Fun With Science / Globe Deconstruction
Twenty-four checkable claims from a 500-page prerelease draft — what each one says, where it sits, and whether we have answered it.
This is an in-progress review of Globe Deconstruction?, a prerelease book by Levi Miller, and of the argument at shapedebate.com, which Miller runs with Alex Kampf. The book calls its position geocentric flat-earth, states three headline claims and twelve numbered science questions, and asks for exactly this sort of scrutiny — “no religions, no beliefs, no biases, no assumptions.” We are taking that at its word. Each claim is answered on its own page, in the author’s own wording; the general argument the pages share is made once, on Method & Foundations.
The finding
Of twenty-four claims, twenty-two are graded exactly as he states them. Thirteen are contradicted by a specific measurement, four are real observations the standard model already predicts, four overrun their own evidence, one is asserted without the evidence to establish it — and the number that survive as a falsification of the globe, a prediction it actually makes measurably failing, is zero. That is a running score, kept by a rubric with a state he can win, and it will change the moment something earns it.
The book is strongest where it tries something new and weakest where it recycles. Its genuine breaks from the genre are of method: falsification-first, a speculation fence around its own weakest material, and tests it builds or proposes itself — a raft towed across a pond, a flight-data challenge, a reflection that should compress with height, a Jupiter photograph read for its shadows. Those get the hardest look here, and several are conceded in part. Its weakest passages are the ones that arrive second-hand: the Chicago time-lapse, the P900 recoveries, the Rampion footage, the Action Lab chamber, the one-fake-voids-all rule and the moon-tilt spread are the standard catalogue of the last fifteen years, largely third-party footage, with published answers already waiting.
It leans on outside references more than a book can bear. Much of the case does not live on the page: it lives in a linked video, a QR code, a third party’s footage, or a technical paper at shapedebate.com/42. The reader who wants to test a claim first has to reconstruct it — from a broadcast, a slide, a transcript — before there is anything to test, and this review has had to do that reconstruction, frame by frame, to answer at all. It makes the book hard to follow and harder to evaluate, and it means the argument as printed is often weaker than the argument as meant.
It does not raise a new argument for a flat Earth. What it does, more than once, is dig up independent data that supports the globe. His pond arithmetic is right to the digit. The Naval Observatory tables on his eclipse slide, subtracted, give the approach he says did not happen. The time-lapse camera whose footage he adopts measures the globe’s prediction to a degree. The book’s own worked example of the moon-tilt, run through the closed form, returns the 45° its source offered as impossible. His air-speed table is right to the mile per hour. The two Jupiter frames his QR codes link to show the shadow offsets a single distant Sun predicts, to about a per cent. At the lake he nominates for his reflection test, a photograph already taken there settles the question by the angle to the summit — against the flat plane. That is the best thing that can be said about the book, and it is not a small thing: it went looking, and it kept what it found.
By verdict — what we concluded about his claim
By our progress — how far the answer has got
Twenty-four claims. The two rows measure different things: a claim can be fully answered and still come back as an observation the mainstream model simply predicts.
Each page answers one specific claim, as its author actually stated it — not a summary, not a weaker version, not the genre it belongs to. Where the observation is real or the arithmetic is right, we say so first. Where the reasoning fails, we show the working from public data and named sources, using the author’s own figures wherever he supplies them, so the reader need not trust us. Every page ends with tests that could go against us.
The standard is higher than usual because most published responses to this material argue with a caricature — the loudest version of a claim rather than the one made. The book notices this and says so, fairly. So several pages here concede real ground: the pond arithmetic is correct, the perspective-compression physics is real, Foucault pendulums are hard to run well, and the book is unusually candid when its own evidence is thin. It also charges the globe side with demanding a rigour it never supplies itself, and on that charge it is right; our page on it concedes the point first and binds this review to the same check — every mainstream figure here is verified against its source.
We are in correspondence with the author, who has been generous about receiving criticism. If any specific claim on any of these pages is wrong, tell us which one and why, and the correction goes up with attribution rather than being quietly swapped out.
