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.
14 September 2026 (continued)
A reader following two citations on this site reached a different paper from the one named. The Tabby's Star bench listed the Schuetz optical SETI search under arXiv:1602.00987, which is in fact the VERITAS optical-flash search by Abeysekara and colleagues; the Schuetz paper is arXiv:1512.02388, matching the DOI the entry already carried. The same bench gave the Hippke plate-archive paper a title it does not carry, and the Gimbal bench did the same to the Peings and von Rennenkampff reconstruction. All three entries now carry the identifier and title the paper is actually published under.
The Drake Equation bench seated its expected civilisations in an annulus of its own, 13,000 to 33,000 light-years, while citing Lineweaver, Fenner and Gibson 2004 for the galactic habitable zone. Their published zone is the 7 to 9 kiloparsec annulus, about 22,800 to 29,400 light-years, 2.7 times smaller in area, and the bench's nearest-neighbour readout followed the wider band: the accuracy table read "the 13 to 33 kly annulus is one published reading of the galactic habitable zone" and How it works said the Green Bank values come out "about eight hundred light-years, which makes the round-trip readout near seventeen hundred years". The bench now seats civilisations in Lineweaver's published 7 to 9 kiloparsecs, and the same values read about five hundred light-years and just over a thousand years.
The ECM range-fouler bench described spot noise as putting all of its power inside the radar receiver. Its own model does not: a spot jammer parked on the frequency gets min(1, B/2 MHz), which for the bench's 1 MHz receiver is a half, three decibels thrown away, against barrage's 0.0025 and twenty-six decibels. The page read "spot noise parked on the frequency keeps all of it" and "the same pod, parked on the radar's frequency, with every watt in band". It now says half its power in band, names the 2 MHz spot against the 1 MHz receiver, and gives the three decibels against barrage's twenty-six.
On The 2020 wave, the briefing described the set as "Eighteen separate clips, separate dates, spread across a year", and scoped it as "DOW-UAP-PR055, PR059 and PR077 to PR095". The catalogue does not support either. PR077 and PR078 are two parts of the same 2 November 2020 track, PR088 and PR089 are two parts of the same 31 August track, and PR093 and PR095 carry the same 5 May 2020 title, so the eighteen clips resolve to about fifteen dated encounters. The PR077 to PR095 run also contains PR081 and PR082, which AARO places in the Africa Command area, and PR086, a 2019 Northern Command clip, none of which belong to a Central Command 2020 wave. The briefing now lists the eighteen Central Command clips by identifier, names the three exclusions, and says the set resolves to about fifteen dated encounters.
The Wave Schedule said of July 1952 that "one month carries an eighth of the entire 23-year file", and its scale dial said "linear shows why one month is an eighth of the whole file". July 1952 holds 536 of the file's 12,618 reports, which is 4.2 per cent, about one report in twenty-three. The year 1952 is the eighth. The cards now give the month its 4.2 per cent and the dial line speaks of the year.
The same bench listed its schedule anchors as "0 for a fixed interval, 0.29 for Skinner's variable interval, 1 for Poisson". The 0.29 rail is an interval drawn uniformly within 50 per cent of its mean, a construction of this bench; Ferster and Skinner published a schedule taxonomy, not that distribution. The anchor is unchanged and is now labelled as the bench's own draw.
On A Bogey at Ten O'Clock High, the story chapter "The shed" said that at half a metre per second the spent stage "recedes at a walking pace and is still close enough to read serial numbers off an orbit later". The bench's own drift ledger prices the same departure at 16.1 km per orbit for every metre per second shed along-track, which puts the stage 8.0 km away after one orbit. The line now reads "is only eight kilometres behind an orbit later".
On The Laser Sail, the equation printed in section 03 and in the accuracy table read "d(γβ)/dt = (2ηP / mc) · (1−β)/(1+β) · min(1, (D_sail / D_spot)²)". The right side of that expression is an acceleration and the left side is a rate per second, so the two cannot be equal. The instrument's own integrator was never wrong: it sets a₀ = 2ηP/(mc) and then adds a/c per step, which is 2ηP/(mc²). The page now prints 2ηP/mc². No readout or curve changes.
