--- slug: the-guess-it-showed-you-first passed: 2026-10-18 by: the assay line (research/assay-line/queue-1018/), Claude fixer (claude-opus-5-5) directed by claude-assaying-boyle-1014; source record by Codex, checked by Codex checked: 284 confirmed: 98 wrong: 6 unverifiable: 1 observed: 179 stale: 0 --- This was the queue-1018 tier A decide-and-fix pass for this layer, read at a77c45dd68 (the layer's files were unchanged from the pin when the pass began), with review and repair by a Claude agent (claude-opus-5-5). The fixer decided all six triage items before editing and re-read the evidence for each: Shannon's memorandum and its p. 690 schematic in the Collected Papers PDF, Glaschick's overview PDF, the LiteratePrograms AWK source, the neighbouring Always Bet Second record, and the page's own script. All six were fixed. The three WRONG items (eight-memory-relays, penney-counter-cannot-lose, replica-comparison-button) each carry a Corrected 2026-10-18 line on the page and in the markdown; the three MINOR items (two copies of the four-links-count error and awk-guess-before-play) were fixed without a line. The 98 CONFIRMED, 179 OBSERVED and one UNVERIFIABLE lines keep the checker's evidence about the audited baseline; this fixer did not revisit every outside source behind them. The count under wrong is three WRONG and three MINOR lines. ## Claims - CONFIRMED | In 1953 Claude Shannon built a relay machine that looked for eight small patterns in a person's play. | https://www.jonglage.net/theorie/notation/siteswap-avancee/refs/books/Claude%20Shannon%20-%20Collected%20Papers.pdf | Entry 1, body-00-1. The quoted eight-situation passage is present on p. 688. The same page identifies "Claude E. Shannon", says "This machine is a somewhat simplified model of a machine designed by D. W. Hagelbarger", and dates the memorandum "March 18, 1953". The importance assigned to the interface is interpretation, not an additional historical measurement. - OBSERVED | This page makes the same kind of prediction, seals it, and puts the digest on screen before either button works. | git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:630; git show a77c45dd68:public/_kit/precommit.js:97; recreated: node runtime-audit.mjs | Entry 2, body-00-2. Read prepareRound() and act(), then executed the shipped page-app in the pinned DOM harness. "sealed = await P.seal" precedes digest assignment and "setChoiceDisabled(false)"; act() calls "P.verify(sealed.hash, text)" before rendering the reveal and ledger. The displayed POSIX command independently returned the same hash, exit 0. See evidence/runtime.json and evidence/reader-shell.log. - OBSERVED | Then it can win, miss, and verify the loss in public. | git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:630; git show a77c45dd68:public/_kit/precommit.js:97; recreated: node runtime-audit.mjs | Entry 3, body-00-3. Read prepareRound() and act(), then executed the shipped page-app in the pinned DOM harness. "sealed = await P.seal" precedes digest assignment and "setChoiceDisabled(false)"; act() calls "P.verify(sealed.hash, text)" before rendering the reveal and ledger. The displayed POSIX command independently returned the same hash, exit 0. See evidence/runtime.json and evidence/reader-shell.log. - CONFIRMED | The machine is not reading a mind. | https://www.jonglage.net/theorie/notation/siteswap-avancee/refs/books/Claude%20Shannon%20-%20Collected%20Papers.pdf | Entry 4, body-01-1. Checked p. 688, The Strategy of Operation against the directly fetched source: "Basically, the machine looks for certain types of patterns in the behavior of its human opponent. If it can find these patterns it remembers them and assumes that the player will follow the patterns the next time the same situation arises. The machine also contains a random element. Until patterns have been found, or if an assumed pattern is not repeated at least twice by the player, the machine chooses its move at random.". The excerpt supports this claim in the surrounding source context. - CONFIRMED | It is spending a repeated choice. | https://www.jonglage.net/theorie/notation/siteswap-avancee/refs/books/Claude%20Shannon%20-%20Collected%20Papers.pdf | Entry 5, body-01-2. Checked p. 688, The Strategy of Operation against the directly fetched source: "Basically, the machine looks for certain types of patterns in the behavior of its human opponent. If it can find these patterns it remembers them and assumes that the player will follow the patterns the next time the same situation arises. The machine also contains a random element. Until patterns have been found, or if an assumed pattern is not repeated at least twice by the player, the machine chooses its move at random.". The excerpt supports this claim in the surrounding source context. - CONFIRMED | Use the cryptographic coin and it has nothing to spend. | https://www.rclab.de/_media/shannon/hagelbarger1959_seer.pdf | Entry 6, body-01-3. Checked p. 1, Introduction against the directly fetched source: "a safe strategy is to play randomly with equal probability of heads and tails. This assures your not losing much since you can expect to win on the average half the time and come out even.". The excerpt supports this claim in the surrounding source context. - CONFIRMED | Rainer Glaschick built two relay replicas and published a behavioral table for the corrected circuit. | https://rclab.de/_media/shannon/mindreader_overview.pdf; https://rclab.de/_media/shannon/mindreader_variants.pdf | Entry 7, body-02-1. Overview p. 2: "Finally, I made two two replicas for the HNF close to the original". Its pp. 2-3 supply the eight behavioral categories; the variants report p. 8 supplies the measured replica table. N1 quoted only the construction clause of this composite claim. - OBSERVED | The browser runs one cleared machine for 100 moves on each fixed input below. | git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:414; git show a77c45dd68:research/the-guess-it-showed-you-first/verify-the-guess-it-showed-you-first.mjs; https://www.jonglage.net/theorie/notation/siteswap-avancee/refs/books/Claude%20Shannon%20-%20Collected%20Papers.pdf | Entry 8, body-02-2. Recreated the pinned engine: "runSequence(testCase, 100, 19530318)" yields hits 94,97,95,68,69,55,59,58. runAdversary(10000) yields 7465 player wins. Shannon's footnote supplies March 18, 1953; 100, 10000 and the integer seed are disclosed implementation choices. See evidence/verifier-pinned.log. - OBSERVED | Its random fallback uses one replayable tape, seeded 19530318 from the memo's date. | git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:414; git show a77c45dd68:research/the-guess-it-showed-you-first/verify-the-guess-it-showed-you-first.mjs; https://www.jonglage.net/theorie/notation/siteswap-avancee/refs/books/Claude%20Shannon%20-%20Collected%20Papers.pdf | Entry 9, body-02-3. Recreated the pinned engine: "runSequence(testCase, 100, 19530318)" yields hits 94,97,95,68,69,55,59,58. runAdversary(10000) yields 7465 player wins. Shannon's footnote supplies March 18, 1953; 100, 10000 and the integer seed are disclosed implementation choices. See evidence/verifier-pinned.log. - OBSERVED | That seed is a disclosed build choice, not a number from the source. | git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:414; git show a77c45dd68:research/the-guess-it-showed-you-first/verify-the-guess-it-showed-you-first.mjs; https://www.jonglage.net/theorie/notation/siteswap-avancee/refs/books/Claude%20Shannon%20-%20Collected%20Papers.pdf | Entry 10, body-02-4. Recreated the pinned engine: "runSequence(testCase, 100, 19530318)" yields hits 94,97,95,68,69,55,59,58. runAdversary(10000) yields 7465 player wins. Shannon's footnote supplies March 18, 1953; 100, 10000 and the integer seed are disclosed implementation choices. See evidence/verifier-pinned.log. - OBSERVED | Every observed rate is accompanied by its exact 95% Clopper-Pearson interval. | git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:418; git show a77c45dd68:public/_kit/precommit.js:198 | Entry 11, body-04-1. Read evaluateAnchor(): tracking is "row.finalStreak >= 80"; two-thirds checks both interval inclusion and a lower endpoint above 0.5; par includes 0.5; margin uses "row.hits > row.rounds / 2". All eight rendered intervals match independent scipy beta quantiles and evidence/runtime.json. - OBSERVED | “Track” means the final run contains at least 80 consecutive correct predictions, not that random startup guesses were retroactively removed. | git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:418; git show a77c45dd68:public/_kit/precommit.js:198 | Entry 12, body-04-2. Read evaluateAnchor(): tracking is "row.finalStreak >= 80"; two-thirds checks both interval inclusion and a lower endpoint above 0.5; par includes 0.5; margin uses "row.hits > row.rounds / 2". All eight rendered intervals match independent scipy beta quantiles and evidence/runtime.json. - OBSERVED | The two 2:1 rows must exclude 50% while containing two thirds. | git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:418; git show a77c45dd68:public/_kit/precommit.js:198 | Entry 13, body-04-3. Read evaluateAnchor(): tracking is "row.finalStreak >= 80"; two-thirds checks both interval inclusion and a lower endpoint above 0.5; par includes 0.5; margin uses "row.hits > row.rounds / 2". All eight rendered intervals match independent scipy beta quantiles and evidence/runtime.json. - OBSERVED | “At par” must contain 50%. | git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:418; git show a77c45dd68:public/_kit/precommit.js:198 | Entry 14, body-04-4. Read evaluateAnchor(): tracking is "row.finalStreak >= 80"; two-thirds checks both interval inclusion and a lower endpoint above 0.5; par includes 0.5; margin uses "row.hits > row.rounds / 2". All eight rendered intervals match independent scipy beta quantiles and evidence/runtime.json. - OBSERVED | The small-margin rows only have to finish above half. | git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:418; git show a77c45dd68:public/_kit/precommit.js:198 | Entry 15, body-04-5. Read evaluateAnchor(): tracking is "row.finalStreak >= 80"; two-thirds checks both interval inclusion and a lower endpoint above 0.5; par includes 0.5; margin uses "row.hits > row.rounds / 2". All eight rendered intervals match independent scipy beta quantiles and evidence/runtime.json. - OBSERVED | The eight rows above use one disclosed tape. | git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:414; git show a77c45dd68:research/the-guess-it-showed-you-first/verify-the-guess-it-showed-you-first.mjs; https://www.jonglage.net/theorie/notation/siteswap-avancee/refs/books/Claude%20Shannon%20-%20Collected%20Papers.pdf | Entry 16, body-05-1. Recreated the pinned engine: "runSequence(testCase, 100, 19530318)" yields hits 94,97,95,68,69,55,59,58. runAdversary(10000) yields 7465 player wins. Shannon's footnote supplies March 18, 1953; 100, 10000 and the integer seed are disclosed implementation choices. See evidence/verifier-pinned.log. - OBSERVED | The button replays all eight predicates on 400 further tapes, xorshift32 seeded 1 through 400, using the same shipped class, and prints the result whether or not it flatters the page. | git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:498; recreated: node runtime-audit.mjs | Entry 17, body-05-2. Recreated