Three claims are published twice over — on shapedebate.com and on page 2 of the book — and are presented as the strongest evidence the position has. They are the load-bearing set by the authors’ own construction, held to the hardest standard here, and none of the three stands. The first is standard physics in the version the site states and contested in the version the book adds. The second has been measured in space, several unrelated ways at once. The third is the one to notice: it is offered as a cumulative case, and the four “hardest to ignore” observations it names are the only entries in its supporting lists that commit to anything — of the twenty-two entries in the red-flag lists at pp. 165–166, twelve ask whether someone should check something rather than stating what is wrong. The four that do commit are answered, and none of them holds. A case whose best evidence is a list of things somebody ought to look into is a request for a case, not a case.
Where the two wordings differ the difference is quoted; on the first claim it decides the verdict.
“Objects can completely disappear across a flat surface even when there is no obstruction in between. Visual limitations with a long-distance surface have not been properly studied.” That is shapedebate.com as it stands today. The prerelease book states the same claim differently at p. 2 and again at p. 209 — “Objects can disappear from the bottom up across a flat surface due to the limitations of angular resolution” — naming a mechanism the site version does not. We answer both, and the difference matters to the verdict.
“Mainstream physics is completely wrong in its understanding of how gas-propelled vehicles work in a vacuum. Thrust from gas must have something to push off, and an infinite vacuum does not provide that.”
“There are too many red flags in celestial observations. The four hardest to ignore are the uniform light distribution of a full moon, the moon-tilt illusion, Jupiter’s moon shadows that fail to align heliocentrically, and the failed trajectory alignment during a solar eclipse. Together, they make a strong case that the Sun is not the source of illumination.”
The groupings below are the book's own — its twelve numbered questions and its chapter titles — sorted into five families by what each claim argues about. The structure the book does not supply, what these claims share and where they collide, is on Method & Foundations.
How claims are graded
Seven verdicts, used unchanged across this project and its siblings Flat Earth Origins and the dome-model review, so a grade means the same thing everywhere. The verdict grades his claim; the smaller chip beside it tracks our progress. A verdict marked prov. rests on a page that has had one pass and no review. Items with no verdict have not been assessed — that is a real state, not an oversight.
Standing rule: grade the claim as he actually hedged it, never a compressed or strengthened version. Where a page answers a stronger claim than the book makes, the item stays unassessed until that is fixed.
The first grouping is the book’s own; the five after it are ours, sorted by what each claim argues about, with the book's order preserved in the page map. Every question is quoted verbatim from the prerelease draft, which numbers two consecutive questions Q4 (pp. 46 and 47); both are listed. The index counts 24 claims rather than his 22 because two long sections — Hydrostatics & Optics at p. 88 and Space Travel Requires Faith at p. 56 — get rows of their own. Each row also carries the two-kinds tag: States asserts something about the world that can be checked; Asks raises a question or proposes a test it does not run. A question counts as asking even when it presupposes an observation; a section is tagged on its thesis. Eleven state, thirteen ask — and the load a case can carry is set by the first number.
Six claims. The strongest part of the book, and where most of what we concede sits.
“Why has no one proved that spacetime curvature (known as gravity) is strong enough to literally bend the laws of hydrostatics in a controlled environment?”
“Do we observe perfectly mirrored reflections over large bodies of water at rest as we increase observer height?”
“Why on clear days with low humidity and high altitude does the Sun observably not set (like shown in movies), but rather shrinks due to linear perspective and disappears completely from a lack of visibility?”
“What scientist proved that sunlight moves in a perfectly straight line through hundreds or thousands of miles of gas? Is this an incorrect assumption when measuring shadow angles to dismiss the possibility of a flat Earth?”
Hydrostatics & Optics — twenty-nine pages.
Celestial Globes Exposed — Polaris's inconsistent calculated height, and whether linear perspective accounts for it.
Two claims. The most decisively testable set, and the one with the most measurement behind it on both sides.
“Why does the scientific method reveal that gas-propelled vehicles utilize the surrounding gases for movement? If verified, all engineering calculations for space travel are based on a faulty model.”
Space Travel Requires Faith — thirty-two pages, including the vacuum-chamber video at pp. 80–83.
Five claims, all answered — the only family where public instrument data settles the question without either side running a new experiment.
“What would the flight data recorder reveal if a pilot maintained perfect positioning on the equator, a constant turn, or a straight line around a sphere?”
“How was the scientific method used to prove that air layers have enough friction to stay in perfect rotational sync across all heights of the atmosphere and the Earth to where we perceive zero motion? Mechanical common sense states that the speed of the air would be faster as you get closer to the rotational source.”
