG | Staked as below (a positive stake buys the bet, a negative one sells it; the sizes of the stakes sum to 1), the total is a loss of at least `guaranteed_loss_per_unit_staked` in every case where any of the bets is on. | mcp.js | F3: for "P(A) <= 0.3; P(A) >= 0.5" buying both, as the sentence says, wins 0.6 if A happens; the real book sells the "<=" bet (checker.mjs negates its gain) - WRONG | a set of bets at your own odds that never wins, and loses whenever any of them is in play. | public/tools/can-these-all-be-true/index.html | F4: bets of the second kind (a strict statement that cannot hold) never lose: for "P(A) = 0.5 / P(A) > 0.5" the page’s own table reads 0, 0 - WRONG | The reason is beside each line above. | public/tools/can-these-all-be-true/app.mjs | F5: for 63 statements every line was shown as read and no reason appeared; the engine’s error had line 0 - MINOR | cannot win and loses wherever any of them is on (the tool page’s structured data) | public/tools/can-these-all-be-true/index.html | F4, the same sentence in the JSON-LD - MINOR | it prints a Dutch book, a set of bets at your own stated odds that loses wherever any of them is on | src/data/tools.mjs | F4, the catalogue description - MINOR | or a Dutch book, the bets at your own odds that lose wherever any of them is on. | src/content/strata/can-these-all-be-true.md | F4, the layer’s dek - MINOR | If the statements cannot all be true it prints a Dutch book. Each statement is a price you have called fair for a bet, and at those prices there is a set of bets that loses wherever any of them is on. | src/content/strata/can-these-all-be-true.md | F4, the layer’s body - MINOR | with the proof (a Dutch book: stakes on the person's own statements that lose in every case where any of the bets is on) | mcp.js | F4, the callable face’s description; its own example "P(A and B) > P(A)" returns "bets that never win" - MINOR | Every statement is needed: drop any one … and the rest of them can all be true together | public/tools/can-these-all-be-true/app.mjs | F6: shown next to "Proved." but not checked by the second program (true on 400 random problems) - MINOR | the nearest thing: a world where the strict statement holds only with equality | public/tools/can-these-all-be-true/app.mjs | F6: the checker accepted a nearest world whose shares summed to 3/4 - MINOR | the most that any stakes on these statements can guarantee | public/tools/can-these-all-be-true/app.mjs | F6: the checker accepted a book whose loss had been halved, which the page would then call the most - MINOR | a_world_that_fits (the callable face’s reply) | mcp.js | F7: a 34-case world came back as 32 entries summing to 61/64, with no note - MINOR | What the site learns when you use this page: its server sends the page’s files and counts a page view by the address up to, not including, the #. | worker.js | F8: the server also records the hour, a source label, the previous page of this site, and separate daily tallies of country and device class; the inline pulse posts the path - MINOR | the probability of anything you ask about has a least and a greatest value over all of them | src/content/strata/can-these-all-be-true.md | F9: the tool’s own "two things more likely than not" has no smallest value, only a limit; the dek’s "with a world at each end" likewise - MINOR | What this one adds is the range of anything asked, a proof for every answer that a second program checks, and the stricter reading of conditional statements | research/can-these-all-be-true/screen/REPORT.md | F10: Pinference (CRAN) computes Hailperin’s bounds; true in a browser, and the layer did not say so - MINOR | Let a condition that cannot happen hide a contradiction. | public/tools/can-these-all-be-true/index.html | F11: the condition meant is one of probability zero, which the same page insists is not "never" - MINOR | it says so when a condition that can never happen is doing the work. | src/content/strata/can-these-all-be-true.md | F11, the layer - MINOR | Narration: The answer is rarely one number. It is a range, with a possible world standing at each end. | public/tv/lib/narration/batch-120.js | F12: the layer makes no claim about how often ## Findings - [fixed] resting on a theorem they cite from earlier work (their Theorem 3): Section 3 now credits Algorithm 1 as the paper does, to Biazzo, Gilio and Sanfilippo (2008), and Theorem 3 to Gilio (1993), and says those theorems are for statements of exact value; the source-record entry is rewritten. - [fixed] This is the procedure Gilio and Sanfilippo set out as their Algorithm 1: The same sentence now says they credit it to that earlier work. - [fixed] What this instrument adds is therefore narrower than it first seemed, and is these things:…: Now "these things, in a browser", with Pinference and CkC credited with ranges and conditional coherence outside one; the layer says the same (new entries body-pinference, body-ckc, body-adds). - [fixed] Lewis Carroll’s readers found exactly such a proof by hand in 1885: "in an answer printed in A Tangled Tale (1885)". - [fixed] An independent agent, sent to attack the idea before it was built on, did.: "sent to attack the first build". - [fixed] The same paper sets out the way of finding them that the instrument follows.: The page now quotes the paper: the method "is taken from the 'Memoir on the Combination of Testimonies and of Judgments,' already referred to". - [fixed] The same paper shows that computing the optimal interval for the union probability is NP-h…: "which is why" is gone: the union problem is a special case, no method is known that is fast on every problem of this kind, and the stop at sixteen comes from going through all 2^n cases. - [fixed] The statements are coherent exactly when no layer fails.: The page now says the theorems are for statements of exact value, that the test is applied to one-sided and strict ones as well, and that families D, K and L test that extension. - [fixed] The mathematics is old and public, and there is software for parts of it. What was found, …: Section 6 now names NILS, the PSAT solvers of Finger and De Bona, Pronto and diff-SAT as found and not opened. - [fixed] The verifier, verify.mjs, runs 17,664 checks in twelve families: Section 7 names family J and prints its count, written by sync-figures.mjs and compared by the verifier; J now also checks the 85.4% and 14.6% figures and the four Delft counts. - [fixed] Sixteen events, and a fixed allowance of work. … took about twenty seconds on the machine …: Section 9 names sixty statements (and what counts as two), six events declared independent at once, the cap on rounds, the second program's time, and slow devices. - [fixed] (The 1978 abstract was read in a copy that is cut short where the first quotation ends.: "cut short where the quoted question breaks off". - [fixed] exact fractions, nothing rounded … every figure in an answer is an exact fraction; decimal…: An answer is now always exact, as a fraction or a decimal that ends; a fraction too long for the line is printed in full under it, and a cut decimal keeps enough digits never to read as zero. The promise is reworded to match. - [fixed] Round, or call a rounding slip a contradiction … When figures conflict as written but woul…: The rounding test now runs for both kinds of proof, and the rounding verdict lists the slack each figure was given. - [fixed] Staked as below (a positive stake buys the bet, a negative one sells it; the sizes of the …: The callable face now lists each bet with its side, buy or sell, and its size, and says what each side means; its checks recompute the loss from the bets as described, a "<=" statement sold among them. - [fixed] a set of bets at your own odds that never wins, and loses whenever any of them is in play.: "a set of bets at your own odds that can never come out ahead", true of both kinds. - [fixed] The reason is beside each line above.: Too many statements is now a refusal with the limit named, and a reason that belongs to no line is printed in the verdict. - [fixed] cannot win and loses wherever any of them is on (the tool page’s structured data): "can never come out ahead". - [fixed] it prints a Dutch book, a set of bets at your own stated odds that loses wherever any of t…: "a set of bets at your own stated odds that can never come out ahead". - [fixed] or a Dutch book, the bets at your own odds that lose wherever any of them is on.: "or bets at your own odds that can never come out ahead". - [fixed] If the statements cannot all be true it prints a Dutch book. Each statement is a price you…: The layer now says a Dutch book is given when the statements could not hold even with every "more than" read as "at least", and otherwise bets called better than fair that never pay. - [fixed] with the proof (a Dutch book: stakes on the person's own statements that lose in every cas…: "bets on the person's own statements, each marked buy or sell, that can never come out ahead: usually a Dutch book". - [fixed] Every statement is needed: drop any one … and the rest of them can all be true together: The record now carries a world for each statement of a conflicting set showing that it is needed, and the checker verifies each. - [fixed] the nearest thing: a world where the strict statement holds only with equality: The record carries that world and the checker verifies that the staked statements hold in it, read as "at least". - [fixed] the most that any stakes on these statements can guarantee: The record carries a world in which no staked statement misses by more than the loss, with the stakes' sizes summing to 1; the checker verifies it, and the sentence is shown only when it does. - [fixed] a_world_that_fits (the callable face’s reply): A world with more than 32 cases is cut with a note giving the count; weights are listed per statement. - [fixed] What the site learns when you use this page: its server sends the page’s files and counts …: The line names the hour, the arrival label, the daily country and device tallies and the confirming ping, each in the source record (t-pulse, t-people, t-arrival, t-ping). - [fixed] the probability of anything you ask about has a least and a greatest value over all of the…: "can range only between two limits", and "a world at each end that is reached". - [fixed] What this one adds is the range of anything asked, a proof for every answer that a second …: The layer names Pinference and CkC and says "in a browser". - [fixed] Let a condition that cannot happen hide a contradiction.: "a condition of probability zero"; the method page's section 3 and the example are renamed to match. - [fixed] it says so when a condition that can never happen is doing the work.: "a condition of probability zero". - [fixed] Narration: The answer is rarely one number. It is a range, with a possible world standing …: "The answer may be one number, or a range"; and the experiment is named by its year, not called famous. - [fixed] The page told readers to run the checker on "record.json" and saved "can-these-all-be-true.record.json"; it now names the file it saves.