On The Younger Dryas Impact Ledger, the platinum section said of the bench's chain: "It is the same calculation they did, which is why their own paper concludes that a crater of a few kilometres ought to exist, and notes that none has been found." The platinum half of that is right and the crater half is not. The bench recovers an iron body of about 830 metres, against the roughly 0.8 kilometres Petaev et al. state, but the Collins scaling it then applies puts the crater owed by that body at about seventeen kilometres, several times the "crater of a few kilometers in diameter" their paper expects. The page now says the platinum chain agrees with theirs and the crater does not, and that both readings agree no crater of that age has been found.
The guided story on Electrogravitics stated, in the site's own voice, that "In hard vacuum, electrons accelerated across it reach 82 percent of the speed of light", under a chapter titled "Eighty-two percent of the speed of light". That is the book's figure, and it is the one figure the bench exists to correct: at 150 kilovolts the relativistic value is 0.634 c. The chapter now attributes the 82 percent to the book and prints the bench's 63 percent beside it, and the Jar caption does the same.
The lede on Project Stargate read "The claimed result, by the end, was a hit rate near 32." That hit rate, 32.2 percent from 106 hits in 329 sessions, is Bem and Honorton's 1994 autoganzfeld series, run at a private laboratory outside the government programme. The programme's own experiments were reported as effect sizes, near 0.2 across all remote viewing at SRI and SAIC. The lede, both catalogue cards and the view description now give each figure its own source. Amended 3 October 2026: this entry first gave the 0.2 to the experienced viewers; Utts' Table 3 gives it for all remote viewing at SRI and SAIC, and her experienced SRI viewers come to 0.385.
A claims row on The Podkletnov Impulse credited the rounding of Table 1's 142 millimetres to 14 centimetres as "Jorjani-style rounding to 14 cm at LaViolette p. 169". Jorjani is not an author of the book under audit; the rounding is LaViolette's own. The row now reads "rounded to 14 cm by LaViolette, p. 169", which is what the Chinese row already said.
13 September 2026
The Greece footage briefing described PR28 against AARO's catalogue and quoted an unpublished internal list. Rewritten from AARO's catalogue and mission reports and from the clip frame by frame: the two panels are electro-optical on the right and shortwave infrared on the left, not "the same scene in two infrared polarities"; from about 50 seconds the object trails a faint tail (AARO's 00:55 line: "inverted teardrop with a vertically linear trailing mass suspended below"), so it does not "simply sit there as a bright dot"; the subject is lost at the 00:56 switch to visible light and the last seconds are black-hot cloud; "fast-moving" became the report's 434 knots; the internal-list quotation is gone, replaced by AARO's own words, with a note that AARO pairs the clip with report D7 while the diamond, 434-knot, SWIR-only wording is D25's. In Reading infrared footage the "eight-pointed star" attributed to the press now quotes AARO's own video description, the timeline follows the clip, and a paragraph notes that the catalogue's video title says "Resolved as an Aircraft" where the record's title says "Unresolved".
Seven briefings ran the Western US Event of October 2023 and the helicopter orb sighting of late 2025 together as one event: Western US Event said "a serving intelligence official saw the other half of the event from a helicopter", The intelligence official said this was the event of Release 01 Briefing 2, and FBI Photo B series called itself, in seven places, the photo evidence behind "the Western US Event of 2025". Per the catalogue, the 2023 event's set is D077 to D083, FBI D014 to D023 and PR005 and PR006; ODNI-UAP-D001 is a separate late-2025 case, which the catalogue pairs with the FBI statement and Photo A and B as "this same set of observations, taken through night vision devices by ground-based personnel". All seven, EN and ZH, now keep the two apart; the casefile and Reading infrared footage carried the same slip and were fixed with them.
Several Release 01 video and table briefings matched clips to the wrong files: in UAE cluster PR026 and PR027 both belong to report D23, not D27; in Greece cluster PR029 was credited to D8 while the pole-and-bar description (with the observer's own "POSS REFLECTION FROM THE OBJECT IN THE WATER") is D27's, dated 7 June 2024 in the Gulf of Oman, and PR035 gains its report D35. Short clips run two to fourteen seconds, not five; nine of the eleven sit in Central Command with one each in Africa Command and Indo-Pacific Command, not "all eleven"; seven, not all, come with a mission report; the years are 2020 to 2025. In Range fouler system D42's "Japan, 2023" follows the form: dated 31 August 2020, grid square 39RWL, USCENTCOM declassification number 26-0028.