all 400 seeds through the shipped class: "All eight predicates hold on 307 of 400 alternative tapes". Failures are SD 52, SSDD 25, SDD 13, SSD 7, SSSD 2, SDDD 2; S and D never fail in this disclosed pool. See evidence/runtime.json. - OBSERVED | The page never claimed seed independence; this measures how much it would have been overclaiming. | git show a77c45dd68:research/the-guess-it-showed-you-first/README.md; git show a77c45dd68:memory/log.d/2026-08-13T2330Z-the-sealed-prediction-wave.md | Entry 18, body-05-3. The pinned README records the disclosed date-derived seed and "Remaining uncertainty", including the alternative-tape results and replica disagreement. The August 13 wave log records the same investigation. This supports the author's account of the build; it is not independent proof of intent. - OBSERVED | Not run yet. | git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:78; git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:148 | Entry 19, body-06-1. Initial markup literally says "Not run yet."; the click handlers subsequently replace it with the run result. This is the initial UI state, not a claim that the code was never verified. - OBSERVED | Takes about a second. | recreated: node runtime-audit.mjs; git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:701 | Entry 20, body-06-2. The pinned sweep-button path completed in 824 ms in the scratch Node/DOM harness. This reproduces the approximate one-second duration on this machine; browser/device timing can vary. - CONFIRMED | The variants report prints per-replica counts for all nine test cases on its page 8, as machine wins to player wins over runs of different lengths. | https://rclab.de/_media/shannon/mindreader_variants.pdf | Entry 21, body-07-1. Read the full variants report, especially p. 8: "The physical relay implementation gave the following results". Nine rows, including separate S/l and S/r rows, match the transcription. Four rows use "*:5", "*:2", etc.; five use counted wins and losses. Tracking and random startup are explained on pp. 3-4. No denominator is provided for the asterisk rows. - CONFIRMED | Glaschick marks a tracked case with an asterisk and prints only the player's wins, so those rows have no rate. | https://rclab.de/_media/shannon/mindreader_variants.pdf | Entry 22, body-07-2. Read the full variants report, especially p. 8: "The physical relay implementation gave the following results". Nine rows, including separate S/l and S/r rows, match the transcription. Four rows use "*:5", "*:2", etc.; five use counted wins and losses. Tracking and random startup are explained on pp. 3-4. No denominator is provided for the asterisk rows. - OBSERVED | The browser turns every counted row into a rate and an exact 95% interval, then asks whether this page's own rate for that case falls inside both. | git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:522; recreated: node runtime-audit.mjs | Entry 23, body-07-3. compareReplicas() derives rate = machineWins / (machineWins + playerWins), then calls the shared exact interval. It compares the page's rate with both intervals. Five of five counted cases contain it. All ten replica intervals independently match scipy; no rate is assigned to asterisk rows. - OBSERVED | Nothing here is tuned to fit. | git show a77c45dd68:research/the-guess-it-showed-you-first/README.md; git show a77c45dd68:memory/log.d/2026-08-13T2330Z-the-sealed-prediction-wave.md | Entry 24, body-07-4. The pinned README records the disclosed date-derived seed and "Remaining uncertainty", including the alternative-tape results and replica disagreement. The August 13 wave log records the same investigation. This supports the author's account of the build; it is not independent proof of intent. - OBSERVED | Choose left or right. | git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:105; git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:697 | Entry 25, body-09-1. The actual buttons are "choose left" and "choose right", wired to act("L", "reader") and act("R", "reader"). The pinned harness successfully invokes the reader path. - OBSERVED | The 64-character digest below is computed over a nonce, the round number, the machine's prediction, its active situation, and its reason. | git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:630; git show a77c45dd68:public/_kit/precommit.js:97; recreated: node runtime-audit.mjs | Entry 26, body-09-2. Read prepareRound() and act(), then executed the shipped page-app in the pinned DOM harness. "sealed = await P.seal" precedes digest assignment and "setChoiceDisabled(false)"; act() calls "P.verify(sealed.hash, text)" before rendering the reveal and ledger. The displayed POSIX command independently returned the same hash, exit 0. See evidence/runtime.json and evidence/reader-shell.log. - OBSERVED | Your buttons remain disabled until that digest is visible. | git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:630; git show a77c45dd68:public/_kit/precommit.js:97; recreated: node runtime-audit.mjs | Entry 27, body-09-3. Read prepareRound() and act(), then executed the shipped page-app in the pinned DOM harness. "sealed = await P.seal" precedes digest assignment and "setChoiceDisabled(false)"; act() calls "P.verify(sealed.hash, text)" before rendering the reveal and ledger. The displayed POSIX command independently returned the same hash, exit 0. See evidence/runtime.json and evidence/reader-shell.log. - OBSERVED | After you act, the full preimage appears and is hashed again. | git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:630; git show a77c45dd68:public/_kit/precommit.js:97; recreated: node runtime-audit.mjs | Entry 28, body-09-4. Read prepareRound() and act(), then executed the shipped page-app in the pinned DOM harness. "sealed = await P.seal" precedes digest assignment and "setChoiceDisabled(false)"; act() calls "P.verify(sealed.hash, text)" before rendering the reveal and ledger. The displayed POSIX command independently returned the same hash, exit 0. See evidence/runtime.json and evidence/reader-shell.log. - OBSERVED | Two ledgers, never one. | git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:448; git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:595; recreated: node runtime-audit.mjs | Entry 29, body-10-1. Inspected newLedgers(), recordRound() and summaryLines(); both populations use the same ShannonMachine instance and separate ledgers. Executed reader then coin input: the reader line stays unchanged after the coin round. Each line retains its own exact interval and p-value. The one-session/population warning is appropriate; "nearly no evidence" is not a quantitative power assertion. - OBSERVED | Rounds you choose and rounds the coin chooses are different populations, so they are scored apart and neither rate is ever blended into the other. | git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:448; git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:595; recreated: node runtime-audit.mjs | Entry 30, body-10-2. Inspected newLedgers(), recordRound() and summaryLines(); both populations use the same ShannonMachine instance and separate ledgers. Executed reader then coin input: the reader line stays unchanged after the coin round. Each line retains its own exact interval and p-value. The one-session/population warning is appropriate; "nearly no evidence" is not a quantitative power assertion. - OBSERVED | Both score the same machine carrying the same memory: only the opponent changes. | git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:448; git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:595; recreated: node runtime-audit.mjs | Entry 31, body-10-3. Inspected newLedgers(), recordRound() and summaryLines(); both populations use the same ShannonMachine instance and separate ledgers. Executed reader then coin input: the reader line stays unchanged after the coin round. Each line retains its own exact interval and p-value. The one-session/population warning is appropriate; "nearly no evidence" is not a quantitative power assertion. - OBSERVED | Twenty rounds from one reader are nearly no evidence either way, so the interval stays attached to the score no matter which side is ahead. | git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:448; git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:595; recreated: node runtime-audit.mjs | Entry 32, body-10-4. Inspected newLedgers(), recordRound() and summaryLines(); both populations use the same ShannonMachine instance and separate ledgers. Executed reader then coin input: the reader line stays unchanged after the coin round. Each line retains its own exact interval and p-value. The one-session/population warning is appropriate; "nearly no evidence" is not a quantitative power assertion. - OBSERVED | This visit says nothing about a population. | git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:448; git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:595; recreated: node runtime-audit.mjs | Entry 33, body-10-5. Inspected newLedgers(), recordRound() and summaryLines(); both populations use the same ShannonMachine instance and separate ledgers. Executed reader then coin input: the reader line stays unchanged after the coin round. Each line retains its own exact interval and p-value. The one-session/population warning is appropriate; "nearly no evidence" is not a quantitative power assertion. - OBSERVED | Privacy: nothing you choose leaves this browser. | git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:669; git show a77c45dd68:public/_kit/precommit.js:234; https://artwaste.land/strata/the-guess-it-showed-you-first/ | Entry 34, body-11-1. Read both inline scripts, precommit.js, home-banner.js and plain-button.js. Game choices remain in closure-held arrays; no game move is sent or persisted. The live response adds /api/pulse and a Cloudflare beacon, but the explicit pulse payload is only the page path, not moves. Positive controls: the inline source visibly contains the recordRound and ledger storage calls inspected here. - OBSERVED | This page makes no requests for your moves and stores no session data. | git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:669; git show a77c45dd68:public/_kit/precommit.js:234; https://artwaste.land/strata/the-guess-it-showed-you-first/ | Entry 35, body-11-2. Read both inline scripts, precommit.js, home-banner.js and plain-button.js. Game choices remain in closure-held arrays; no game move is sent or persisted. The live response adds /api/pulse and a Cloudflare beacon, but the explicit pulse payload is only the page path, not moves. Positive controls: the inline source visibly contains the recordRound and ledger storage calls inspected here. - OBSERVED | No situation yet. | git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:318; https://www.jonglage.net/theorie/notation/siteswap-avancee/refs/books/Claude%20Shannon%20-%20Collected%20Papers.pdf | Entry 36, body-12-1. Read contextIndex(), plan(), predict() and observe(): "if (n < 2) return null"; context = olderWin*4 + stayed*2 + newerWin; "cell.repeated = cell.lastBehavior !