“If you travel to the exact location of the rotational south pole, why does a high-altitude time-lapse show no movement of the surrounding Earth landscape relative to the camera? Keep in mind that the rotational speed of the air at the axis point would be zero. Why has no scientist tested the Compton generator in Antarctica?”
The Equator Flight Data Challenge.
Perfectly Synchronized Gas Rotating with Earth? — co-rotation, gyroscopes, the Foucault pendulum and the rotational axis.
Five claims, four answered. The largest open gap in this review.
“Why is there no difference in visual speed between stars and planets as the Earth spins during a single night? A large distance gap requires that visual speeds will not be equal to the observer.”
“Why do the moons of Jupiter produce shadows that are not aligned with the singular light source of the Sun? Are we assuming them to be shadows or can it be proven?”
“If the Moon is a spherical rock in a vacuum, how does the light of the Sun perfectly distribute across the entire surface during a full Moon without any highlight points?”
“Why do we never observe the entire circular silhouette of the Moon when approaching a solar eclipse? Photo manipulation always fails to detect evidence of the Moon.”
Unknown Luminaries — the planets, Jupiter's moons and their shadows, and what the objects in the sky actually are.
Three claims across 271 pages. Nothing here is an observation about the shape of the Earth, which is why the family sits last — and why one of the three is catalogued rather than answered.
Extraordinary Evidence or Fallacy — the standard of proof demanded of the two models; the closing summary.
All Construction Records Missing? — 174 pages on nineteenth-century architecture and World's Fairs.
Antarctica's Magnetic North — and the speculative map that removes magnetic declination.
Whether space-agency footage is authentic. The book’s case that ISS and Apollo material is staged runs through pp. 10–11 and Space Travel Requires Faith at pp. 56–87. We do not assess it. The physics in that same section is not exempt: whether a rocket can push against vacuum has an experiment attached, and two pages of its own.
Whether The Final Experiment was staged. On p. 11 the author puts himself at 90 per cent certain that the December 2024 Antarctic expedition was faked. We do not assess that either.
Both turn on the authenticity of footage and the good faith of named people, not on a measurement a reader can repeat; a review that graded them would trade arithmetic anybody can check for an opinion about character. Silence here is not agreement and not a concession; it is the edge of what this method can do. Neither is one of the twenty-four; they meet the catalogue only in the Space Travel Requires Faith row, marked unassessed for exactly this reason.
A review that cannot concede anything is not a review. These are places where the book is right, or the observation is real and only the cause is in dispute.
Over 540 feet a globe hides nothing. His curvature figure is right to the digit, and something did disappear.
Angular resolution and perspective compression really do make distant objects vanish over flat ground; the question is at what range.
Gas reaches the chamber wall about two seconds of playback before the syringe moves. Neither account forbids that ordering.
They are difficult to run well, and the demonstration is often presented with more confidence than the apparatus earns.
It is correct, and it lands on this review too; the page concedes it first, then binds this review to the same check.
Two hundred and thirty-three pages are stamped SPECULATION and excluded from the shape debate by the author himself. That is more discipline than the genre usually shows.
Every answered page ends with tests that could go against it; thirteen of them are collected as a hand-off in the book's own falsification-first, Team A versus Team B form, each with its globe and flat-plane predictions stated in advance — the Self-Test Protocol. The project as a whole has two.
On any single claim: a claim that survives scrutiny and genuinely counts against the globe model as stated is graded Sustained. The count is zero today. If one lands it goes at the top of this page and the scoreboard changes the same day.
On the thesis of this page: the mesh argument says the book needs a different world per chapter. It would be retracted by a single stated set of auxiliary positions — one optics, one atmosphere, one sun — that holds across every row of the collision table at once. Not a reply to each row in turn, which is always available; one set, applied to all of them. We have not been able to construct one on his behalf, and it is what we would want to see.
noindex, is not linked as an answer in the catalogue above, and carries a notice naming what is still outstanding on it. A draft that passes a review pass moves to the catalogue above as an answer; a page that was never a claim of his sits with the supporting pages below.Celestial Globes Exposed, p. 148 — the blueprint challenge. Concedes the premise outright: engineering documents mostly do not say "Earth's curvature". The argument is about procedures that would visibly fail if the term were absent — levelling that defines level as perpendicular to local gravity, radio path planning, tunnel geodesy.
