--- slug: the-mirror-that-stops-it passed: 2026-10-18 by: the assay line (research/assay-line/queue-1018/), Codex fixer directed by claude-assaying-boyle-1014; source record by Codex, checked by Codex checked: 180 confirmed: 161 wrong: 18 unverifiable: 0 observed: 0 stale: 1 --- This pass decided all 19 queue-1018 findings before editing, re-read their evidence, and repaired the supported claims in this layer. Its starting layer files matched d9d5d37097; the 161 CONFIRMED lines below carry forward the Codex checker's source readings at 20ca47ceb1, with their evidence and attribution, rather than claiming a new independent reading of every source. MINOR findings are counted in wrong. The archive quotation remains a declined MINOR because its original is unavailable. The film source is repaired and the page discloses the error, but the existing MP4 still says 972 and remains unfinished under the prohibition on media rendering. The authored dek is corrected; its two prohibited shared search copies await the normal rebuild. The requested pass and correction date is 2026-10-18; the live HTTP recheck occurred on 2026-10-19. ## Claims - MINOR | Every figure on screen is read out of this page's own data at render time | git show 20ca47ceb1:research/the-mirror-that-stops-it/film/film.html:248 | [X1; film-figures-data-driven] Read film/film.html:248: "Over 972 settings the largest answer is 1.3e-15." is literal prose, although lines 38-40 fetch model, blur and traces. The caption overstates how its figures are checked and supplied. Under the explicit rule for similar checking overclaims, change WRONG to MINOR and describe the mixture of data and prose. The false mirror-test count itself is separately WRONG under X14/X19. - WRONG | the total, unsigned energy is untouched. | git show 20ca47ceb1:public/strata/the-mirror-that-stops-it/data/model.json | [X2; unsigned-energy-unchanged] Read data/model.json modes.forward and modes.balanced: forward has "right":1898888776.221533 and "left":1906495056.5317829; balanced has "right":988720651.8786422 and "left":988720651.8786423. Their sums are 3805383832.753316 and 1977441303.7572846. "Untouched" says the total is unchanged, which is false, so upgrade MINOR to WRONG. Say it remains nonzero; do not claim conservation of total energy. - MINOR | recomputes the whole curve and checks the crossing. | git show 20ca47ceb1:verify-the-mirror-that-stops-it.mjs:182 | [X3; whole-curve-recomputed] Read verify-the-mirror-that-stops-it.mjs:182: "i += 8" recomputes six of the 41 blurCurve points; the following loop derives the crossing from the committed points. This is an overclaim about the page's own checking, explicitly MINOR under the task rule, not a reason to expand an expensive computation. State sampled recomputation and the stored-curve crossing, and make the verifier's message equally precise. - MINOR | The resulting images visually follow much of the requested brief | git show 20ca47ceb1:research/the-mirror-that-stops-it/README.md:21 ; git show 20ca47ceb1:research/the-mirror-that-stops-it/extract-traces.mjs:6 | [X4; archive-quote-discrepancy] Read README.md:21 and extract-traces.mjs:6-8, which retain "The resulting images visually follow much of the requested grammar"; index.html:309 instead has "requested brief". The original private guide is not supplied, so these copies establish a discrepancy but do not establish which preserves the original. A supported restoration is not available cheaply. Leave all quoted words unchanged and record this MINOR as [declined]; do not invent a restoration. - WRONG | taking the largest autocorrelation over a range always returns the smallest lag | git show 20ca47ceb1:public/strata/the-mirror-that-stops-it/drift.mjs:481 ; recreated: node audit-rendered.mjs | [X5; autocorrelation-smallest-lag] Read drift.mjs:481-510, including "Circular autocorrelation of a centred array at integer lag." Re-ran that autocorr on a 256-sample period-16 sine: lag 6 is -0.7071067811865483 and lag 16 is 1.0000000000000002, so the maximum over lags 6-31 cannot be at the smallest lag. The universal mathematical assertion is false, so upgrade MINOR to WRONG. Retain the recorded bug but say a range maximum can select its smallest lag, rather than always doing so; correct the page, notebook and module comment. - MINOR | An artwork can therefore only be read against controls at its own size, and the position column below does that. | git show 20ca47ceb1:public/strata/the-mirror-that-stops-it/index.html:744 ; git show 20ca47ceb1:public/strata/the-mirror-that-stops-it/data/calibration.json | [X6; exact-size-controls] Read index.html:744-749: "const d = Math.abs(s - size)" selects the nearest available size, not an exact match. The original stationary-drift record is 720 by 540, while calibration.json has native controls at 2400, 1800, 1254, 1100, 800 and 550; that row uses 800. The other nine match. This is a cheap precision fix: describe nearest-size controls and disclose the 720-to-800 exception in the page and notebook. - WRONG | Reducing in linear light and reducing on the encoded eight-bit values agree to three decimal places | git show 20ca47ceb1:public/strata/the-mirror-that-stops-it/data/calibration.json | [X7; resampling-three-decimals] Read calibration.json rows balanced-2400/2-linear and balanced-2400/2-encoded: "asymmetry":0.09196 and "asymmetry":0.09482. These round to 0.092 and 0.095. Recomputed all six paired differences: the largest is 0.0028600000000000014. Read verifier:328, which accepts "worstSpace < 0.005" while claiming three-decimal agreement. State the observed maximum gap, correct the associated no-visible-difference inference, and bind the verifier to that measured maximum. - MINOR | Measured at native resolution, against controls at matching size. | git show 20ca47ceb1:public/strata/the-mirror-that-stops-it/index.html:744 ; git show 20ca47ceb1:public/strata/the-mirror-that-stops-it/data/calibration.json | [X8; exact-size-controls] Read the same nearest-size bracket in index.html:744-749 and the stationary-drift original dimensions in traces.json: 720 by 540 is bracketed by the 800-pixel controls. Change "matching size" to "nearest available size" and point out this single exception. Same error as X6 and X15. - WRONG | Presence at 0.46 overlaps none of them. | git show 20ca47ceb1:public/strata/the-mirror-that-stops-it/data/traces.json ; git show 20ca47ceb1:public/strata/the-mirror-that-stops-it/data/calibration.json | [X9; presence-interval-separation] Read the original records in traces.json: Presence has "ci95":[0.18528,0.22182] and Undertow has "ci95":[0.21171,0.37]. The raw intervals overlap. Applying the page's calibration formula with the 1254 and 2400 controls gives [0.3923442961,0.5462163642] and [0.2429211892,1.1766753185], also overlapping. Correct the non-overlap claim and its inference; distinguish a lower point estimate from interval separation. Same error as X18. - WRONG | It is served here at its native 1800 pixels, unlike everything else on the page | git show 20ca47ceb1:public/strata/the-mirror-that-stops-it/data/images.json | [X10; native-image-exclusivity] Read images.json entries stationary-drift and depth-lattice. Both say "resample":"none (native size)"; stationary-drift has sourceSize and servedSize [720,540], while depth-lattice has [1800,1800]. Remove the false exclusivity and name the other native image. - MINOR | Every number on this page is recomputed from scratch by one program | git show 20ca47ceb1:verify-the-mirror-that-stops-it.mjs:334 | [X11; all-numbers-recomputed] Read verify-the-mirror-that-stops-it.mjs:334: "const cc = findings.colourCost" compares stored original and archiveStated measurements rather than decoding the originals. The model and blur loops also use "i += 37" and "i += 8". This exactly matches the explicit MINOR rule for every-number checking overclaims. Describe sampled recomputation, summary reconstruction and comparisons with stored measurements, and narrow the verifier header's matching claim. - STALE | The -L is load-bearing: this site answers the directory form of a page with a redirect, and without it you get a file of length zero and a confusing failure. | git show 20ca47ceb1:memory/log.d/2026-09-04T2230Z-the-source-mirror.md ; https://artwaste.land/strata/the-mirror-that-stops-it/ | [X12; directory-redirect-required] Read the 2026-09-04 source-mirror log: the "index.html spelling" returned "307". That supports the historical redirect advice, not a claim that the directory URL must always redirect. A fresh curl without -L on 2026-10-19 returned HTTP 200, 77968 bytes and zero redirects, matching N2. Make -L conditional on redirects being present. Group all STALE items into ONE Updated 2026-10-18 line (this is the only STALE item), date the retained September test and do not claim its index.html result was a measurement of this directory URL. - MINOR | rebuilding the whole thing from nothing is five commands: | git show 20ca47ceb1:public/strata/the-mirror-that-stops-it/index.html:420 | [X13; rebuild-command-count] Read index.html:422-427: the block has six node commands, ending with "blur-colour.mjs". Change five to six. Retain the following caveat that original archive measurements require the private files. - WRONG | across 972 parameter settings the largest departure from zero is 1.3e-15 | git show 20ca47ceb1:public/strata/the-mirror-that-stops-it/data/model.json | [X14; mirror-sweep-count] Read model.json exact: "checked":49 and "maxBalanced":1.3310186133423186e-15. The grid has 972 forward-profile rows; model-results.mjs:117-127 only adds a reversed and balanced test when "i % 20 === 0". The 1.3e-15 maximum belongs to those 49 checks. Correct the dek and any prose conflating the two sweeps. Share one correction and error label with X19. - MINOR | placed against controls at matching size | git show 20ca47ceb1:public/strata/the-mirror-that-stops-it/index.html:744 ; git show 20ca47ceb1:public/strata/the-mirror-that-stops-it/data/calibration.json | [X15; exact-size-controls] Read the markdown claim and index.html:744-749 nearest-size selector. traces.json gives stationary-drift width 720 and calibration.json selects 800 as the nearest available native control; the other nine exact sizes are present. Use nearest available size in the markdown too. Same error as X6/X8. - WRONG | (added) Peripheral Drift Illusion: Why the Order of the Four Greys Is the Whole Effect | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC10989047/fullTextXML | [X16; order-whole-effect] Fetched and read Europe PMC fullTextXML for PMC10989047. Its abstract says "the presence of colour affects the magnitude of illusion" after comparing colour variants with equated luminances; its introduction assigns the ordering a role in direction. The title's "Whole Effect" excludes that measured colour contribution, a false claim about the effect rather than only metadata precision. Upgrade MINOR to WRONG and narrow the title in both authoring locations to how order sets direction. - WRONG | (added) here is the whole thing from an empty directory. | git show 20ca47ceb1:public/strata/the-mirror-that-stops-it/index.html:400 ; git show 20ca47ceb1:verify-the-mirror-that-stops-it.mjs:60 | [X17; reader-recipe-missing-blur] Read index.html:410: "for f in model traces calibration images findings; do" omits blur. The verifier unconditionally executes "const blur = read('blur.json')" at line 60, and the retained source-mirror closure includes that file. Add blur to the fetch loop and run the complete ten-file recipe in an empty directory. - WRONG | (added) Only one of those two is a separation the intervals will carry | git show 20ca47ceb1:public/strata/the-mirror-that-stops-it/data/traces.json ; git show 20ca47ceb1:public/strata/the-mirror-that-stops-it/data/calibration.json | [X18; presence-interval-separation] Re-read the original Presence and Undertow rows in traces.json: "ci95":[0.18528,0.22182] and "ci95":[0.21171,0.37]. Recomputed the same affine calibration used by the page: their position intervals overlap at [0.3923442961,0.5462163642] within [0.2429211892,1.1766753185]. Neither generator-made image has an interval disjoint from every program-made image. Remove the claimed separation, sharing the correction with X9. - WRONG | (added) Over 972 settings the largest answer is 1.3e-15. | git show 20ca47ceb1:research/the-mirror-that-stops-it/film/film.html:248 ; git show 20ca47ceb1:public/strata/the-mirror-that-stops-it/data/model.json | [X19; mirror-sweep-count] Read film/film.html:248: "Over 972 settings the largest answer is 1.3e-15." Read model.json exact: "checked":49. The film-facts gate wrongly tests model.grid.settings for this count. Fix the film source to interpolate model.exact.checked and maxBalanced, and correct that assertion. Share the correction with X14 and explicitly tell readers that the existing video still displays 972. The user prohibits rendering or re-encoding media, so the existing MP4 will remain NOT FIXED and must be listed under not done; no media rebuild is authorized. - CONFIRMED | A picture that never changes can still look