Fun With Science / Globe Deconstruction / Q5 + Q6 · pages 48–49 / Draft
Globe Deconstruction Review — a reference check on the paper behind the perspective argument: what it contains, who wrote it, and which of its evidence survives inspection.
One paper: no — fiveLead monograph: unattributedFabrication indicators: three, all checked
Where this lands
This page exists so that the physics response does not have to lean on it. That page stands whether or not a word here is true: the paper’s own page 5 refutes the use it is put to, and that argument needs no view about who wrote it.
But the document is cited as a technical paper that “100% proves” its claim, and a reader who follows the link deserves to know what they are looking at. What they are looking at is an anthology, and its lead item has problems that go beyond argument.
The PDF restarts its internal page numbering five times. The five parts do not share typography, argument, sourcing conventions or failure modes. Treating them as a single work is the first mistake available here, and it flatters the weakest part by association with the strongest.
| PDF pages | Document | Byline |
|---|---|---|
| 1–6 “paper 1” | ‘Empirical and Mathematical Derivation of Perspective Line Curvature in Biological and Optical Projections’ | None. Affiliation given as a “Division of Vision Science, Optical Physics & Perceptual Geometries”, which we could not identify as an existing institution |
| 7–8 | ‘Spatial Mislocalization of High-Altitude Visual Targets Along the Optical Centerline’ | None. No author, affiliation or date |
| 9–39 | ‘Perspective Appearances and Representations’ — a theorem-and-corollary treatise in the style of Euclid’s Optics | Vortexpuppy |
| 40–45 | ‘Oops we forgot perspective. lol.’ — on Polaris-elevation geometry | Informal, same milieu |
| 46–52 | ‘When does a circle appear as a circle?’ — on Euclid’s Optics Theorem 34 | Vortexpuppy |
The item actually cited as “the technical paper” is the six-page monograph at the front — paper 1, as the physics page calls it throughout. The rest is supporting anthology, and it is worth saying plainly that the Euclidean treatise is the better work: careful classical geometry, applied correctly and then extended past what it can carry. Our disagreement with it is about reach. Our difficulty with the monograph is different.
At p. 127, immediately before the QR code numbered 42, the book tells readers: “If you have any doubt that our vision is globular, read through this technical paper from Vortexpuppy (PhD in mathematics).”
That is a single named author with a claimed credential, offered for the whole file. The file does not match it. Vortexpuppy is named on p. 46, over the Euclidean treatise and the circle commentary — the part of the anthology we think is the better work. The six-page monograph at the front, which is the item the argument actually rests on and the one with the problems below, carries no author, no affiliation that we could identify, and no date.
We are not in a position to say who wrote the monograph, and the book may simply be describing the file by its one visible byline. But the effect is that a reader following that sentence arrives at an unattributed document believing it has an author and a doctorate behind it. The credential itself we have not attempted to check, and would not report on if we had — it is a pseudonymous author’s claim about himself, and nothing in our argument turns on it.
The citation skeleton is genuine. Hillebrand 1902, Luneburg 1947, Brown 1966, Koenderink & van Doorn 1992, Pepperell and colleagues 2018, Hertzmann 2024 all exist and are real work in this area. What does not survive inspection is the experimental detail attached to them.
The text attributes the alley experiments to “Franz Hillebrand (1902) and Steel Blumenfeld (1913).” Its own reference list gives “Blumenfeld, W.” The modern literature gives the same: W. Blumenfeld, Untersuchungen über die scheinbare Grösse im Sehraume, Zeitschrift für Psychologie 65 (1913), 241–404, alongside F. Hillebrand (1902). There is no Steel Blumenfeld. A document whose prose disagrees with its own citation list about the name of a researcher has not been read by its author against its sources.
The monograph describes the 1902 and 1913 alley apparatus as using “24 point-source green LED lamps mounted on bilateral motorized transverse tracks”, with dimensions and a fixation rig quoted to two decimal places. Practical visible light-emitting diodes date from 1962, and green ones later still. No such apparatus existed in 1902 or 1913; the historical experiments used flames and rods.
This matters beyond the anachronism. A quantitative result is attributed to those same studies — a hyperbolic-cosine law for alley separation — and a description invented in this much detail cannot be relied on to have preserved a real number.
The “empirical optical bench proof” presents a table of measured sub-pixel deflections at sensor rows running to 3,900 px, credited to Duane Brown’s 1966 lens-calibration work.