Work in progressA reference check on the same file: five documents concatenated rather than one paper, an unattributed lead monograph, and experimental detail that does not survive inspection — including an apparatus using LEDs in 1913. Kept separate so the physics does not rest on it.
Tool · no claim of his. Enter what a camera saw; it returns the refraction a globe and a flat plane each need, as a temperature gradient against the weather that occurs.
Map · no new verdict. His Team A versus Team B scorecard (pp. 208–231), item by item: 102 Team B items sorted by what they bear on, each with a reader-check and a reference.
List · the other side of the ledger. Fourteen structural claims, each with a number, a test, a falsifier and a reader-check.
Foundation page · no claim of his. Several claims turn on the Moon, so its shape is established once: five independent lines, four ground-based and three doable with a camera — the terminator a facing disc cannot have, shadows that diverge where the light grazes, features swinging a hundred arcseconds across a face whose outline never deforms, a radar echo saying the Moon is as deep as it is wide to four parts in ten thousand, and a limb with relief that returns to the same place.
Housekeeping · no claim of his. Production slips in the book and the site, kept apart so they are never mistaken for objections: three QR codes labelled SUN, MOON and STARS resolved one position out of step, since corrected, and the question cards run Q1, Q2, Q4, Q4, Q5. The test for inclusion: the item would still want fixing if he is entirely right and this review entirely wrong.
Foundation page · a source he cites, not a claim of his. The technical paper at shapedebate.com/42 is named as the root of three arguments at once — the sunset, Polaris and the pond — so it is answered once. Its central equation is correct and granted; on its own worked example the mechanism it describes can never put anything below a horizon, and its own deflection table reads zero at the horizon, the one place every claim it is cited for actually happens.
Fifteen NOAA ocean drifters recover the length of the sidereal day to within half a per cent, and their loops reverse direction at the equator. Three models of Earth predict three different pictures; the data picks one. Supporting evidence for Q10 and Q11.
Thirteen falsifiable tests built from the book's own stated methodology — falsification-first, Team A versus Team B — that the author, or any reader, can run, each with a globe prediction and a flat-plane prediction stated before the result.
Ground chambers can always be accused of not being empty enough, so engines measured firing in real space: a 1993 NASA cross-check — accelerometer and orbit change in flight against a ground thrust stand on one ion engine, reaction-wheel momentum on a second — agrees to a few per cent and puts the chamber artefact at about one per cent, in the direction that flatters the chamber.
The book argues at p. 20 that a third party's vacuum-chamber video disproves rocket thrust, because the syringe moves only once its exhaust reaches the wall. The ordering is real and conceded, and neither account forbids it; what the footage rules out is the wall as the cause — the delay is 25 to 220 times too slow for a pressure wave, and thrust answers its own flare at the nozzle within a frame or two.
In August 2026 he towed a four-inch raft 540 feet across a pond and filmed it vanishing from a lens a third of an inch above the water. His curvature arithmetic is correct to the digit — a globe hides nothing at that range — so something local did the work, and his own source files say what: from that vantage the raft and the whole pond behind it fit inside a single blur width of a camera with the equivalent of 20/100 vision. Refraction is the one candidate the footage measurably excludes.
A point-by-point response to the Rampion Wind Farm curvature video the book cites at pp. 91–95 as real-world evidence, including the two-century record of superior mirage on that stretch of English coast.
The render, the balloon frames and the rocket footage of pp. 234–237, measured on one rule; the rocket's horizon bows as a globe predicts, and the same footage answers Claim #2.
A separate review, from an adjacent project: whether ring-laser and fibre-optic gyroscopes detect Earth's rotation, what instrument drift does and doesn't explain, and the 180° test that separates the two.
Of the twenty-four items, twenty-two carry a verdict and two do not — 13 refuted, 4 misleading, 4 standard physics, 1 not demonstrated. The four unassessed are so for stated reasons: one sits inside the author's speculation fence, two are sections whose wider argument we have not answered, one of them by choice (the agency-footage material above), and one is a charge against this review that we accept. This is an active review of a prerelease book, so both sides are moving; pages are revised in place, and where one has changed materially in response to the author, it says so.
One thing in scope has no page yet: the geocentric model the project’s launch video endorses. It is described, and the burden it moves, on Method & Foundations.