like it is moving. | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC10989047/fullTextXML | [intro-still] Uesaki et al.: "The order in which those areas are arranged within each micropattern dictates the direction of illusory rotational motion". The same paper separately finds colour changes magnitude, so this supports direction at fixed levels, not an assertion that colour never matters. The model and hashing clauses were checked against drift.mjs and index.html. PMC initially returned a challenge to curl; the complete article was read through Europe PMC. - CONFIRMED | What carries the direction is not the contrast and not the colour: it is the order of four luminance levels around a repeating unit. | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC10989047/fullTextXML | [intro-direction] Uesaki et al.: "The order in which those areas are arranged within each micropattern dictates the direction of illusory rotational motion". The same paper separately finds colour changes magnitude, so this supports direction at fixed levels, not an assertion that colour never matters. The model and hashing clauses were checked against drift.mjs and index.html. PMC initially returned a challenge to curl; the complete article was read through Europe PMC. - CONFIRMED | Reverse that order and the same four greys, in exactly the same amounts, drift the other way. | https://michaelbach.de/ot/mot-snakesLum/ | [intro-reversal] Bach: "If this sequence is reversed (between wheels) using the checkbox Alternate, the seeming motion changes its direction." For symmetry: "When the asymmetric sequence is replaced by a symmetric one" the rotation vanishes; his correction retains shimmer. The page implements the separate profile and hashing operations. - CONFIRMED | Give the unit a mirror axis and a whole family of motion models is forced to predict not a small answer but exactly zero. | git show 20ca47ceb1:public/strata/the-mirror-that-stops-it/drift.mjs | [intro-zero] drift.mjs states "R depends on x ONLY through L(x)" and computes "const opponent = right - left;". Reflection exchanges the even/odd filter combinations; a reflection-invariant input has equal integrated directional energies. The reader run independently reproduced reversal and zero over 14 settings. - CONFIRMED | Below is a field of curved bands built from four luminance levels. | git show 20ca47ceb1:public/strata/the-mirror-that-stops-it/drift.mjs | [p003-01] The code defines "LEVELS = Object.freeze([0.012, 0.16, 0.91, 0.55])", reverses the sampled profile with "out.reverse()", and balances with "const m = (LEVELS[1] + LEVELS[3]) / 2". fieldSample keeps the band geometry; regionChroma/nullChroma provide distinct colour offsets. Exact symmetry and equal histograms refer to the nominal repeating luminance profile, not to quantized shaded pixels. Perceived disappearance is not measured here. - CONFIRMED | Nothing in the file is creeping: the canvas is drawn once per setting and never touched again, and the page hashes the pixels it has just drawn, four times a second, and prints the digest. | git show 20ca47ceb1:public/strata/the-mirror-that-stops-it/index.html | [p003-04] index.html calls "setInterval(digest, 250)"; draw writes pixels on setting/fullscreen changes, while digest reads getImageData and calls "crypto.subtle.digest('SHA-256', d)". No periodic redraw is scheduled. The zeros in the source are initial counters, not a measurement after interaction. - CONFIRMED | frames drawn since you arrived: 0 · pixel digests taken: 0 · digests that differed: 0 SHA-256 of the canvas taking the first reading | git show 20ca47ceb1:public/strata/the-mirror-that-stops-it/index.html | [p004-01] index.html calls "setInterval(digest, 250)"; draw writes pixels on setting/fullscreen changes, while digest reads getImageData and calls "crypto.subtle.digest('SHA-256', d)". No periodic redraw is scheduled. The zeros in the source are initial counters, not a measurement after interaction. - CONFIRMED | Three orders, one geometry. | git show 20ca47ceb1:public/strata/the-mirror-that-stops-it/drift.mjs | [p006-01] The code defines "LEVELS = Object.freeze([0.012, 0.16, 0.91, 0.55])", reverses the sampled profile with "out.reverse()", and balances with "const m = (LEVELS[1] + LEVELS[3]) / 2". fieldSample keeps the band geometry; regionChroma/nullChroma provide distinct colour offsets. Exact symmetry and equal histograms refer to the nominal repeating luminance profile, not to quantized shaded pixels. Perceived disappearance is not measured here. - CONFIRMED | Built order is the recipe as the archive's own programs lay it down. | git show 20ca47ceb1:public/strata/the-mirror-that-stops-it/drift.mjs | [p006-02] The code defines "LEVELS = Object.freeze([0.012, 0.16, 0.91, 0.55])", reverses the sampled profile with "out.reverse()", and balances with "const m = (LEVELS[1] + LEVELS[3]) / 2". fieldSample keeps the band geometry; regionChroma/nullChroma provide distinct colour offsets. Exact symmetry and equal histograms refer to the nominal repeating luminance profile, not to quantized shaded pixels. Perceived disappearance is not measured here. - CONFIRMED | Reversed is the same period read backwards, which is an exact spatial mirror of the whole thing, so its histogram is identical bin for bin: the same four greys, in the same amounts, in the same shapes. | git show 20ca47ceb1:public/strata/the-mirror-that-stops-it/drift.mjs | [p006-03] The code defines "LEVELS = Object.freeze([0.012, 0.16, 0.91, 0.55])", reverses the sampled profile with "out.reverse()", and balances with "const m = (LEVELS[1] + LEVELS[3]) / 2". fieldSample keeps the band geometry; regionChroma/nullChroma provide distinct colour offsets. Exact symmetry and equal histograms refer to the nominal repeating luminance profile, not to quantized shaded pixels. Perceived disappearance is not measured here. - CONFIRMED | Mirrored replaces the two intermediate greys with their common mean, which gives the unit a mirror axis while keeping both extrema, all four widths and the mean luminance. | git show 20ca47ceb1:public/strata/the-mirror-that-stops-it/drift.mjs | [p006-04] The code defines "LEVELS = Object.freeze([0.012, 0.16, 0.91, 0.55])", reverses the sampled profile with "out.reverse()", and balances with "const m = (LEVELS[1] + LEVELS[3]) / 2". fieldSample keeps the band geometry; regionChroma/nullChroma provide distinct colour offsets. Exact symmetry and equal histograms refer to the nominal repeating luminance profile, not to quantized shaded pixels. Perceived disappearance is not measured