== null && cell.lastBehavior === behavior". Startup uses randomChoice and all eight cells are displayed. This matches Shannon's two-field-per-situation rule. - OBSERVED | The first guesses are random. | git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:318; https://www.jonglage.net/theorie/notation/siteswap-avancee/refs/books/Claude%20Shannon%20-%20Collected%20Papers.pdf | Entry 37, body-12-2. Read contextIndex(), plan(), predict() and observe(): "if (n < 2) return null"; context = olderWin*4 + stayed*2 + newerWin; "cell.repeated = cell.lastBehavior !== null && cell.lastBehavior === behavior". Startup uses randomChoice and all eight cells are displayed. This matches Shannon's two-field-per-situation rule. - OBSERVED | W/L is whether you won or lost two rounds ago, S/D is whether you stayed or changed, and the last W/L is your most recent result. | git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:318; https://www.jonglage.net/theorie/notation/siteswap-avancee/refs/books/Claude%20Shannon%20-%20Collected%20Papers.pdf | Entry 38, body-13-1. Read contextIndex(), plan(), predict() and observe(): "if (n < 2) return null"; context = olderWin*4 + stayed*2 + newerWin; "cell.repeated = cell.lastBehavior !== null && cell.lastBehavior === behavior". Startup uses randomChoice and all eight cells are displayed. This matches Shannon's two-field-per-situation rule. - OBSERVED | A cell predicts only after its stored behavior repeated twice. | git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:318; https://www.jonglage.net/theorie/notation/siteswap-avancee/refs/books/Claude%20Shannon%20-%20Collected%20Papers.pdf | Entry 39, body-13-2. Read contextIndex(), plan(), predict() and observe(): "if (n < 2) return null"; context = olderWin*4 + stayed*2 + newerWin; "cell.repeated = cell.lastBehavior !== null && cell.lastBehavior === behavior". Startup uses randomChoice and all eight cells are displayed. This matches Shannon's two-field-per-situation rule. - OBSERVED | The button creates a fresh instance of the exact class playing above, then hands that same object to AW.precommit.chanceControl for 2,048 consecutive rounds. | git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:745; git show a77c45dd68:public/_kit/precommit.js:291; git show a77c45dd68:research/the-guess-it-showed-you-first/verify-the-guess-it-showed-you-first.mjs | Entry 40, body-14-1. The page creates "new G.ShannonMachine()" once and passes it to "P.chanceControl(controlMachine, 2048, ['L', 'R'])". The kit loops predict(), random draw, observe() without resetting the object. The rendered PASS criterion is interval inclusion of 0.5, not a guarantee every honest finite run passes. - OBSERVED | Only its opponent changes, to crypto.getRandomValues. | git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:745; git show a77c45dd68:public/_kit/precommit.js:291; git show a77c45dd68:research/the-guess-it-showed-you-first/verify-the-guess-it-showed-you-first.mjs | Entry 41, body-14-2. The page creates "new G.ShannonMachine()" once and passes it to "P.chanceControl(controlMachine, 2048, ['L', 'R'])". The kit loops predict(), random draw, observe() without resetting the object. The rendered PASS criterion is interval inclusion of 0.5, not a guarantee every honest finite run passes. - OBSERVED | Its memory is never reset inside the run. | git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:745; git show a77c45dd68:public/_kit/precommit.js:291; git show a77c45dd68:research/the-guess-it-showed-you-first/verify-the-guess-it-showed-you-first.mjs | Entry 42, body-14-3. The page creates "new G.ShannonMachine()" once and passes it to "P.chanceControl(controlMachine, 2048, ['L', 'R'])". The kit loops predict(), random draw, observe() without resetting the object. The rendered PASS criterion is interval inclusion of 0.5, not a guarantee every honest finite run passes. - OBSERVED | Not run yet. | git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:78; git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:148 | Entry 43, body-15-1. Initial markup literally says "Not run yet."; the click handlers subsequently replace it with the run result. This is the initial UI state, not a claim that the code was never verified. - OBSERVED | A valid run has an exact interval containing 50%. | git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:745; git show a77c45dd68:public/_kit/precommit.js:291; git show a77c45dd68:research/the-guess-it-showed-you-first/verify-the-guess-it-showed-you-first.mjs | Entry 44, body-15-2. The page creates "new G.ShannonMachine()" once and passes it to "P.chanceControl(controlMachine, 2048, ['L', 'R'])". The kit loops predict(), random draw, observe() without resetting the object. The rendered PASS criterion is interval inclusion of 0.5, not a guarantee every honest finite run passes. - CONFIRMED | A correct machine still shows a red FAIL THIS RUN on about one honest run in twenty: that is what a 95% interval means, and the red is built in rather than hidden. | https://stat.ethz.ch/R-manual/R-devel/library/stats/html/binom.test.html | Entry 45, body-16-1. R states: "This guarantees that the confidence level is at least conf.level". This alone gives an upper bound on rejection, not exactly 5%. Independently summing Binomial(2048, 0.5) tails gives rejection for hits below 980 or above 1068, probability 0.049198161497. Thus the page's "about one" in twenty is accurate for its disclosed run length. See evidence/independent-arithmetic.log. - OBSERVED | Chance misses one time in twenty; a machine that could read the generator misses essentially never, which is what the verifier's cheating predictor demonstrates offline. | git show a77c45dd68:research/the-guess-it-showed-you-first/verify-the-guess-it-showed-you-first.mjs; recreated: pinned verifier; evidence/independent-arithmetic.log | Entry 46, body-16-3. Recreated the verifier controls. FrozenMemoryMachine fails tracking; the deliberately readable RNG lets the cheating predictor score "400" of 400 and the chance control rejects it. At 2048 honest independent coin rounds the rejection probability is 0.049198161497, consistent with the approximate one-in-twenty wording. - CONFIRMED | A real coin does not make the machine lose. | https://www.rclab.de/_media/shannon/hagelbarger1959_seer.pdf | Entry 47, body-17-1. Checked p. 1, Introduction against the directly fetched source: "a safe strategy is to play randomly with equal probability of heads and tails. This assures your not losing much since you can expect to win on the average half the time and come out even.". The excerpt supports this claim in the surrounding source context. - CONFIRMED | It makes both sides approach parity. | https://www.rclab.de/_media/shannon/hagelbarger1959_seer.pdf | Entry 48, body-17-2. Checked p. 1, Introduction against the directly fetched source: "a safe strategy is to play randomly with equal probability of heads and tails. This assures your not losing much since you can expect to win on the average half the time and come out even.". The excerpt supports this claim in the surrounding source context. - OBSERVED | Use the cryptographic-coin button in the live game and watch that ledger's interval settle around one half. | git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html; git show a77c45dd68:public/_kit/precommit.js; git show a77c45dd68:research/the-guess-it-showed-you-first/verify-the-guess-it-showed-you-first.mjs; https://www.rclab.de/_media/shannon/hagelbarger1959_seer.pdf | Entry 49, body-17-3. The fresh coin draw is made independently after the prediction; matching a fair coin has conditional probability 1/2. The same class against the deterministic control tape scores 2094/4096, interval containing 1/2; the state-tracking adversary wins 7465/10000. These are disclosed finite demonstrations of the parity/defeat mechanisms, not universal finite-run equalities. - OBSERVED | Coin rounds are scored in their own ledger, so the coin's parity is never averaged into the machine's rate against you, and one rate is never reported across two populations. | git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:448; git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:595; recreated: node runtime-audit.mjs | Entry 50, body-17-4. Inspected newLedgers(), recordRound() and summaryLines(); both populations use the same ShannonMachine instance and separate ledgers. Executed reader then coin input: the reader line stays unchanged after the coin round. Each line retains its own exact interval and p-value. The one-session/population warning is appropriate; "nearly no evidence" is not a quantitative power assertion. - OBSERVED | The machine is not failing. | git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html; git show a77c45dd68:public/_kit/precommit.js; git show a77c45dd68:research/the-guess-it-showed-you-first/verify-the-guess-it-showed-you-first.mjs; https://www.rclab.de/_media/shannon/hagelbarger1959_seer.pdf | Entry 51, body-17-5. The fresh coin draw is made independently after the prediction; matching a fair coin has conditional probability 1/2. The same class against the deterministic control tape scores 2094/4096, interval containing 1/2; the state-tracking adversary wins 7465/10000. These are disclosed finite demonstrations of the parity/defeat mechanisms, not universal finite-run equalities. - OBSERVED | It has been denied a repeatable behavior. | git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html; git show a77c45dd68:public/_kit/precommit.js; git show a77c45dd68:research/the-guess-it-showed-you-first/verify-the-guess-it-showed-you-first.mjs; https://www.rclab.de/_media/shannon/hagelbarger1959_seer.pdf | Entry 52, body-17-6. The fresh coin draw is made independently after the prediction; matching a fair coin has conditional probability 1/2. The same class against the deterministic control tape scores 2094/4096, interval containing 1/2; the state-tracking adversary wins 7465/10000. These are disclosed finite demonstrations of the parity/defeat mechanisms, not universal finite-run equalities. - CONFIRMED | Shannon's memo says the best possible play beats his machine three games to one. | https://www.jonglage.net/theorie/notation/siteswap-avancee/refs/books/Claude%20Shannon%20-%20Collected%20Papers.pdf; https://www.rclab.de/_media/shannon/hagelbarger1959_seer.pdf | Entry 53, body-18-1. Read and visually checked p. 689: "it can be beaten by the best possible play in the ratio 3:1" and "The player should repeat a behavior pattern twice". The rule and dependence on previous results are on pp. 688-689. The fair random-opponent component of md-plain-3 is additionally supported by Hagelbarger p. 1: "win on the average half the time". - CONFIRMED | The recipe is exact: for each of the eight situations, repeat one same-or-different behavior twice, then change it when the machine is prepared to follow. | https://www.jonglage.net/theorie/notation/siteswap-avancee/refs/books/Claude%20Shannon%20-%20Collected%20Papers.pdf; https://www.rclab.de/_media/shannon/hagelbarger1959_seer.pdf | Entry 54, body-18-2. Read and visually checked p. 689: "it can be beaten by the best possible play in the ratio 3:1" and "The player should repeat a behavior pattern twice". The rule and dependence on previous results are on pp. 688-689. The fair random-opponent component of md-plain-3 is additionally supported by Hagelbarger p. 1: "win on the average half the time". - CONFIRMED | That requires tracking every cell. | https://www.jonglage.net/theorie/notation/siteswap-avancee/refs/books/Claude%20Shannon%20-%20Collected%20Papers.pdf; https://www.rclab.de/_media/shannon/hagelbarger1959_seer.pdf | Entry 55, body-18-3. Read and visually checked p. 689: "it can be beaten by the best possible play in the ratio 3:1" and "The player should repeat a behavior pattern twice". The rule and dependence on previous results are on pp. 688-689. The fair random-opponent component of md-plain-3 is additionally supported by Hagelbarger p. 1: "win on the average half the time". - OBSERVED | This button does so with the same predictor class for 10,000 rounds. | git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:430; recreated: node runtime-audit.mjs | Entry 56, body-18-4. Recreated runAdversary(10000), seeded 19530318: "Player 7465 · machine 2535. Player rate 74.7%, exact 95% CI 73.8% to 75.5%. PASS: contains Shannon's 75%." It uses plan() to repeat an unconfirmed response and invert a repeated one. No fresh empirical evidence is created by rerunning the same tape. - OBSERVED | The seeded random fallback makes this a deterministic replay: the same 10,000 rounds and the same badge every time it is pressed, by anyone. | git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:430; recreated: node runtime-audit.mjs | Entry 57, body-19-1. Recreated runAdversary(10000), seeded 19530318: "Player 7465 · machine 2535. Player rate 74.7%, exact 95% CI 73.8% to 75.5%. PASS: contains Shannon's 75%." It uses plan() to repeat an unconfirmed response and invert a repeated one. No fresh empirical evidence is created by rerunning the same tape. - OBSERVED | It is a reproduction of a derivation, not a fresh verdict, and pressing it twice is not two pieces of evidence. | git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:430; recreated: node runtime-audit.mjs | Entry 58, body-19-2. Recreated runAdversary(10000), seeded 19530318: "Player 7465 · machine 2535. Player rate 74.7%, exact 95% CI 73.8% to 75.5%. PASS: contains Shannon's 75%." It uses plan() to repeat an unconfirmed response and invert a repeated one. No fresh empirical evidence is created by rerunning the same tape. - OBSERVED | The target is the player's 75%, and the exact interval must contain it. | git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:430; recreated: node runtime-audit.mjs | Entry 59, body-19-3. Recreated runAdversary(10000), seeded 19530318: "Player 7465 · machine 2535. Player rate 74.7%, exact 95% CI 73.8% to 75.5%. PASS: contains Shannon's 75%." It uses plan() to repeat an unconfirmed response and invert a repeated one. No fresh empirical evidence is created by rerunning the same tape. - CONFIRMED | William Poundstone reported doing better on later software games by hiding their score bars and trying to avoid over-switching and short runs. | https://william-poundstone.com/blog/2015/7/30/how-i-beat-the-mind-reading-machine | Entry 60, body-20-1. Checked Paragraph beginning I found I did better against the directly fetched source: "I found I did better when I tried to ignore the feedback, and better yet when I made sure I couldn't see the bars. (I resized the window so that the bars weren't visible in Mauboussin's game; covered the top part of my phone screen with Wong's app).". The excerpt supports this claim in the surrounding source context. - CONFIRMED | He did not demonstrate a truly random feed, and he did not claim to simulate all eight cells. | https://william-poundstone.com/blog/2015/7/30/how-i-beat-the-mind-reading-machine | Entry 61, body-20-2. Checked Paragraph beginning I am not saying against the directly fetched source: "I'm not saying that I succeeded in being random. But, having written a book on the subject, I was at least aware of the common biases. In general we switch back and forth too much between choices and avoid long streaks of the same choice. In a truly random series of 50 binary choices, there is generally a streak of six consecutive identical choices (six "heads" or "tails" in a row). I didn't count, but I made an effort to stick on the same choice repeatedly, relative to what my instincts told me.". The excerpt supports this claim in the surrounding source context. - CONFIRMED | Glaschick objected that true randomness can only produce parity, while forcing losses needs state tracking. | https://rclab.de/_media/shannon/mindreader_overview.pdf | Entry 62, body-20-3. Checked p. 2, Verification and Testing against the directly fetched source: "Shannon and Hagelbarger claim that to force the machine to loose requires simulating its state and a response dependent on the history. To supply random input (as claimed by [Poundstone2014] to beat the machine by not loosing) is evidently futile. If the input sequence is truly random, the result can only be 50% wins, otherwise the machine could predict the claimed random sequence, in whatever direction.". The excerpt supports this claim in the surrounding source context. - OBSERVED | This page separates those claims: the coin control tests parity, and the adversary tests the 3:1 route. | git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:430; recreated: node runtime-audit.mjs | Entry 63, body-20-4. Recreated runAdversary(10000), seeded 19530318: "Player 7465 · machine 2535. Player rate 74.7%, exact 95% CI 73.8% to 75.5%. PASS: contains Shannon's 75%." It uses plan() to repeat an unconfirmed response and invert a repeated one. No fresh empirical evidence is created by rerunning the same tape. - CONFIRMED | The machine does not keep a global count of left and right. | https://www.jonglage.net/theorie/notation/siteswap-avancee/refs/books/Claude%20Shannon%20-%20Collected%20Papers.pdf | Entry 64, body-21-1. Read all eight situations on p. 688 and the complete update rule on p. 689. They use two results and the intervening same/different action, not global left/right frequencies. Positive control for that scope: the source explicitly enumerates situations 1 through 8. The shipped contextIndex() implements the same three binary fields. - CONFIRMED | Before a move it describes the last two results from your point of view and whether you stayed with the same side between them. | https://www.jonglage.net/theorie/notation/siteswap-avancee/refs/books/Claude%20Shannon%20-%20Collected%20Papers.pdf | Entry 65, body-21-2. Read all eight situations on p. 688 and the complete update rule on p. 689. They use two results and the intervening same/different action, not global left/right frequencies. Positive control for that scope: the source explicitly enumerates situations 1 through 8. The shipped contextIndex() implements the same three binary fields. - CONFIRMED | That makes one of eight situations. | https://www.jonglage.net/theorie/notation/siteswap-avancee/refs/books/Claude%20Shannon%20-%20Collected%20Papers.pdf | Entry 66, body-21-3. Read all eight situations on p. 688 and the complete update rule on p. 689. They use two results and the intervening same/different action, not global left/right frequencies. Positive control for that scope: the source explicitly enumerates situations 1 through 8. The shipped contextIndex() implements the same three binary fields. - CONFIRMED | Each situation owns two bits of useful memory. | https://www.jonglage.net/theorie/notation/siteswap-avancee/refs/books/Claude%20Shannon%20-%20Collected%20Papers.pdf | Entry 67, body-22-1. P. 689: "Within the cell two things are registered"; (1) the previous same/different response and (2) whether it repeated the preceding response in that situation. The next sentences explicitly say the machine follows a repeat and otherwise uses its random element. N1's excerpt for body-22-4 ended before that following sentence. - CONFIRMED | The first records whether you stayed or changed the last time this situation occurred. | https://www.jonglage.net/theorie/notation/siteswap-avancee/refs/books/Claude%20Shannon%20-%20Collected%20Papers.pdf | Entry 68, body-22-2. P. 689: "Within the cell two things are registered"; (1) the previous same/different response and (2) whether it repeated the preceding response in that situation. The next sentences explicitly say the machine follows a repeat and otherwise uses its random element. N1's excerpt for body-22-4 ended before that following sentence. - CONFIRMED | The second records whether that behavior matched the time before. | https://www.jonglage.net/theorie/notation/siteswap-avancee/refs/books/Claude%20Shannon%20-%20Collected%20Papers.pdf | Entry 69, body-22-3. P. 689: "Within the cell two things are registered"; (1) the previous same/different response and (2) whether it repeated the preceding response in that situation. The next sentences explicitly say the machine follows a repeat and otherwise uses its random element. N1's excerpt for body-22-4 ended before that following sentence. - CONFIRMED | If it matched, the machine predicts you will do it again. | https://www.jonglage.net/theorie/notation/siteswap-avancee/refs/books/Claude%20Shannon%20-%20Collected%20Papers.pdf | Entry 70, body-22-4. P. 689: "Within the cell two things are registered"; (1) the previous same/different response and (2) whether it repeated the preceding response in that situation. The next sentences explicitly say the machine follows a repeat and otherwise uses its random element. N1's excerpt for body-22-4 ended before that following sentence. - CONFIRMED | If not, the rotating commutator supplied a random choice. | https://www.jonglage.net/theorie/notation/siteswap-avancee/refs/books/Claude%20Shannon%20-%20Collected%20Papers.pdf | Entry 71, body-22-5. Checked p. 689, random element against the directly fetched source: "The random element in the machine is actually a commutator rotating at about 10 revolutions per second.". The excerpt supports this claim in the surrounding source context. - CONFIRMED | This is a tiny conditional sequence model, not a mind reader. | https://www.jonglage.net/theorie/notation/siteswap-avancee/refs/books/Claude%20Shannon%20-%20Collected%20Papers.pdf; https://www.rclab.de/_media/shannon/hagelbarger1959_seer.pdf | Entry 72, body-23-1. Read and visually checked p. 689: "it can be beaten by the best possible play in the ratio 3:1" and "The player should repeat a behavior pattern twice". The rule and dependence on previous results are on pp. 688-689. The fair random-opponent component of md-plain-3 is additionally supported by Hagelbarger p. 1: "win on the average half the time". - CONFIRMED | Its leverage is also its weakness: the state depends on the machine's earlier guesses, so a player who follows the state can arrange the history that fools it. | https://www.jonglage.net/theorie/notation/siteswap-avancee/refs/books/Claude%20Shannon%20-%20Collected%20Papers.pdf; https://www.rclab.de/_media/shannon/hagelbarger1959_seer.pdf | Entry 73, body-23-2. Read and visually checked p. 689: "it can be beaten by the best possible play in the ratio 3:1" and "The player should repeat a behavior pattern twice". The rule and dependence on previous results are on pp. 688-689. The fair random-opponent component of md-plain-3 is additionally supported by Hagelbarger p. 1: "win on the average half the time". - CONFIRMED | Both primary sources give the same order, and both give it plainly. | https://www.rclab.de/_media/shannon/hagelbarger1959_seer.pdf; https://www.jonglage.net/theorie/notation/siteswap-avancee/refs/books/Claude%20Shannon%20-%20Collected%20Papers.pdf | Entry 74, body-24-1. Read both primary accounts. Hagelbarger p. 1 says "The