There was no image sensor in 1966. The charge-coupled device was invented at Bell Labs on 17 October 1969 — three years after the paper. The first experimental CCD, built in 1970, was an eight-bit linear array. The first two-dimensional imaging CCD, in 1971, was 100 × 100 pixels. A sensor with 3,900 rows is a device of the mid-2000s, roughly forty years downstream.
So a table of deflections indexed by sensor row, quoted to sub-pixel precision, cannot have come from a 1966 publication by anyone. The measurement it describes was not physically possible when the work it is credited to was done.
That argument runs on the calendar alone, and it is sound on the calendar alone. But Brown’s paper settles the same point directly, and it is worth saying what is in it rather than only what could not be.
It is Duane C. Brown, ‘Decentering Distortion of Lenses’, Photogrammetric Engineering Vol. 32 No. 3, May 1966, pp. 444–462 — nineteen pages, nine figures, two tables. The experiment at its centre is a stellar calibration of a pair of Pth 60 phototheodolites, 600 mm focal length and 17° × 17° field, photographing the same star field side by side. Brown gives the recording medium in one sentence:
“Kodak Microflat glass plates coated with 103 F emulsion were employed. The plates were photographically processed together and were measured on a calibrated Mann comparator by the same operator on different days.”
Glass plates, 6 mm thick and 190 × 215 mm, wet-processed, then measured by hand — 155 star images per plate, two settings on each image. Every measurement in the paper is a plate coordinate in millimetres and every error in it is quoted in microns; the scale bars on the residual-vector figures read “0 5 10 microns”. The words pixel, sensor, digital, electronic and charge-coupled do not occur anywhere in the nineteen pages.
Nor is there a table of the shape described. There are two of them. Table 1 gives mean errors for the two cameras under three adjustment cases — six numbers. Table 2 gives two calibrated decentering parameters per camera. Neither is indexed by position across a detector; neither runs to thousands of rows.
There is a mismatch of subject as well. Barrel and pincushion are symmetric radial distortion. This paper is not about those — it treats them as a solved part of the reduction and works on what is left over. Its subject, as the title says, is the decentering residue: what a slight misalignment of a lens’s elements adds, which Brown describes in his first paragraph as tangential distortion plus an asymmetric radial component. He also puts the scale of it on record in that same paragraph. Suppressing it physically “to a value not exceeding five microns over the plate format” takes “appreciable skill and patience” from the optical technician, and getting below two microns “calls for luck in addition to skill and patience.” Five microns across a 215 mm plate is two parts in a hundred thousand. That is the size of the phenomenon the deflection table is credited to.
The paper is real and foundational, and the Brown–Conrady model that carries his name is the standard treatment used everywhere in camera calibration today. That is exactly what makes the citation effective: the reference is genuine, and the data attributed to it is not.
No named author anywhere in the lead document. A “CONFIDENTIAL & PROPRIETARY — ACADEMIC RESEARCH MONOGRAPH” footer on a file distributed by public link. A section headed “The 10/10 Unassailable Physics”. A “Plain-English Translation” block after each equation. An abstract that says “definitively refutes”. Individually each is a style choice. Together with the two checked indicators they describe a familiar pattern: real references, invented experimental content.
It licenses one thing: the monograph’s empirical claims cannot be relied on, and anyone citing it as established work should know that its two historical anchors do not hold up.
It does not license anything about Miller. He cites the document as a source; he does not present it as his own, and there is no indication he wrote it or checked it. Citing a bad source is an ordinary failure and it is one this review has to be careful about, having made sourcing errors of its own on these pages. The right response is to say what the document is and let the physics stand on the other page, which it does independently.
It also does not license dismissing the anthology. Vortexpuppy’s geometry is mostly correct Euclid carefully applied. Where we disagree is that classical results about how things appear are extended into claims about what can be hidden, and that step is not in Euclid.
A credit line can be loose without being false. If the monograph means only that its bench used Brown’s method, and the numbers are its own measurements on its own modern camera, then the anachronism does not bite the way we have described it. We think the plain reading of “empirical optical bench proof” attached to a named 1966 source is that the data comes from the source, and the table is presented as evidence rather than as illustration. But the looser reading is available, and on that reading the finding changes shape rather than disappearing: the table would then have no stated apparatus, no operator and no date at all.
We identified no institution, which is not the same as none existing. The affiliation on the lead monograph returned nothing we could confirm; absence of a search result is weak evidence and we are not treating it as more.
We are reading one distributed copy. The file’s embedded metadata gives an August 2026 creation date and a macOS PDF pipeline. If a different copy of the paper carries a byline, that changes the first finding.
shapedebate.com/42. The structure table above is read from that file.