here. - CONFIRMED | Switch to MIRRORED and then to greyscale, and the symmetry becomes plain to look at: each unit reads the same forwards and backwards about its black sliver. | git show 20ca47ceb1:public/strata/the-mirror-that-stops-it/drift.mjs | [p007-01] The code defines "LEVELS = Object.freeze([0.012, 0.16, 0.91, 0.55])", reverses the sampled profile with "out.reverse()", and balances with "const m = (LEVELS[1] + LEVELS[3]) / 2". fieldSample keeps the band geometry; regionChroma/nullChroma provide distinct colour offsets. Exact symmetry and equal histograms refer to the nominal repeating luminance profile, not to quantized shaded pixels. Perceived disappearance is not measured here. - CONFIRMED | Switch the colour back on and it does not, because the violet and the gold are two different colours sitting at one luminance. | git show 20ca47ceb1:public/strata/the-mirror-that-stops-it/drift.mjs | [p007-02] The code defines "LEVELS = Object.freeze([0.012, 0.16, 0.91, 0.55])", reverses the sampled profile with "out.reverse()", and balances with "const m = (LEVELS[1] + LEVELS[3]) / 2". fieldSample keeps the band geometry; regionChroma/nullChroma provide distinct colour offsets. Exact symmetry and equal histograms refer to the nominal repeating luminance profile, not to quantized shaded pixels. Perceived disappearance is not measured here. - CONFIRMED | The theorem is about luminance and only luminance, and the chroma offsets are built to carry none, which is section 5's whole subject. | git show 20ca47ceb1:public/strata/the-mirror-that-stops-it/drift.mjs | [p007-04] The code defines "LEVELS = Object.freeze([0.012, 0.16, 0.91, 0.55])", reverses the sampled profile with "out.reverse()", and balances with "const m = (LEVELS[1] + LEVELS[3]) / 2". fieldSample keeps the band geometry; regionChroma/nullChroma provide distinct colour offsets. Exact symmetry and equal histograms refer to the nominal repeating luminance profile, not to quantized shaded pixels. Perceived disappearance is not measured here. - CONFIRMED | The mirrored field is exactly symmetric in the one quantity the argument uses, and visibly asymmetric in a quantity the argument never mentions, and the drift still goes. | git show 20ca47ceb1:public/strata/the-mirror-that-stops-it/drift.mjs | [p007-05] The code defines "LEVELS = Object.freeze([0.012, 0.16, 0.91, 0.55])", reverses the sampled profile with "out.reverse()", and balances with "const m = (LEVELS[1] + LEVELS[3]) / 2". fieldSample keeps the band geometry; regionChroma/nullChroma provide distinct colour offsets. Exact symmetry and equal histograms refer to the nominal repeating luminance profile, not to quantized shaded pixels. Perceived disappearance is not measured here. - CONFIRMED | The thing above wants to be still while you look at it, and a film playing beside it would be the one moving object on the page. | git show 20ca47ceb1:public/strata/the-mirror-that-stops-it/index.html | [p009-02] The stage is static between control operations and the video is the page's time-varying image. index.html contains " 2·S·f. Every claim proven live, and the same fold made audible by genuinely under-sampling a 1000 Hz tone down to the 500 Hz you hear. A playable instrument on temporal aliasing, honest about the one part that is still unsettled: the contested continuous-light version." - CONFIRMED | Both are a direction read out of a spatial pattern by a mechanism with a time constant, and in both the direction flips when nothing changes but a ratio: there the ratio of spoke frequency to frame rate, here the amount of blur. | git show 20ca47ceb1:src/content/strata/the-wheel-that-spins-backward.md | [related-the-wheel-that-spins-backward-3] Read the related layer at the pin. Its description: dek: "In old films a stagecoach races forward and its wheels turn slowly backward. No trick, no editing. The camera reports the literal truth about each frame it caught, and that truth is a lie about the motion. Set a wheel's real spin, its spokes, and a frame rate, and watch the reversal happen; then watch why, as the spoke frequency folds into the band a frame rate can hold. The wheel freezes exactly when its spin is a whole multiple of fs/S, runs backward just below each freeze, and is captured faithfully only above the Nyquist rate fs > 2·S·f. Every claim proven live, and the same fold made audible by genuinely under-sampling a 1000 Hz tone down to the 500 Hz you hear. A playable instrument on temporal aliasing, honest about the one part that is still unsettled: the contested continuous-light version." - CONFIRMED | The precedent for the two rulers that died here, and it is unusually close. | git show 20ca47ceb1:src/content/strata/the-wedge-the-scanner-never-saw.md | [related-the-wedge-the-scanner-never-saw-1] Read the related layer at the pin. Its description: dek: "A CT scan that cannot go all the way round does not lose quality, it loses directions. This page takes a real walnut measured in Helsinki, computes what fraction of its boundary a missing wedge erases at each width against a prediction registered before the run, and prints beside it the value a featureless object would have scored. Three of the ten registered predictions came out wrong, a fourth could not be scored on this instrument at all, and the worst of them was undone by the page's own control rather than by the walnut." - CONFIRMED | There, three of ten registered predictions came out wrong and the worst was undone by the page's own control rather than by the specimen. | git show 20ca47ceb1:src/content/strata/the-wedge-the-scanner-never-saw.md | [related-the-wedge-the-scanner-never-saw-2] Read the related layer at the pin. Its description: dek: "A CT scan that cannot go all the way round does not lose quality, it loses directions. This page takes a real walnut measured in Helsinki, computes what fraction of its boundary a missing wedge erases at each width against a prediction registered before the run, and prints beside it the value a featureless object would have scored. Three of the ten registered predictions came out wrong, a fourth could not be scored on this instrument at all, and the worst of them was undone by the page's own control rather than by the walnut." - CONFIRMED | What resampling costs, measured rather than asserted. | git show 20ca47ceb1:src/content/strata/slanted-edge-mtf.md | [related-slanted-edge-mtf-1] Read the related layer at the pin. Its description: dek: "Photograph a straight edge tilted a few degrees and this page hands back your own imaging chain's spatial frequency response: the frequency at which it has lost half the contrast, in cycles per pixel of the frame you were given, and in cycles per millimetre of paper once you have printed the target and measured the printed ruler with a ruler of your own. Before it will report anything about you, the same estimator has to recover four blurs whose true answers are closed forms and refuse six specimens it is supposed to refuse. A response that peaks above 1.000 is arithmetic no lens performs. If yours does, the page shows you the bands the sharpener left beside the edge in your own picture." - CONFIRMED | The four regions here have hard interior boundaries at 0.135 and 0.635 of a unit, and any rescale rounds them, so publishing these artworks at web size is a real loss. | git show 20ca47ceb1:public/strata/the-mirror-that-stops-it/drift.mjs ; git show 20ca47ceb1:research/the-mirror-that-stops-it/publish-images.py | [related-slanted-edge-mtf-2] The specific delivered-image reduction is LANCZOS in linear light. BOUNDS includes 0.135 and 0.635; the calibration quantifies the loss under these reductions. The statement is read in this resampling context, not as a theorem about every possible enlargement operator. - CONFIRMED | That layer is the instrument for seeing the damage in an image you were served; this one puts a number on it by rendering the same field at six sizes and watching both the ceiling and the floor of the measurement move. | git show 20ca47ceb1:src/content/strata/slanted-edge-mtf.md | [related-slanted-edge-mtf-3] Read the related layer at the pin. Its description: dek: "Photograph a straight edge tilted a few degrees and this page hands back your own imaging chain's spatial frequency response: the frequency at which it has lost half the contrast, in cycles per pixel of the frame you were given, and in cycles per millimetre of paper once you have printed the target and measured the printed ruler with a ruler of your own. Before it will report anything about you, the same estimator has to recover four blurs whose true answers are closed forms and refuse six specimens it is supposed to refuse. A response that peaks above 1.000 is arithmetic no lens performs. If yours does, the page shows you the bands the sharpener left beside the edge in your own picture." - CONFIRMED | The visual member that portal was missing. | git show 20ca47ceb1:src/content/strata/the-percept-the-world-never-sent.md | [related-the-percept-the-world-never-sent-1] Read the related layer at the pin. Its description: dek: "Why can't you tickle yourself, how do you hear a bass note that carries no energy at its own frequency, why does a Shepard tone climb forever and arrive nowhere? Three illusions across hearing and touch turn out to be one law: what you perceive is the residual between the world's signal and your brain's prediction of it, so the model can supply a percept the world never sent, or cancel one it did. Two add, one subtracts, and that they are the same operation with opposite sign is the thing no single piece shows. Drive all three by hand; every number recomputed live and re-checked offline, 31/31." - CONFIRMED | Its cases are hearing and touch: a bass note carrying no energy at its own frequency, a tone that climbs forever and arrives nowhere. | git show 20ca47ceb1:src/content/strata/the-percept-the-world-never-sent.md | [related-the-percept-the-world-never-sent-2] Read the related layer at the pin. Its description: dek: "Why can't you tickle yourself, how do you hear a bass note that carries no energy at its own frequency, why does a Shepard tone climb forever and arrive nowhere? Three illusions across hearing and touch turn out to be one law: what you perceive is the residual between the world's signal and your brain's prediction of it, so the model can supply a percept the world never sent, or cancel one it did. Two add, one subtracts, and that they are the same operation with opposite sign is the thing no single piece shows. Drive all three by hand; every number recomputed live and re-checked offline, 31/31." - CONFIRMED | A still image with no temporal frequency at all that is nonetheless read as motion is the same argument in sight, and it comes with a numeric grammar in the stimulus that says which way the supplied percept should run. | git show 20ca47ceb1:src/content/strata/the-percept-the-world-never-sent.md | [related-the-percept-the-world-never-sent-3] Read the related layer at the pin. Its description: dek: "Why can't you tickle yourself, how do you hear a bass note that carries no energy at its own frequency, why does a Shepard tone climb forever and arrive nowhere? Three illusions across hearing and touch turn out to be one law: what you perceive is the residual between the world's signal and your brain's prediction of it, so the model can supply a percept the world never sent, or cancel one it did. Two add, one subtracts, and that they are the same operation with opposite sign is the thing no single piece shows. Drive all three by hand; every number recomputed live and re-checked offline, 31/31." - CONFIRMED | The governing warning for a layer whose whole claim is about what an image looks like. | git show 20ca47ceb1:src/content/strata/the-engine-and-the-eye.md | [related-the-engine-and-the-eye-1] Read the related layer at the pin. Its description: dek: "A check can drive a page's arithmetic over every input it has, read the shipped HTML and match every sentence in it, and still never once look at what the page puts on the screen. On 6 August two tools were built here in one night with careful verifiers, green at 120 of 120 and 174 of 174, and a reviewer multiplied every displayed count on one by ten and reversed the shipped ranking rule on the other while both stayed green. This page hands you a working tool with five checks attached: break its arithmetic and they react, break only its display and four of five stay green while every number in front of you is wrong. Then the same question asked of every page on this site at once, by injecting a script that changes the last digit of every number a page renders, running the page's own checks against it, and controlling with a placebo one byte away. The population is watched rather than guessed, the instrument reports whether it actually changed anything before any green light is believed, and the result is an upper bound, because every number was corrupted at once." - CONFIRMED | A check can drive every line of a page's arithmetic, match every sentence in its shipped HTML, and never once look at the picture. | git show 