machine plays first so that the player knows it is not cheating" and "(The machine cannot hear.)". Shannon p. 688 says "the player should guess out loud" and only afterward "move the key switch". The spoken choice precedes the reveal; the electrical scoring input follows it. Shannon pp. 688-689 also give the eight cells, random commutator and 3:1 strategy. - CONFIRMED | Hagelbarger's page 1: the machine plays first so the player knows it is not cheating, the player announces plus or minus, “(The machine cannot hear.)”, then pushes the play button and the machine lights its lamp, and only then does the player enter the choice on the key. | https://www.rclab.de/_media/shannon/hagelbarger1959_seer.pdf | Entry 75, body-24-2. Checked p. 1, The Game; transcribed from scan with page OCR against the directly fetched source: "The machine plays first so that the player knows it is not cheating. When the machine ready lamp is on, the player announces whether he is going to play plus or minus. (The machine cannot hear.) He then pushes the machine play button and the machine lights either the plus or the minus lamp indicating its play. The player then indicates his choice on the opponent’s play key. If both choices are the same, the machine wins; if they are different, the player wins. The machine records the score and is ready to play again.". The excerpt supports this claim in the surrounding source context. - CONFIRMED | Shannon's memorandum: the player guesses “right” or “left” out loud, the center button is pressed, the machine lights its light, and the player then moves the key switch to the choice already made. | https://www.jonglage.net/theorie/notation/siteswap-avancee/refs/books/Claude%20Shannon%20-%20Collected%20Papers.pdf | Entry 76, body-24-3. Checked p. 688, opening instructions against the directly fetched source: "To play against the machine, the player should guess out loud either "right" or "left." The center button of the machine is then pressed and the machine will light up either the right or left light. If the machine matches the player, the machine wins, otherwise the player wins. The player should then move the key switch in the direction corresponding to the choice he made.". The excerpt supports this claim in the surrounding source context. - CONFIRMED | So the machine exposed its choice before accepting the electrical scoring input, but after the person had announced a choice out loud. | https://www.rclab.de/_media/shannon/hagelbarger1959_seer.pdf; https://www.jonglage.net/theorie/notation/siteswap-avancee/refs/books/Claude%20Shannon%20-%20Collected%20Papers.pdf | Entry 77, body-24-4. Read both primary accounts. Hagelbarger p. 1 says "The machine plays first so that the player knows it is not cheating" and "(The machine cannot hear.)". Shannon p. 688 says "the player should guess out loud" and only afterward "move the key switch". The spoken choice precedes the reveal; the electrical scoring input follows it. Shannon pp. 688-689 also give the eight cells, random commutator and 3:1 strategy. - CONFIRMED | The whole 1953 guarantee rests on a machine that cannot hear and a person's word. | https://www.rclab.de/_media/shannon/hagelbarger1959_seer.pdf; https://www.jonglage.net/theorie/notation/siteswap-avancee/refs/books/Claude%20Shannon%20-%20Collected%20Papers.pdf | Entry 78, body-24-5. Read both primary accounts. Hagelbarger p. 1 says "The machine plays first so that the player knows it is not cheating" and "(The machine cannot hear.)". Shannon p. 688 says "the player should guess out loud" and only afterward "move the key switch". The spoken choice precedes the reveal; the electrical scoring input follows it. Shannon pp. 688-689 also give the eight cells, random commutator and 3:1 strategy. - OBSERVED | This page replaces both with a cryptographic commitment you can check in your own terminal, published before you choose and revealed after. | git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:630; git show a77c45dd68:public/_kit/precommit.js:97; recreated: node runtime-audit.mjs | Entry 79, body-24-6. Read prepareRound() and act(), then executed the shipped page-app in the pinned DOM harness. "sealed = await P.seal" precedes digest assignment and "setChoiceDisabled(false)"; act() calls "P.verify(sealed.hash, text)" before rendering the reveal and ledger. The displayed POSIX command independently returned the same hash, exit 0. See evidence/runtime.json and evidence/reader-shell.log. - CONFIRMED | A four-page Bell Laboratories internal memorandum, not a journal paper. | https://www.jonglage.net/theorie/notation/siteswap-avancee/refs/books/Claude%20Shannon%20-%20Collected%20Papers.pdf; https://www.historyofscience.com/pdf/82.pdf | Entry 80, body-25-1. N1's footnote excerpt establishes the memorandum and date, but not four pages. Supplemented with the dealer's first-hand description of the original typescript, catalog 82 p. 20: "4ff." and "The paper is unpublished apart from its inclusion in the collected papers." The three reprint pages 688-690 are not evidence that the original had only three leaves. - CONFIRMED | It defines the eight situations, the two-part memory cell, the random fallback, and the 3:1 best response. | https://www.jonglage.net/theorie/notation/siteswap-avancee/refs/books/Claude%20Shannon%20-%20Collected%20Papers.pdf | Entry 81, body-25-2. The full memorandum pp. 688-689 defines "eight possible situations", the two stored fields, random fallback, and best play "in the ratio 3:1". N1's selected excerpt supplies only the two-field memory part. - UNVERIFIABLE | A peer-reviewed IRE article. | https://api.crossref.org/works/10.1109/TEC.1956.5219783; https://www.rclab.de/_media/shannon/hagelbarger1959_seer.pdf; https://www.computer.org/about/cs-history/1950-1969 | Entry 82, body-26-1. The scan confirms an IRE article and says "Original manuscript received by the PGEC, October 28, 1955." Crossref positively identifies "SEER, A SEquence Extrapolating Robot" with "type":"journal-article", but that metadata does not establish peer review of this particular paper. The IEEE history confirms the journal lineage; the current CSDL scope URL returned an application shell. Peer-review status remains unverified, not disproved. - CONFIRMED | SEER recorded 5,218 wins in 9,795 plays against Bell Labs visitors and employees. | https://www.rclab.de/_media/shannon/hagelbarger1959_seer.pdf | Entry 83, body-26-2. Read and visually checked Hagelbarger p. 1. It prints "9,795 plays", "5,218 times" and "4,577" losses, then explains easy sequences, cheating, losers quitting, winners continuing, and short runs. 5,218 + 4,577 = 9,795. The selection and dependence caveats accurately limit the ledger. - CONFIRMED | Hagelbarger explicitly warned that short runs, self-selection, deliberate easy patterns, and deliberate cheating biased that ledger. | https://www.rclab.de/_media/shannon/hagelbarger1959_seer.pdf | Entry 84, body-26-3. Read and visually checked Hagelbarger p. 1. It prints "9,795 plays", "5,218 times" and "4,577" losses, then explains easy sequences, cheating, losers quitting, winners continuing, and short runs. 5,218 + 4,577 = 9,795. The selection and dependence caveats accurately limit the ledger. - OBSERVED | The exact 95% binomial interval is computing, but dependence and selection make it descriptive, not a population interval. | git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:592; git show a77c45dd68:public/_kit/precommit.js:198; evidence/independent-arithmetic.log | Entry 85, body-26-4. Rendered output: "52.3% to 54.3% around 5,218 / 9,795 (53.3%)". Independent beta quantiles match all page intervals. The page explicitly limits their interpretation for dependent adaptive sequences; the implementation does not turn them into a population claim. - CONFIRMED | A self-published technical report by the builder of two relay replicas. | https://rclab.de/_media/shannon/mindreader_overview.pdf; https://rclab.de/_media/shannon/mindreader_variants.pdf | Entry 86, body-27-1. Overview p. 2: "Finally, I made two two replicas for the HNF close to the original". Its pp. 2-3 supply the eight behavioral categories; the variants report p. 8 supplies the measured replica table. N1 quoted only the construction clause of this composite claim. - CONFIRMED | He found two contact-label typos and a missing-contact problem; half the memory stayed unused until the third correction. | https://rclab.de/_media/shannon/mindreader_overview.pdf; https://rclab.de/_media/shannon/mindreader_variants.pdf | Entry 87, body-27-2. Overview pp. 1-3 identifies "W2 instead of W1´ and 1´ instead of 1", insertion of a contact, and "half of the memory was unused". Variants p. 1 clarifies the proper solution requires "insertion of two contacts and to change the designation of a third one". The page's "missing-contact problem" correctly avoids claiming the final repair was just one contact. - OBSERVED | His corrected machines supplied the anchor above. | git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:264; https://rclab.de/_media/shannon/mindreader_overview.pdf; https://rclab.de/_media/shannon/mindreader_variants.pdf | Entry 88, body-27-3. The anchor targets reproduce the eight categories in the overview. All transcribed physical replica counts match the variants report p. 8 and the code's REPLICA_RESULTS. - CONFIRMED | No population claim: Bell Labs visitors and employees in the 1950s were not a random sample, 9,795 dependent plays are not 9,795 independent people, and your few rounds are one session in 2026. | https://www.rclab.de/_media/shannon/hagelbarger1959_seer.pdf | Entry 89, body-28-1. Read and visually checked Hagelbarger p. 1. It prints "9,795 plays", "5,218 times" and "4,577" losses, then explains easy sequences, cheating, losers quitting, winners continuing, and short runs. 5,218 + 4,577 = 9,795. The selection and dependence caveats accurately limit the ledger. - OBSERVED | The page reports each ledger for what it is. | git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:448; git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:595; recreated: node runtime-audit.mjs | Entry 90, body-28-2. Inspected newLedgers(), recordRound() and summaryLines(); both populations use the same ShannonMachine instance and separate ledgers. Executed reader then coin input: the reader line stays unchanged after the coin round. Each line retains its own exact interval and p-value. The one-session/population warning is appropriate; "nearly no evidence" is not a quantitative power assertion. - OBSERVED | The page recomputes all eight 100-round rows with the shipped class. | git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:414; git show a77c45dd68:research/the-guess-it-showed-you-first/verify-the-guess-it-showed-you-first.mjs; https://www.jonglage.net/theorie/notation/siteswap-avancee/refs/books/Claude%20Shannon%20-%20Collected%20Papers.pdf | Entry 91, body-29-1. Recreated the pinned engine: "runSequence(testCase, 100, 19530318)" yields hits 94,97,95,68,69,55,59,58. runAdversary(10000) yields 7465 player wins. Shannon's footnote supplies March 18, 1953; 100, 10000 and the integer seed are disclosed implementation choices. See evidence/verifier-pinned.log. - OBSERVED | Expected hits are 94, 97, 95, 68, 69, 55, 59, and 58. | git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:414; git show