When this site gets something wrong, the fix is logged here, in public, with a date. A site that hides its corrections is not more reliable, only quieter about its mistakes.
25 September 2026 (continued)
The Render Engine bench labelled its Rizwan Virk quotations "Virk 2019" and gave one as "(2019/2025, p. 164)", while its page references are to the 2025 second edition its source list names. Every label now says 2025.
The Fermi Paradox bench said Fermi's lunch companions "all agree Fermi then ran envelope numbers", and its story added that "the witnesses agree, everyone goes quiet". Eric Jones's 1985 Los Alamos report, which the bench cites, records that the question drew laughter, and only Herbert York recalls the calculation that followed. The story also had the four "joking about a New Yorker cartoon" on "missing trash-can lids" and Fermi putting "his fork down"; in the report Konopinski recalls a cartoon about trash cans vanishing from New York's streets, and there is no fork. The introduction said Fermi asked "walking to lunch" and called the question "four words"; the accounts place it at the table in Fuller Lodge. The bench now follows the report.
The Electrogravitics bench, the Secrets of Antigravity Propulsion note and the Tom Turman wiki entry said Brown told Turman "that no device had ever fully left the ground", and the note and the entry called it his "admission that no device of his ever fully lifted from the ground". Brown's letter of 23 January 1968, reproduced in the book's appendix A, answers a narrower question: whether he had ever got an anti-gravitic device to lift from the ground, "its weight being entirely removed. The answer is no." His letter of 1 November 1971 in the same appendix describes an electrode structure of about 100 grams that lifted about 125 grams at 170 kV, "actually self-levitating". The three pages now give the 1968 answer as the letter puts it.
The Younger Dryas Impact bench quoted Petaev et al. as noting that their platinum ingrowth "significantly exceeds the expected lifetime of about 5 years of fine dust". Those words are not in their 2013 paper, which says the decay of the signal is "consistent with the ∼5-y lifetime of dust in the stratosphere"; the point is made in their reply to Boslough later that year, which gives an ingrowth "significantly exceeding the expected lifetime (∼5 y) of fine dust in the atmosphere". The bench also quoted LeCompte et al. 2012 as saying "the peaks" are "not reliably correlated with the onset of the Younger Dryas", as if of the spherules. The paper says it of magnetic grains: "peaks in the total magnetic grain density at the three sites are not reliably correlated with the peaks in spherules or the onset of the Younger Dryas." Both quotations now follow the papers.
The Crust Displacement bench said that Hapgood, "shown the arithmetic, conceded it" and withdrew the ice engine. The bench's own sources give the reason his 1970 preface does: the gravity surveys of the Geophysical Year, showing the Antarctic cap largely compensated. The card now says only that he dropped the engine himself.
The Tiamat bench said the asteroid belt weighs 1/2,500 of the lower half its story needs "and the book concedes it". The book does not give that figure: on p. 95 it says the debris "is insufficient to account for even one whole planet, to say nothing of two". The 1/2,500 is the bench's own arithmetic from Pitjeva 2018. The bench now credits the book only with calling the debris insufficient.
The Electrogravitics bench's card on the disc body said "The 1950 carousel discs were 1.5 and 3 feet across". In LaViolette's Secrets of Antigravity Propulsion, the 1950 carousel used 1.5-foot discs (p. 43); the 3-foot pair belongs to Brown's second Pearl Harbor demonstration, around 1953 or 1954 (p. 50). The card now says "the carousel discs", as the book does on p. 83.
The same bench's card on the end of the open literature began "By 1959 and 1960 the subject leaves the open literature" and counted "Bahnson's fatal crash" among the signs. Agnew Bahnson filed his electrogravitic patent in September 1964 and was killed shortly after (pp. 109-110). The card now says "From 1959 on" and places the crash after the 1964 patent.
The Electrogravitics bench marked its page references to LaViolette's book as approximate, and several were off, among them the voltage exponents (p. 48, not "approx. p. 53"), Cady's measurements (pp. 47-48, not "approx. p. 52"), the 1960 Electrokinetics report (pp. 89-92, not "approx. pp. 93-95") and Talley's experiment (pp. 359-360, not "approx. p. 356"). Every card now gives the printed page, read from the book's running heads. The Wave Schedule bench's two Vallée page references were right and no longer say "approx."
The Rendlesham Light bench quoted Halt's tape as "There's a round abrasion on the tree about three and a half, four inches in diameter... it looks like it might be old... there's some sap marks or something on it." The bench takes its tape quotations from Ian Ridpath's transcript, which does not contain those words. At that point the transcript has Englund saying "It looks like an abrasion pointing into the centre of the..." and, after Halt's "It is.", "...landing area"; Halt then says "It may be old, though" and "there's some sap marks or something on it." The card now quotes that passage.
The Nuclear Lure bench said Medina, Brewer and Kirkpatrick (2023) found that report density tracks "proximity to military installations", in about 98,000 civilian reports, and that ordinary objects plus opportunity "reproduce the geography". The paper measured neither distance nor density: it models each county's rate of reports per resident against, among other things, the share of the county's area inside military installations, and finds a small positive relationship (1.3 per cent per standard deviation). Its data are "over 98,000 publicly reported UAP sightings", and it reads its results as people seeing more where they have more opportunity to, not as reproducing the geography. The bench now says so throughout.
14 September 2026
The Project Orion page filed "the ~20 m steel pusher plate" under the established record in its accuracy table. The comparison table in the page's own first source gives the 4,000-ton interplanetary design a 40 m plate and the 880-ton orbital test article 25 m; 20 m is the bench's drawing, not the record's, and the accuracy table now says so in a stylised row of its own. The same page, the two panels and the interactive's part cards and story said the plasma strikes the plate for a few microseconds. By the model's own impulse of 48 MN·s at the cited 10,000 g the strike lasts about a millisecond, and the plasma's flight across the 28 m standoff alone takes most of a millisecond; the source's microseconds refer to the pulse unit's channel filler. Every place now says about a millisecond.
The Wormhole bench printed two readouts in the scene's drawing units while labelling them in throat radii r₀, so the distance crossed and the throat circumference were right only at a throat radius of 1.00 and wrong at every other setting (at r₀ = 2.40 the crossing read 6.0 r₀ where it is 2.5), and the negative-energy readout printed the 1/r₀² factor twice. All three now read in throat radii, and the energy line names the factor once.
The Astrophage bench said three of the book's numbers "lock into one to 0.11%". By the bench's own figures only two do: the thermal proton at 96.415 °C and the 25.984 µm photon agree to 0.11%, and the neutrino at one twenty-billionth of a proton lands about 1.7% off them, which the page's claims row had said all along. The tank card said the fuel held about 900 years of world energy use; at the 20 TW the bench then used that was about 285 years, and at the 19.0 TW every bench now shares it is about 300. The clock spread printed as ×430 is ×440 by the bench's own clocks, and "a 1.08° disk" for the Sun seen from the apex was its angular radius: the disc is about 2.2° across.
The Casimir bench miscounted its own fourth-power law in three illustrations: 10 µm to 100 nm is two decades of separation and eight decades of pressure, not four and sixteen; the pressure at 10 µm on a square centimetre is about 13 piconewtons, not femtonewtons; and a milli-pascal over a square metre weighs about 130 milligrams, some five grains of rice, not a few flecks of dust. The plotted figures were right; the words around them were not.