Houston,
we have a skeptic.
And he has a PhD.
A Research Scientist at NVIDIA — a man who builds billion-parameter models on top of satellite data for a living — has announced that he "questions the moon landing."
This site is a public service. Every claim is linked to a primary source, which is a courtesy the good doctor did not extend to us.
Dr. Jason Stock
"From Neuro-Inspired Attention Methods to Generative Diffusion: Applications to Weather and Climate" — trained on data from, wait for it, satellites.
"SRViT: Vision Transformers for Estimating Radar Reflectivity from Satellite Observations at Scale" — funded in part by the GOES-R Program. That's a NOAA/NASA satellite program. He cashed the check.
Co-built AERIS, a billion-parameter Earth systems model. 2025 ACM Gordon Bell Prize finalist. Genuinely brilliant. Which makes this so much worse.
"I'm not full on tin foil hat man but I definitely question it." — the subject, this morning, 8:02 AM, in a group chat.
The Exhibits
"If we really went, why did we never go back?"
Jason's whole case rests on theory 5 of the Royal Museums Greenwich debunk list — the one the museum itself stamps "Conspiracy status: debunked." He cited a page that debunks him. Respect.
The answer isn't a cover-up. It's the least mysterious force in government:
Apollo cost $25.8B — about $309B in today's dollars (Planetary Society). Congress cancelled Apollo 18–20 in 1970. Vietnam ate the budget, the race was won, and priorities moved to the Shuttle and ISS.
The Evidence, Doctor
Each card follows the format you failed to use: claim → mechanism → primary source. Non-American sources included, per your own methodology ("Non American source that has some good debunks lol" — your co-counsel, Parker).
You can laser the mirrors tonight
Apollo 11, 14 & 15 left retroreflector arrays. Observatories still bounce lasers off them at millimeter precision — including the French station at Grasse, which has done it since the 1980s. The hardware is physically there. Anyone with a big enough laser can check.
We photographed the landing sites
NASA's Lunar Reconnaissance Orbiter imaged every Apollo site: descent stages, experiments, rover tracks, and the astronauts' actual foot trails in the dust — which will persist for millions of years.
So did India. And Japan.
India's Chandrayaan-2 imaged the Apollo 11 & 12 hardware (2021). Japan's Kaguya probe reconstructed, from its own 2008 terrain data, the exact landscape in a photo taken by Apollo 15's crew in 1971. You can't pre-render a 3D match for a probe that wouldn't exist for 37 years.
382 kg of rocks, five decades of audits
Apollo samples have been loaned to thousands of labs in dozens of countries. Isotopes, zero hydrated minerals, cosmic-ray exposure, micrometeorite zap pits — unfakeable in 1969, unfaked since. And they match the USSR's own robotic Luna samples.
The Soviets tracked it live. And said nothing.
The one adversary with full capability and every geopolitical incentive to expose a hoax monitored Apollo in real time — and even flew Luna 15 to the Moon during Apollo 11. Zero objections, 1969–1991.
400,000 conspirators, zero leaks since 1969
A peer-reviewed model of conspiracy longevity — calibrated on real conspiracies that actually leaked — gives a moon-hoax of this size ~3.7 years before exposure. It has held for 56 years and 363 days. See "The Math" below. You like math.
The UK and Germany heard it from the Moon
Jodrell Bank (Manchester) tracked the Eagle's descent; Bochum Observatory (Germany) independently received the lunar module's transmissions. Radio direction-finding doesn't lie: the signals came from the Moon, not Nevada.
1969 couldn't fake the footage
Faking lunar gravity needs slow motion. The only 1969 slow-mo videodisc (Ampex HS-100) held 30 seconds; Apollo 11's EVA ran 143 unbroken minutes. Film would've needed visible reel-change cuts. There are none. Also: dust arcs in perfect ballistic parabolas — vacuum only.
"90% of rockets still explode" — that's the point
Rocketry is hard, which is why Apollo needed 4% of the federal budget and 400,000 people — and why nobody casually repeats it for fun. SpaceX optimizes for reusable cost-per-kg, not flags-and-footprints. Different problem, Doctor. You of all people know what an objective function is.
We literally just went back
Artemis II: four humans around the Moon, April 1–10, 2026. Farthest-from-Earth record: 406,770 km (NASA says 406,770.34 — we refuse to round up; see Revision History). Crew includes a Canadian, for non-American-source purposes. Artemis III lands next.
The Math (You Like Math)
Grimes (2016, PLOS ONE) modeled how long conspiracies survive before an insider leaks, calibrated on real conspiracies that actually failed (NSA PRISM, Tuskegee, FBI forensics).
A man who trains diffusion models is skeptical of the moon landing but not of a 400,000-person, 56.99-year, zero-leak secret. The loss function has diverged.
Buy the watch, Jason.
The Omega × Swatch MoonSwatch "Mission to the Moon 1969": Moonshine Gold made by melting down Omega parts dating to 1969, honoring the Speedmasters worn on the lunar surface.
Your only stated objection to owning it was that the event it commemorates didn't happen. That objection is hereby resolved (see Evidence 01–10). We await your lottery application.