a77c45dd68:research/the-guess-it-showed-you-first/verify-the-guess-it-showed-you-first.mjs; https://www.jonglage.net/theorie/notation/siteswap-avancee/refs/books/Claude%20Shannon%20-%20Collected%20Papers.pdf | Entry 92, body-29-2. Recreated the pinned engine: "runSequence(testCase, 100, 19530318)" yields hits 94,97,95,68,69,55,59,58. runAdversary(10000) yields 7465 player wins. Shannon's footnote supplies March 18, 1953; 100, 10000 and the integer seed are disclosed implementation choices. See evidence/verifier-pinned.log. - OBSERVED | The offline verifier derives them again from the primary-source state rule. | git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:745; git show a77c45dd68:public/_kit/precommit.js:291; git show a77c45dd68:research/the-guess-it-showed-you-first/verify-the-guess-it-showed-you-first.mjs | Entry 93, body-29-3. The page creates "new G.ShannonMachine()" once and passes it to "P.chanceControl(controlMachine, 2048, ['L', 'R'])". The kit loops predict(), random draw, observe() without resetting the object. The rendered PASS criterion is interval inclusion of 0.5, not a guarantee every honest finite run passes. - OBSERVED | A fresh, unmodified ShannonMachine carries its memory through 2,048 cryptographic opponent moves. | git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:745; git show a77c45dd68:public/_kit/precommit.js:291; git show a77c45dd68:research/the-guess-it-showed-you-first/verify-the-guess-it-showed-you-first.mjs | Entry 94, body-30-1. The page creates "new G.ShannonMachine()" once and passes it to "P.chanceControl(controlMachine, 2048, ['L', 'R'])". The kit loops predict(), random draw, observe() without resetting the object. The rendered PASS criterion is interval inclusion of 0.5, not a guarantee every honest finite run passes. - OBSERVED | A run fails visibly if its exact interval excludes 50%. | git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:745; git show a77c45dd68:public/_kit/precommit.js:291; git show a77c45dd68:research/the-guess-it-showed-you-first/verify-the-guess-it-showed-you-first.mjs | Entry 95, body-30-2. The page creates "new G.ShannonMachine()" once and passes it to "P.chanceControl(controlMachine, 2048, ['L', 'R'])". The kit loops predict(), random draw, observe() without resetting the object. The rendered PASS criterion is interval inclusion of 0.5, not a guarantee every honest finite run passes. - OBSERVED | The verifier freezes the machine's memory and requires the anchor check to fail. | git show a77c45dd68:research/the-guess-it-showed-you-first/verify-the-guess-it-showed-you-first.mjs; recreated: pinned verifier; evidence/independent-arithmetic.log | Entry 96, body-31-1. Recreated the verifier controls. FrozenMemoryMachine fails tracking; the deliberately readable RNG lets the cheating predictor score "400" of 400 and the chance control rejects it. At 2048 honest independent coin rounds the rejection probability is 0.049198161497, consistent with the approximate one-in-twenty wording. - OBSERVED | It also gives the chance harness a predictor that can see the deterministic test RNG and requires the interval to exclude 50%. | git show a77c45dd68:research/the-guess-it-showed-you-first/verify-the-guess-it-showed-you-first.mjs; recreated: pinned verifier; evidence/independent-arithmetic.log | Entry 97, body-31-2. Recreated the verifier controls. FrozenMemoryMachine fails tracking; the deliberately readable RNG lets the cheating predictor score "400" of 400 and the chance control rejects it. At 2048 honest independent coin rounds the rejection probability is 0.049198161497, consistent with the approximate one-in-twenty wording. - OBSERVED | Every choice waits for a SHA-256 digest. | git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:630; git show a77c45dd68:public/_kit/precommit.js:97; recreated: node runtime-audit.mjs | Entry 98, body-32-1. Read prepareRound() and act(), then executed the shipped page-app in the pinned DOM harness. "sealed = await P.seal" precedes digest assignment and "setChoiceDisabled(false)"; act() calls "P.verify(sealed.hash, text)" before rendering the reveal and ledger. The displayed POSIX command independently returned the same hash, exit 0. See evidence/runtime.json and evidence/reader-shell.log. - OBSERVED | The reveal is the exact canonical preimage from the shared kit, verified again in-page and printed with a POSIX shell command. | git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:630; git show a77c45dd68:public/_kit/precommit.js:97; recreated: node runtime-audit.mjs | Entry 99, body-32-2. Read prepareRound() and act(), then executed the shipped page-app in the pinned DOM harness. "sealed = await P.seal" precedes digest assignment and "setChoiceDisabled(false)"; act() calls "P.verify(sealed.hash, text)" before rendering the reveal and ledger. The displayed POSIX command independently returned the same hash, exit 0. See evidence/runtime.json and evidence/reader-shell.log. - OBSERVED | The anchor uses 100 rounds and a xorshift32 tape seeded from 1953-03-18. | git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:414; git show a77c45dd68:research/the-guess-it-showed-you-first/verify-the-guess-it-showed-you-first.mjs; https://www.jonglage.net/theorie/notation/siteswap-avancee/refs/books/Claude%20Shannon%20-%20Collected%20Papers.pdf | Entry 100, body-33-1. Recreated the pinned engine: "runSequence(testCase, 100, 19530318)" yields hits 94,97,95,68,69,55,59,58. runAdversary(10000) yields 7465 player wins. Shannon's footnote supplies March 18, 1953; 100, 10000 and the integer seed are disclosed implementation choices. See evidence/verifier-pinned.log. - OBSERVED | The adversary uses 10,000 rounds. | git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:430; recreated: node runtime-audit.mjs | Entry 101, body-33-2. Recreated runAdversary(10000), seeded 19530318: "Player 7465 · machine 2535. Player rate 74.7%, exact 95% CI 73.8% to 75.5%. PASS: contains Shannon's 75%." It uses plan() to repeat an unconfirmed response and invert a repeated one. No fresh empirical evidence is created by rerunning the same tape. - OBSERVED | These lengths and seeds make the demonstrations replayable; they are not source measurements. | git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:414; git show a77c45dd68:research/the-guess-it-showed-you-first/verify-the-guess-it-showed-you-first.mjs; https://www.jonglage.net/theorie/notation/siteswap-avancee/refs/books/Claude%20Shannon%20-%20Collected%20Papers.pdf | Entry 102, body-33-3. Recreated the pinned engine: "runSequence(testCase, 100, 19530318)" yields hits 94,97,95,68,69,55,59,58. runAdversary(10000) yields 7465 player wins. Shannon's footnote supplies March 18, 1953; 100, 10000 and the integer seed are disclosed implementation choices. See evidence/verifier-pinned.log. - OBSERVED | The tape matters: the sweep button above replays the eight predicates on 400 other tapes and reports live how often all eight survive. | git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:498; recreated: node runtime-audit.mjs | Entry 103, body-33-4. Recreated all 400 seeds through the shipped class: "All eight predicates hold on 307 of 400 alternative tapes". Failures are SD 52, SSDD 25, SDD 13, SSD 7, SSSD 2, SDDD 2; S and D never fail in this disclosed pool. See evidence/runtime.json. - OBSERVED | Rounds you choose and rounds the cryptographic coin chooses are scored in separate ledgers with separate intervals. | git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:448; git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:595; recreated: node runtime-audit.mjs | Entry 104, body-34-1. Inspected newLedgers(), recordRound() and summaryLines(); both populations use the same ShannonMachine instance and separate ledgers. Executed reader then coin input: the reader line stays unchanged after the coin round. Each line retains its own exact interval and p-value. The one-session/population warning is appropriate; "nearly no evidence" is not a quantitative power assertion. - OBSERVED | A single blended rate across both would let the coin's parity be reported as evidence about you, or your rounds be reported as a machine that beats a CSPRNG. | git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:448; git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:595; recreated: node runtime-audit.mjs | Entry 105, body-34-2. Inspected newLedgers(), recordRound() and summaryLines(); both populations use the same ShannonMachine instance and separate ledgers. Executed reader then coin input: the reader line stays unchanged after the coin round. Each line retains its own exact interval and p-value. The one-session/population warning is appropriate; "nearly no evidence" is not a quantitative power assertion. - OBSERVED | Every rate in an instrument carries an exact 95% Clopper-Pearson interval. | git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:592; git show a77c45dd68:public/_kit/precommit.js:198; evidence/independent-arithmetic.log | Entry 106, body-35-1. Rendered output: "52.3% to 54.3% around 5,218 / 9,795 (53.3%)". Independent beta quantiles match all page intervals. The page explicitly limits their interpretation for dependent adaptive sequences; the implementation does not turn them into a population claim. - CONFIRMED | An honest chance control lands outside its 95% interval about one run in twenty, so an occasional red is expected and is not an indictment. | https://stat.ethz.ch/R-manual/R-devel/library/stats/html/binom.test.html | Entry 107, body-35-2. R states: "This guarantees that the confidence level is at least conf.level". This alone gives an upper bound on rejection, not exactly 5%. Independently summing Binomial(2048, 0.5) tails gives rejection for hits below 980 or above 1068, probability 0.049198161497. Thus the page's "about one" in twenty is accurate for its disclosed run length. See evidence/independent-arithmetic.log. - CONFIRMED | Exact binomial p-values assume independent trials, which adaptive rounds are not. | https://csrc.nist.gov/glossary/term/binomial_distribution | Entry 108, body-35-3. NIST defines the binomial variable using "n independent Bernoulli experiments" and a common probability p. This supports the limitation for adaptive human or deterministic-sequence ledgers. A fair independent coin still gives conditional hit probability 1/2 even against an adaptive predictor, which is why the disclosed chance control is valid. - OBSERVED | They are shown as diagnostics, not repaired into population evidence. | git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:592; git show a77c45dd68:public/_kit/precommit.js:198; evidence/independent-arithmetic.log | Entry 109, body-35-4. Rendered output: "52.3% to 54.3% around 5,218 / 9,795 (53.3%)". Independent beta quantiles match all page intervals. The page explicitly limits their interpretation for dependent adaptive sequences; the implementation does not turn them into a population claim. - OBSERVED | Four often-cited emulator projects were checked, and every URL tried is printed below with what each returned. | git show a77c45dd68:research/the-guess-it-showed-you-first/README.md; evidence/fetch.log; https://web.media.mit.edu/ | Entry 110, body-36-1. The dated README records five attempted URLs for four projects, three reachable project pages and one 404. Direct curl reproduces all five