20ca47ceb1:src/content/strata/the-engine-and-the-eye.md | [related-the-engine-and-the-eye-2] Read the related layer at the pin. Its description: dek: "A check can drive a page's arithmetic over every input it has, read the shipped HTML and match every sentence in it, and still never once look at what the page puts on the screen. On 6 August two tools were built here in one night with careful verifiers, green at 120 of 120 and 174 of 174, and a reviewer multiplied every displayed count on one by ten and reversed the shipped ranking rule on the other while both stayed green. This page hands you a working tool with five checks attached: break its arithmetic and they react, break only its display and four of five stay green while every number in front of you is wrong. Then the same question asked of every page on this site at once, by injecting a script that changes the last digit of every number a page renders, running the page's own checks against it, and controlling with a placebo one byte away. The population is watched rather than guessed, the instrument reports whether it actually changed anything before any green light is believed, and the result is an upper bound, because every number was corrupted at once." - CONFIRMED | The answer taken here is to have the check import the page's own module rather than a copy of it, and to anchor the image chain on the bytes of a served file by its hash. | git show 20ca47ceb1:verify-the-mirror-that-stops-it.mjs | [related-the-engine-and-the-eye-3] The verifier imports the served drift.mjs and checks the SHA-256 of stationary-drift.png against images.json. Both assertions passed on downloaded bytes. - CONFIRMED | The neighbouring beat, and the fringe-spacing law belongs to that layer, not this one. | git show 20ca47ceb1:src/content/strata/moire.md | [related-moire-1] Read the related layer at the pin. Its description: dek: "Lay one set of fine lines over another, turn it a single degree, and a huge slow pattern rolls across the page that is in neither set of lines: the moiré. It looks like a glitch. It is one of the most exact things in optics: the fringe spacing obeys D = p₁p₂ ⁄ √(p₁² + p₂² − 2p₁p₂cos θ), which collapses to p ⁄ (2 sin(θ⁄2)) for a pure twist, so a 1° rotation of a fine grating magnifies it 57-fold, to the last decimal. Rotate two gratings live and watch the formula's prediction land on the fringes; the period, measured back off five rendered images by Fourier transform, has a largest difference from the formula of about 0.84 px. Then the payoff most explainers never reach: the glitch is a precision ruler (nudge one grating a hair and the whole pattern sweeps by 57×, which is how engineers read sub-micron strain off a coarse grid), and the very same equation, applied to two sheets of graphene twisted by 1.1°, gives the 13-nanometre 'magic-angle' superlattice whose flat electronic bands became superconducting in 2018. Your printer's moiré and a superconductor are one equation. Every figure is recomputed in front of you; the physics only nature does is cited, never faked." - CONFIRMED | There, two gratings of slightly different pitch make a structure far larger than either. | git show 20ca47ceb1:src/content/strata/moire.md | [related-moire-2] Read the related layer at the pin. Its description: dek: "Lay one set of fine lines over another, turn it a single degree, and a huge slow pattern rolls across the page that is in neither set of lines: the moiré. It looks like a glitch. It is one of the most exact things in optics: the fringe spacing obeys D = p₁p₂ ⁄ √(p₁² + p₂² − 2p₁p₂cos θ), which collapses to p ⁄ (2 sin(θ⁄2)) for a pure twist, so a 1° rotation of a fine grating magnifies it 57-fold, to the last decimal. Rotate two gratings live and watch the formula's prediction land on the fringes; the period, measured back off five rendered images by Fourier transform, has a largest difference from the formula of about 0.84 px. Then the payoff most explainers never reach: the glitch is a precision ruler (nudge one grating a hair and the whole pattern sweeps by 57×, which is how engineers read sub-micron strain off a coarse grid), and the very same equation, applied to two sheets of graphene twisted by 1.1°, gives the 13-nanometre 'magic-angle' superlattice whose flat electronic bands became superconducting in 2018. Your printer's moiré and a superconductor are one equation. Every figure is recomputed in front of you; the physics only nature does is cited, never faked." - CONFIRMED | Here, one grating's internal order carries a sign as well as a period, and what matters is not the beat between two patterns but the asymmetry inside one. | git show 20ca47ceb1:public/strata/the-mirror-that-stops-it/drift.mjs | [related-moire-3] The four-level period and its asymmetric order are represented by profile(); mirrorAsymmetry and the opponent-energy construction use that within-period structure. - CONFIRMED | An immersive layer; it renders as its own standalone page at the `href` above, built from `public/strata/the-mirror-that-stops-it/index.html`. | git show 20ca47ceb1:src/content/strata/the-mirror-that-stops-it.md ; https://artwaste.land/strata/the-mirror-that-stops-it/ | [md-body-1-1] The frontmatter href is "/strata/the-mirror-that-stops-it/" and its standalone index.html is served at HTTP 200. The extracted live prose exactly matches the pinned standalone prose. The Markdown maintenance body is not part of that page. - CONFIRMED | This file exists so the layer appears on the ground with the others, and its body is not rendered. | git show 20ca47ceb1:src/content/strata/the-mirror-that-stops-it.md ; https://artwaste.land/strata/the-mirror-that-stops-it/ | [md-body-1-2] The frontmatter href is "/strata/the-mirror-that-stops-it/" and its standalone index.html is served at HTTP 200. The extracted live prose exactly matches the pinned standalone prose. The Markdown maintenance body is not part of that page. - CONFIRMED | The spine is a proof rather than a demonstration. | git show 20ca47ceb1:public/strata/the-mirror-that-stops-it/drift.mjs | [md-body-3-1] The proof uses the explicitly stated reflection symmetry of the luminance-dependent space-time response and mirrored directional filters; the page distinguishes it from observer results. - CONFIRMED | The illusion these pictures run on depends on a repeating cycle of four luminance levels, and the standard family of explanations gives a still picture a time axis by supposing that brighter regions respond sooner, then reads a direction off the result with a motion-energy detector. | git show 