Case study · working instrument · rebuilt 2026-08-06

Mars Arena — two populations, one globe

A rotatable Mars with real orbital imagery, carrying two sets of places that are normally discussed in different rooms: the 1,062 cave and lava-tube candidates NASA and USGS catalogued as somewhere a mission could shelter, and the nine places American Alchemy named on air as possibly built. One switch moves between them. Every marker in the second set carries the established scientific reading in the same card as the claim.

The problem

A claim about a place on Mars is almost never checked against the place. It is checked against other claims — a screenshot, a coordinate with no datum, a photograph nobody re-fetched. Meanwhile the actual imagery is public, free, and one HTTP request away. This instrument puts the claim and the coordinate on the same screen as the picture, so the check takes seconds instead of an afternoon, and so the answer is visible to a reader who does not already know how to do it.

Pipeline

01 ArchiveNASA PDS4 cave catalogue, parsed against its label rather than its header
02 CorpusPlaces named across 136 canonical episodes, pulled from the claim graph
03 LocateEach one geolocated against the IAU Gazetteer and mission product labels
04 LookOrbital imagery fetched live at every coordinate, in the browser

The instrument

No backend. The page asks NASA Trek for the basemap tiles and the USGS Astrogeology WMS for each close crop directly from the browser — both send access-control-allow-origin: *, which is what makes a static host enough. Nothing here is a screenshot of a product; the globe, the table, the inspector and the imagery are the working thing.

app/ · Mars Arena Open full screen

The two layers

The switch at the foot of the globe is the only choice a first-time reader has to make, so it carries its own counts and both layers start on. The distinction it draws is not cosmetic — the two populations have completely different evidentiary status, and the interface never lets them blur.

LayerCountWhat each marker is
Habitability caves1,062A record in a peer-reviewed NASA/USGS catalogue. No claim attached to it by anyone; the archive’s own targeting priority and type code are shown verbatim.
Potential ET objects9A place named in an episode. Carries the claim, who is associated with it, the established reading, and a link to the primary source for each.

Five further places from the corpus are in the dataset and deliberately not drawn: two have no cited coordinate at all, one is on Phobos, one is a meteorite in Antarctica, and one is the planet itself. A Mars globe is the wrong surface for four of those, and inventing a pin for the fifth would be the exact error this page exists to avoid. They are listed with the reason rather than dropped.

Four corrections found while building it

These are not polish. Each one was shipping, each one was invisible, and each one would have been repeated by anyone reading this project as a source.

The pyramid, measured

A post circulating in July 2026 showed a four-lobed mound beside the Great Pyramid at matching scale, under the words “NASA announces discovery of a pyramid on Mars”, with a coordinate and no datum. It is on the map as object 9. Three things were separable, and separating them is the whole method:

exactThe coordinate resolves8.894173° N, 78.631929° E — the pair was unlabelled and could be read either way; both were fetched and only one lands on the feature
3.1–4.9 kmThe size claim holdsmeasured on the HiRISE ESP_077288_1890 browse image; the post’s “3 miles corner to corner” = 4.83 km sits at the top of that range
not announcedThe provenance claim does not“pyramidal mound” is a descriptive title on a routine HiRISE/CTX product, not an announcement of anything

So the interesting answer was not “debunked”. Two of the three components check out, including the one that sounds most extravagant, and the one that fails is the one nobody was arguing about. That is the shape most of these turn out to have, and it is only visible if you check the parts separately instead of the story whole.

What is real

All 1,062 catalogue records with their archive coordinates, type codes, priorities and verbatim comments. The corpus graph — 56 nodes, 52 edges, 72 claims — as extracted from the episodes. Every coordinate on the anomaly layer, with the source that published it. Every orbital image, fetched live from USGS and NASA Trek at request time. The one size figure this project measured itself is labelled as measured by this project, with its method and its uncertainty.

What is computed, and says so

The habitability score is this app’s hypothesis, not a finding: a weighted sum whose every term, weight and substitution is shown in the inspector so a reader can re-derive or reject it. No NASA, USGS or JPL validation of any ranking here. The anomaly layer is a record of what was said and what is established — it endorses neither, and where the two disagree the page does not resolve it for you.