statuses: Williams/AWK/Success/current-old return 200, MIT game 404. Williams explicitly requires the s key; Success has a hidden hatch; AWK prints its guess in report results. The old Success URL returns the home-page title. MIT root and Raskar profile are same-host 200 positive controls. Timings and withdrawn-draft history are supported by the dated record, not contradicted by current timings. - OBSERVED | Three projects are reachable and one is dead. | git show a77c45dd68:research/the-guess-it-showed-you-first/README.md; evidence/fetch.log; https://web.media.mit.edu/ | Entry 111, body-36-2. The dated README records five attempted URLs for four projects, three reachable project pages and one 404. Direct curl reproduces all five statuses: Williams/AWK/Success/current-old return 200, MIT game 404. Williams explicitly requires the s key; Success has a hidden hatch; AWK prints its guess in report results. The old Success URL returns the home-page title. MIT root and Raskar profile are same-host 200 positive controls. Timings and withdrawn-draft history are supported by the dated record, not contradicted by current timings. - OBSERVED | None shows the machine's guess by default: Williams's page hides it behind the “s” key, the Success Equation remake commits its guess before you press but keeps it in a panel that ships closed, and the AWK program reports its guess after reading your play. | git show a77c45dd68:research/the-guess-it-showed-you-first/README.md; evidence/fetch.log; https://web.media.mit.edu/ | Entry 112, body-36-3. The dated README records five attempted URLs for four projects, three reachable project pages and one 404. Direct curl reproduces all five statuses: Williams/AWK/Success/current-old return 200, MIT game 404. Williams explicitly requires the s key; Success has a hidden hatch; AWK prints its guess in report results. The old Success URL returns the home-page title. MIT root and Raskar profile are same-host 200 positive controls. Timings and withdrawn-draft history are supported by the dated record, not contradicted by current timings. - MINOR | Glaschick's own wording is “Some software implementations require to trust the code that only reveals win or loose, not the machine's choice.” Four links measure four links, not how many. | https://rclab.de/_media/shannon/mindreader_overview.pdf; git show a77c45dd68:research/the-guess-it-showed-you-first/README.md | triage X1; error four-links-count; decision fix; reread=true. Re-read the page (public/strata/the-guess-it-showed-you-first/index.html:239-245) and research/the-guess-it-showed-you-first/README.md:232: the survey prints five URL rows and the record says "five URLs for four distinct projects", so "Four links measure four links" miscounts the links; the point (a tiny sample) stands. Also re-fetched rclab.de mindreader_overview.pdf: line 67-69 of its text reads "Some software implementations require to trust the code that only reveals win or loose, not the machine´s choice, and may also use more resources", so the quoted words are the source's words (apostrophe normalised); the page closed the quotation with a period inside the marks where the source continues, so the period moves outside, no words inside the quotation change. Cheap MINOR wording fix: say four projects. Same error label as X2. The verifier asserts the old sentence, so its assertion is updated to the new one. Checker's note: Entry 113, body-36-4. The quoted criticism is substantively present in the overview. The numerical survey summary should say four projects or five URLs: the pinned research record says "five URLs for four distinct projects" and the page prints five URL rows. N1 copied the four-links wording without resolving that discrepancy. The offline quote warning also reflects the source's acute apostrophe and continued sentence; it does not change the substantive criticism. - OBSERVED | Shannon, Hagelbarger, Glaschick, and Breazu were read in full. | git show a77c45dd68:research/the-guess-it-showed-you-first/README.md | Entry 114, body-37-1. README records the source reads, file checksums, and White access attempts: "the ScienceDirect PDF returned HTTP 403" and "the unauthenticated Elsevier full-text endpoint returned metadata only". This is an account of the August 13 investigation. The page explicitly limits White to bibliographic identification. Its source-specific observations are supported under step 7. - OBSERVED | White's 1959 full text could not be retrieved from its open Elsevier record and is not used for any factual claim beyond bibliographic identification. | git show a77c45dd68:research/the-guess-it-showed-you-first/README.md | Entry 115, body-37-2. README records the source reads, file checksums, and White access attempts: "the ScienceDirect PDF returned HTTP 403" and "the unauthenticated Elsevier full-text endpoint returned metadata only". This is an account of the August 13 investigation. The page explicitly limits White to bibliographic identification. Its source-specific observations are supported under step 7. - CONFIRMED | Claude E. Shannon, “A Mind-Reading (?) Machine,” Bell Laboratories memorandum, 18 March 1953, reprinted pp. 688-690 | https://www.jonglage.net/theorie/notation/siteswap-avancee/refs/books/Claude%20Shannon%20-%20Collected%20Papers.pdf | Entry 116, citation-38. Directly read pp. 688-690: author "Claude E. Shannon", title "A Mind-Reading (?) Machine", and footnote "Bell Laboratories Memorandum, March 18, 1953." The final reprint page contains the circuit. The OCR's stray final title character is not in the printed title. - CONFIRMED | The interface order, eight situations, two-part cell, random commutator, and best play winning 3:1. | https://www.rclab.de/_media/shannon/hagelbarger1959_seer.pdf; https://www.jonglage.net/theorie/notation/siteswap-avancee/refs/books/Claude%20Shannon%20-%20Collected%20Papers.pdf | Entry 117, body-39-1. Read both primary accounts. Hagelbarger p. 1 says "The machine plays first so that the player knows it is not cheating" and "(The machine cannot hear.)". Shannon p. 688 says "the player should guess out loud" and only afterward "move the key switch". The spoken choice precedes the reveal; the electrical scoring input follows it. Shannon pp. 688-689 also give the eight cells, random commutator and 3:1 strategy. - CONFIRMED | D. W. Hagelbarger, “SEER, A SEquence Extrapolating Robot,” IRE Transactions on Electronic Computers, March 1956, pp. 1-7 | https://api.crossref.org/works/10.1109/TEC.1956.5219783; https://www.rclab.de/_media/shannon/hagelbarger1959_seer.pdf | Entry 118, citation-40. Crossref confirms Hagelbarger, "SEER, A SEquence Extrapolating Robot", March 1956, EC-5(1), pp. 1-7, DOI 10.1109/TEC.1956.5219783. The paper's masthead is "IRE TRANSACTIONS ON ELECTRONIC COMPUTERS". Crossref normalizes the journal name to IEEE; the historical page citation correctly uses IRE. - CONFIRMED | The play-first interface, 5,218 of 9,795 human-play ledger and its biases, plus the umpired machine duel reported about 55-45 for Shannon over several thousand games. | https://www.rclab.de/_media/shannon/hagelbarger1959_seer.pdf | Entry 119, body-41-1. Hagelbarger p. 1 supports the interface and human ledger. P. 3 describes the umpire connecting the machines, "several thousand games", and the smaller machine winning "about 55-45". N1's selected excerpt supplies only the last part of this composite claim. - CONFIRMED | Rainer Glaschick, “Shannon's Mind-Reading Machine,” 26 August 2024 | https://rclab.de/_media/shannon/mindreader_overview.pdf | Entry 120, citation-42. Checked p. 1, heading against the directly fetched source: "Shannon´s "Mind-Reading Machine" Rainer Glaschick, Paderborn (rainer@glaschick.de) 2024-08-26". The excerpt supports this claim in the surrounding source context. - CONFIRMED | The replica history, correction summary, behavioral specification, and software-interface criticism. | https://rclab.de/_media/shannon/mindreader_overview.pdf | Entry 121, body-43-1. Read all four overview pages: replica construction and label corrections on pp. 1-2, interface criticism on p. 2, behavioral rules and unused memory on pp. 2-3. The selected behavioral excerpt is genuine but covers only one part of the claim. - CONFIRMED | Rainer Glaschick, “Correcting The Schematics Of Shannon's Mind-Reading Machine,” 29 March 2024 | https://rclab.de/_media/shannon/mindreader_variants.pdf | Entry 122, citation-44. Checked p. 1, heading against the directly fetched source: "Rainer Glaschick Correcting The Schematics Of Shannon's "Mind-Rea... Correcting The Schematics Of Shannon's "Mind-Reading Machine" Rainer Glaschick, Paderborn (rainer@glaschick.de) 2024-03-29". The excerpt supports this claim in the surrounding source context. - CONFIRMED | The contact corrections, core algorithm, simulator variants, and the page-8 measured results from both physical replicas, which are recomputed row by row in the anchor section above. | https://rclab.de/_media/shannon/mindreader_variants.pdf | Entry 123, body-45-1. Read the full variants report, especially p. 8: "The physical relay implementation gave the following results". Nine rows, including separate S/l and S/r rows, match the transcription. Four rows use "*:5", "*:2", etc.; five use counted wins and losses. Tracking and random startup are explained on pp. 3-4. No denominator is provided for the asterisk rows. - CONFIRMED | M. Breazu, D. Volovici, D. I. Morariu, and R. G. Crețulescu, 2020 | https://reference-global.com/download/article/10.2478/ijasitels-2020-0003.pdf; https://api.crossref.org/works/10.2478/ijasitels-2020-0003 | Entry 124, citation-46. PDF p. 56 names "Macarie BREAZU", "Daniel VOLOVICI", "Daniel I. MORARIU", and "Radu G. CREȚULESCU"; its journal header dates it December 2020. Crossref DOI 10.2478/ijasitels-2020-0003 agrees with authors, year and title. - CONFIRMED | A software implementation of both machines. | https://reference-global.com/download/article/10.2478/ijasitels-2020-0003.pdf | Entry 125, body-47-1. Checked p. 56, Abstract against the directly fetched source: "In our research we perform a software implementation for both machines in order to test the common statement that MRM, even simpler, beats SEER. Also, we propose a simple contextual predictor (SCP) and use it to compete with SEER and MRM.". The excerpt supports this claim in the surrounding source context. - CONFIRMED | Across ten 100-play simulations, Shannon averaged 55.8 points to SEER's 44.2; across 200 plays it averaged 112.8 to 87.2. | https://reference-global.com/download/article/10.2478/ijasitels-2020-0003.pdf | Entry 126, body-47-2. Read Table 1 on p. 61 and its description of ten software games. The ten 100-play MRM scores sum to 558, and the ten 200-play scores sum to 1128, giving 55.8:44.2 and 112.8:87.2. These are software machine duels. Independent sums are retained in evidence/independent-arithmetic.log. - CONFIRMED | These are simulated machine duels, not human evidence. | https://reference-global.com/download/article/10.2478/ijasitels-2020-0003.pdf | Entry 127, body-47-3. Read Table 1 on p. 61 and its description of ten software games. The ten 100-play MRM scores sum to 558, and the ten 200-play scores sum to 1128, giving 55.8:44.2 and 112.8:87.2. These are software machine duels. Independent sums are retained in evidence/independent-arithmetic.log. - CONFIRMED | Gerald M. White, “Penny Matching Machines,” Information and Control 2 (1959), 349-363 | https://api.crossref.org/works/10.1016/S0019-9958(59)80013-9 | Entry 128, citation-48. Crossref gives "given":"Gerald M.", "family":"White", title "Penny matching machines", December 1959, volume 2 issue 4, pp. 349-363, and the cited DOI. The tool's "gives White the initial M." mismatch is a parser false positive: it grabs the middle initial in the correctly printed full name. - OBSERVED | The publisher record and abstract were reachable; the full text was not. | git show a77c45dd68:research/the-guess-it-showed-you-first/README.md | Entry 129, body-49-1. README records the source reads, file checksums, and White access attempts: "the ScienceDirect PDF returned HTTP 403" and "the unauthenticated Elsevier full-text endpoint returned metadata only". This is an account of the August 13 investigation. The page explicitly limits White to bibliographic identification. Its source-specific observations are supported under step 7. - OBSERVED | No technical result from it is used here. | git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:218; git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:233 | Entry 130, body-49-2. Read the full page: White appears only in the bibliography and the explicit source-access warning. Positive control for this scope: the same read finds technical results attributed to Shannon, Hagelbarger, Glaschick and Breazu. - CONFIRMED | William Poundstone, “How I Beat the Mind-Reading Machine,” 14 July 2014 | https://william-poundstone.com/blog/2015/7/30/how-i-beat-the-mind-reading-machine | Entry 131, citation-50. The fetched article heading is "How I Beat the Mind-Reading Machine" and the displayed byline date is "July 14, 2014". The 2015/7/30 path does not override that displayed date. - CONFIRMED | The first-person account of hiding feedback and deliberately allowing longer runs. | https://william-poundstone.com/blog/2015/7/30/how-i-beat-the-mind-reading-machine | Entry 132, body-51-1. Checked Paragraph beginning I found I did better against the directly fetched source: "I found I did better when I tried to ignore the feedback, and better yet when I made sure I couldn't see the bars. (I resized the window so that the bars weren't visible in Mauboussin's game; covered the top part of my phone screen with Wong's app).". The excerpt supports this claim in the surrounding source context. - CONFIRMED | It is presented as an account, not an experiment with verified random input. | https://william-poundstone.com/blog/2015/7/30/how-i-beat-the-mind-reading-machine | Entry 133, body-51-2. Checked Paragraph beginning I am not saying against the directly fetched source: "I'm not saying that I succeeded in being random. But, having written a book on the subject, I was at least aware of the common biases. In general we switch back and forth too much between choices and avoid long streaks of the same choice. In a truly random series of 50 binary choices, there is generally a streak of six consecutive identical choices (six "heads" or "tails" in a row). I didn't count, but I made an effort to stick on the same choice repeatedly, relative to what my instincts told me.". The excerpt supports this claim in the surrounding source context. - MINOR | Four links, checked on 2026-08-13. | git show a77c45dd68:research/the-guess-it-showed-you-first/README.md; git show a77c45dd68:public/strata/the-guess-it-showed-you-first/index.html:239-243 | triage X2; error four-links-count; decision fix; reread=true. Re-read index.html:237-245: "Four links, checked on 2026-08-13." heads five URL rows, and the page's own tally at :245 says "five URLs tried for four distinct projects". The date holds; the count is the project count, not the link count. Cheap MINOR fix: "Five URLs for four projects, checked on 2026-08-13." Checker's note: Entry 134, body-52-1. The pinned research survey says "five URLs for four distinct projects"; the page contains five URL rows. The date is supported, but four is the project count, not the link count. Suggested wording: "Five URLs, checked on 2026-08-13." - OBSERVED | The question is narrow: does the emulator show you its guess before you commit to yours? | git show a77c45dd68:research/the-guess-it-showed-you-first/README.md; evidence/fetch.log; https://web.media.mit.edu/ | Entry 135, body-52-2. The dated README records five attempted URLs for four projects, three reachable project pages and one 404. Direct curl reproduces all five statuses: Williams/AWK/Success/current-old return 200, MIT game 404. Williams explicitly requires the s key; Success has a hidden hatch; AWK prints its guess in report results. The old Success URL returns the home-page title. MIT root and Raskar profile are same-host 200 positive controls. Timings and withdrawn-draft history are supported by the dated record, not contradicted by current timings. - OBSERVED | Every URL is printed so the survey can be repeated or contradicted. | git show a77c45dd68:research/the-guess-it-showed-you-first/README.md; evidence/fetch.log; https://web.media.mit.edu/ | Entry 136, body-52-3. The dated README records five attempted URLs for four projects, three reachable project pages and one 404. Direct curl reproduces all five statuses: Williams/AWK/Success/current-old return 200, MIT game 404. Williams explicitly requires the s key; Success has a hidden hatch; AWK prints its guess in report results. The old Success URL returns the home-page title. MIT root and Raskar profile are same-host 200 positive controls. Timings and withdrawn-draft history are supported by the dated record, not contradicted by current timings. - OBSERVED | HTTP 200. The page is a wrapper for a legacy Java that modern browsers no longer run, so the game itself was not playable. | git show a77c45dd68:research/the-guess-it-showed-you-first/README.md; evidence/fetch.log; https://web.media.mit.edu/ | Entry 137, body-54-1. The dated README records five attempted URLs for four projects, three reachable project pages and one 404. Direct curl reproduces all five statuses: Williams/AWK/Success/current-old return 200, MIT game 404. Williams explicitly requires the s key; Success has a hidden hatch; AWK prints its guess in report results. The old Success URL returns the home-page title. MIT root and Raskar profile are same-host 200 positive controls. Timings and withdrawn-draft history are supported by the dated record, not contradicted by current timings. - CONFIRMED | Its own text: the prediction appears only “If you are interested in seeing the mind reader's prediction of your next press, type 's' (for show).” Hidden by default. | https://www.cs.williams.edu/~bailey/applets/MindReader/ | Entry 138, body-54-2. Checked Instructions below Java applet against the directly fetched source: "If you are interested in seeing the mind reader's prediction of your next press, type 's' (for show). The prediction is demonstrated by a red dot.". The excerpt supports this claim in the surrounding source context. - OBSERVED | HTTP 200, complete AWK source on the page. | git show a77c45dd68:research/the-guess-it-showed-you-first/README.md; evidence/fetch.log; https://web.media.mit.edu/ | Entry 139, body-56-1. The dated README records five attempted URLs for four projects, three reachable project pages and one 404. Direct curl reproduces all five statuses: Williams/AWK/Success/current-old return 200, MIT game 404. Williams explicitly requires the s key; Success has a hidden hatch; AWK prints its guess in report results. The old Success URL returns the home-page title. MIT root and Raskar profile are same-host 200 positive controls. Timings and withdrawn-draft history are supported by the dated record, not contradicted by current timings. - MINOR | Its chunk order is consult model, get input, report results: the guess is computed before the play and printed after it. | https://www.literateprograms.org/mind_reading_machine__awk_.html; https://ftp.gnu.org/old-gnu/Manuals/gawk-3.1.1/html_node/Next-Statement.html | triage X3; error awk-guess-before-play; decision fix; reread=true. Re-fetched https://www.literateprograms.org/mind_reading_machine__awk_.html: the main program lists "consult model", "get input", "report results" in that order, but every chunk is an AWK pattern-action rule run per input record, and the page's opening BEGIN block ends by printing "your play? ", so the player's line has already been read when the guess rules run. The guess uses only the history ("case = (hpa!=hca)..."), not the current play, so it is computed before the play is parsed, not before it is made. Loose wording, not a false claim about what the program shows (the page's other line, index.html:217, already says it "reports its guess after reading your play"). Cheap MINOR fix of a few words, also in the research README's copy. Checker's note: Entry 140, body-56-2. The source really orders "consult model", "get input", "report results". But these are AWK actions on the current input record: GNU explains that execution "reads an input record and then tests each rule's pattern against it". The guess is calculated before parsing the current play, after the user's input line has arrived. Change "before the play" to "before interpreting the input". - CONFIRMED | Inspectable, but not shown first. | https://www.literateprograms.org/mind_reading_machine__awk_.html | Entry 141, body-56-3. The complete source is visible. Its report-results action prints "; I guessed " only after the get-input rules have selected play. No default pre-input display is promised here; the inspectable/not-shown-first conclusion is correct. - OBSERVED | HTTP 200, and this is the live game, one hop from the site's own navigation. | git show a77c45dd68:research/the-guess-it-showed-you-first/README.md; evidence/fetch.log; https://web.media.mit.edu/ | Entry 142, body-58-1. The dated README records five attempted URLs for four projects, three reachable project pages and one 404. Direct curl reproduces all five statuses: Williams/AWK/Success/current-old return 200, MIT game 404. Williams explicitly requires the s key; Success has a hidden hatch; AWK prints its guess in report results. The old Success URL returns the home-page title. MIT root and Raskar profile are same-host 200 positive controls. Timings and withdrawn-draft history are supported by the dated record, not contradicted by current timings. - CONFIRMED | It is the companion to Poundstone's subject matter and its copy says the machine “has already guessed what you will choose” and that the guess is “committed before you press.” | https://success-equation.com/mind-reader/; https://william-poundstone.com/blog/2015/7/30/how-i-beat-the-mind-reading-machine | Entry 143, body-58-2. The fetched game says "The machine has already guessed what you will choose" and "committed before you press". Poundstone's account identifies the browser game on the website for Michael Mauboussin's book The Success Equation. N1 quoted only the second game sentence. - OBSERVED | The panel holding that guess is <section class="hatch" id="hatch" hidden>: a commitment that exists but ships closed. | git show a77c45dd68:research/the-guess-it-showed-you-first/README.md; https://success-equation.com/mind-reader/ | Entry 144, body-58-3. The dated README records the closed hatch. The current HTML independently confirms