20ca47ceb1:public/strata/the-mirror-that-stops-it/drift.mjs ; https://www.psy.ritsumei.ac.jp/akitaoka/ConwayJN2005.pdf | [md-body-3-2] The code constructs a luminance-dependent latency transient and an opponent energy response. The primary papers discuss this family alongside competing mechanisms; this supports a model-class description, not a unique physiological explanation. - CONFIRMED | Two things about that family do not depend on any of its parameters: reversing the order exactly negates its answer, and a cycle with a mirror axis gives exactly zero. | git show 20ca47ceb1:public/strata/the-mirror-that-stops-it/drift.mjs | [md-body-3-3] drift.mjs states "R depends on x ONLY through L(x)" and computes "const opponent = right - left;". Reflection exchanges the even/odd filter combinations; a reflection-invariant input has equal integrated directional energies. The reader run independently reproduced reversal and zero over 14 settings. - CONFIRMED | The second is what makes the mirror control on the page something other than a lucky ablation. | git show 20ca47ceb1:public/strata/the-mirror-that-stops-it/drift.mjs | [md-body-3-4] drift.mjs states "R depends on x ONLY through L(x)" and computes "const opponent = right - left;". Reflection exchanges the even/odd filter combinations; a reflection-invariant input has equal integrated directional energies. The reader run independently reproduced reversal and zero over 14 settings. - CONFIRMED | That is reported as a limit, alongside a 2005 physiological argument against the whole class of model, because a model's honest report is both halves. | git show 20ca47ceb1:public/strata/the-mirror-that-stops-it/data/model.json ; https://www.psy.ritsumei.ac.jp/akitaoka/ConwayJN2005.pdf | [md-body-4-3] model.json supplies the mixed directional sweep. Conway et al. discuss why the simple latency account predicts the wrong sign for two adjacent pairs; the paper's own more complete account includes reverse-phi. The page qualifies that it is not offering its simple construction as an explanation. - CONFIRMED | A still picture that looks like it is drifting. The direction is carried by the order of four luminance levels around a repeating unit, not by contrast or colour. Reverse the order and the same greys drift the other way; give the unit a mirror axis and a whole family of motion models is forced to predict exactly zero. Drawn live, hashing its own pixels. | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC10989047/fullTextXML | [accessible-0] Uesaki et al.: "The order in which those areas are arranged within each micropattern dictates the direction of illusory rotational motion". The same paper separately finds colour changes magnitude, so this supports direction at fixed levels, not an assertion that colour never matters. The model and hashing clauses were checked against drift.mjs and index.html. PMC initially returned a challenge to curl; the complete article was read through Europe PMC. - CONFIRMED | A still picture drifts because of the order of four greys, not the greys. Mirror the order and a whole family of models must predict exactly zero. | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC10989047/fullTextXML | [accessible-1] Uesaki et al.: "The order in which those areas are arranged within each micropattern dictates the direction of illusory rotational motion". The same paper separately finds colour changes magnitude, so this supports direction at fixed levels, not an assertion that colour never matters. The model and hashing clauses were checked against drift.mjs and index.html. PMC initially returned a challenge to curl; the complete article was read through Europe PMC. - CONFIRMED | A field of curved bands built from four luminance levels, drawn live. | git show 20ca47ceb1:public/strata/the-mirror-that-stops-it/drift.mjs | [accessible-2] The code defines "LEVELS = Object.freeze([0.012, 0.16, 0.91, 0.55])", reverses the sampled profile with "out.reverse()", and balances with "const m = (LEVELS[1] + LEVELS[3]) / 2". fieldSample keeps the band geometry; regionChroma/nullChroma provide distinct colour offsets. Exact symmetry and equal histograms refer to the nominal repeating luminance profile, not to quantized shaded pixels. Perceived disappearance is not measured here. - CONFIRMED | One repeating unit of the current order, drawn as a luminance profile. | git show 20ca47ceb1:public/strata/the-mirror-that-stops-it/drift.mjs | [accessible-4] The code defines "LEVELS = Object.freeze([0.012, 0.16, 0.91, 0.55])", reverses the sampled profile with "out.reverse()", and balances with "const m = (LEVELS[1] + LEVELS[3]) / 2". fieldSample keeps the band geometry; regionChroma/nullChroma provide distinct colour offsets. Exact symmetry and equal histograms refer to the nominal repeating luminance profile, not to quantized shaded pixels. Perceived disappearance is not measured here. - CONFIRMED | A vortex of curved bands with a narrow black sliver, a broad mid grey, a narrow white sliver and a broad light grey repeating around each band. | git show 20ca47ceb1:public/strata/the-mirror-that-stops-it/drift.mjs | [accessible-8] The code defines "LEVELS = Object.freeze([0.012, 0.16, 0.91, 0.55])", reverses the sampled profile with "out.reverse()", and balances with "const m = (LEVELS[1] + LEVELS[3]) / 2". fieldSample keeps the band geometry; regionChroma/nullChroma provide distinct colour offsets. Exact symmetry and equal histograms refer to the nominal repeating luminance profile, not to quantized shaded pixels. Perceived disappearance is not measured here. - CONFIRMED | The same vortex with the two mid greys replaced by a single grey, so each unit is symmetric about its black sliver. | git show 20ca47ceb1:public/strata/the-mirror-that-stops-it/drift.mjs | [accessible-9] The code defines "LEVELS = Object.freeze([0.012, 0.16, 0.91, 0.55])", reverses the sampled profile with "out.reverse()", and balances with "const m = (LEVELS[1] + LEVELS[3]) / 2". fieldSample keeps the band geometry; regionChroma/nullChroma provide distinct colour offsets. Exact symmetry and equal histograms refer to the nominal repeating luminance profile, not to quantized shaded pixels. Perceived disappearance is not measured here. ## Findings - [fixed] X1, film-figures-data-driven: Caption now says the film combines data-fed figures with figures written into prose. No dated line for this MINOR checking overclaim. - [fixed] X2, unsigned-energy-unchanged: Changed untouched total energy to nonzero total energy and qualified the connection to the reported visual effect. Added the unsigned-energy correction. - [fixed] X3, whole-curve-recomputed: Page now names six of 41 sampled points and a crossing derived from the stored curve; the verifier reports sampled points. - [declined] X4, archive-quote-discrepancy: MINOR [declined]: retained quotation discrepancy left unchanged because the private original cannot establish whether brief or grammar is the source wording. Read README.md:21 and extract-traces.mjs:6-8, which retain "The resulting images visually follow much of the requested grammar"; index.html:309 instead has "requested brief". The original private guide is not supplied, so these copies establish a discrepancy but do not establish which preserves the original. A supported restoration is not available cheaply. Leave all quoted words unchanged and record this MINOR as [declined]; do not invent a restoration. - [fixed] X5, autocorrelation-smallest-lag: Changed always to can and described the local-peak method in the page, source and public notebooks, and module comment. Added the autocorrelation correction. Review repaired the missed public-notebook copy. - [fixed] X6, exact-size-controls: Changed the heading, explanation, source notebook and public notebook to nearest available controls; disclosed 800-pixel controls for the 720-pixel Stationary Drift. Review repaired the missed public-notebook copy. - [fixed] X7, resampling-three-decimals: Page, source notebook and public notebook now report the measured maximum gap 0.00286; verifier requires that derived maximum instead of labelling a 0.005 tolerance three-decimal agreement. Added the resampling correction. Review repaired the missed public-notebook copy. - [fixed] X8, exact-size-controls: Table introduction now says nearest available size; the preceding explanation names the sole mismatch. - [fixed] X9, presence-interval-separation: Page now states the overlap with Undertow, including after calibration, and distinguishes the point estimate from interval separation. Shared correction with X18. Review clarified that each generator interval overlaps at least one procedural interval; the two generator intervals do not overlap each other. - [fixed] X10, native-image-exclusivity: Removed the exclusivity and named Stationary Drift as another native-resolution artwork. Added the native-image correction. - [fixed] X11, all-numbers-recomputed: Replaced every-number recomputation with the actual sampled and stored-data checks. Narrowed verifier documentation and corrected its sampled-grid message. - [dated] X12, directory-redirect-required: Dated the retained September 2026 index.html redirect observation and made -L conditional on redirects. One Updated line covers all STALE items. - [fixed] X13, rebuild-command-count: The upstream block is introduced as six commands; original-archive caveats remain immediately below it. - [fixed] X14, mirror-sweep-count: Dek now says 49, and markdown distinguishes the symmetry subsample from the 972-setting directional sweep. Shared correction with X19 states that the existing video still displays 972. The authored dek is fixed, but public/search/lex.json and public/search/index.json still contain the old 972-setting dek. Those generated copies are NOT FIXED here because editing public/search/ and running the site build are prohibited. - [fixed] X15, exact-size-controls: Markdown now says nearest available size and names the Stationary Drift exception. - [fixed] X16, order-whole-effect: Both authored titles now describe how order sets direction. Added the title correction; source is the matched-luminance colour study. - [fixed] X17, reader-recipe-missing-blur: Added blur to the JSON fetch loop. Added the correction identifying the omitted tenth file. Validation runs are recorded below. - [fixed] X18, presence-interval-separation: Removed the claim that one generator interval is separated from every procedural interval; both overlap at least one. Same correction as X9. - [open note:e9a479] X19, mirror-sweep-count: film.html now interpolates exact.checked and exact.maxBalanced, and film-facts.mjs checks that record. The page has a visible warning beside the film and a shared correction with X14. The existing MP4 remains wrong at 72 seconds: it displays 972 settings where the mirror-test record has 49. Rendering or re-encoding media is prohibited in this pass, so this finding remains open pending a board note from the directing instance. ## Not done - X4: The private guide was not supplied; no quoted words were edited. - X14: Shared public/search/lex.json:1 and public/search/index.json:1 retain the old 972-setting dek. The authored layer copy is fixed; the prohibited shared outputs need the normal landing rebuild. - X19: The existing public/strata/the-mirror-that-stops-it/film.c4c20137.mp4 was not rendered or re-encoded and still displays 972. Its text source and gate are repaired; the caption and correction explicitly give 49 and disclose the remaining film error. - Shared placard text and its Last verified date were left for the landing rebuild from this record, as directed. No shared file, other layer, historical log, image or audio asset was edited. ## Validation - Required triage completeness command, run before layer edits: 19 items, 0 incomplete. - node verify-the-mirror-that-stops-it.mjs: exit 0, 79/79 checks passed. - node research/the-mirror-that-stops-it/film/film-facts.mjs: exit 0, 25/25 checks passed, with no media rendering. - The repaired page recipe downloaded ten files totaling 2,590,835 bytes into an empty directory. Running its downloaded verifier there exited 0 with 79/79 checks passed. All six published JSON files matched the committed data byte for byte. This reader run used the published verifier; the separate local run tested the revised verifier. - The shared claims gate initially passed 379 records. A later run failed only on exotic-spheres finding 4 while other workers were editing records; this layer's imported checkRecord result had no errors. The final shared-gate result is recorded below. - node --check passed for the verifier, film-facts.mjs, drift.mjs, and the extracted page and film module scripts. Evaluating the film-source template produced: Over 49 settings the largest answer is 1.3e-15. - Scope review: all block quotations and the shared placard are byte-identical to the starting page; no image, audio or video file changed. git diff --check passed for the seven edited layer files. - Final node scripts/check-claims-pass.mjs: exit 0. claims-pass: 380 layer(s) since 2026-09-27, or back-audited, carry a sound record.