The Western U.S. Event

A file published whole

Most of this database is reconstruction. The documents arrive decades late, in fragments, with the working removed. This one arrived intact: the agency’s own analysis, the five witness narratives it rests on, and a map of where the four incidents sit relative to one another, all published together by the department that produced them.

The analysis does not claim a resolution. It says, in the Director’s own signature block, that the case remains unresolved, and it says exactly how much of it is unresolved and why.

That is the case this project was built to hold. It is also, on our scale, a class C case with material gaps — and the distance between those two sentences is the whole point.

Four incidents

The map published with the file names them.

Incident 1, “Orbs Launching Orbs.” At around dusk, a luminous orange orb appears for one or two seconds, releases a cluster of two to four smaller red orbs, and vanishes. It repeats over several hours. Witness 2 counted the cycle about five times between 19:00 and 19:30. Witness 3 described the red orbs as “grapes being expelled from a basketball.” Witness 4 lost count after five. The red orbs move off in coordinated formations; in at least one case one of them stayed above a ridgeline for hours.

Incident 2, “Fiery Orb.” A large orb against a hillside at close range. Witness 1 ranged the hillside behind it at about 1,100 metres with a rangefinder on a later day, and put the lit disc at 20 feet or more across: “a circle of light that looked like a swirly pattern of bright lava,” with no structure visible at its edges and no glow cast on the ground behind it.

Incident 3, “Dark Kite.” A thin, dark, kite-shaped object carrying one red light and one white light, seen at close range.

Incident 4, “Translucent Kite.” The same shape, transparent. Witness 1 watched one drift past at 15 to 20 feet up and could see a star, the crest of the hill and the vegetation through it, “almost like looking at something through water.”

Two other threads run through the narratives and are not in the map’s four: objects that behaved like road vehicles and were not — a single centre headlight that stayed ahead of a pursuit at 80 mph and loosely followed the road rather than the lane; a pair of lights that left the highway across open desert without raising dust or leaving tracks — and the floating filament the agents twice describe, which the older literature calls angel hair and nobody collected.

The sixty and the forty

AARO did the work that makes a case checkable. It cross-correlated the narratives against commercial and military flight logs, radar, ADS-B and spatial estimates. Military aircraft were in that airspace and were dispensing infrared countermeasure flares as part of a standard exercise. Position, apparent direction of travel and time of observation line up with the recorded tracks for about 60 percent of the reported activity.

The agents, who know what flares look like, said these were not flares. AARO notes in the same paragraph that the specific countermeasure types carried that night do not look like the illumination flares people picture. That is the correct way to handle a witness objection: record it, then say why it may not be decisive.

The remaining 40 percent does not close. For those observations radar and ADS-B put no known aircraft inside the observers’ estimated line of sight. And a flare cannot hover: a red orb reported stationary above a ridgeline for several hours is incompatible with the burn time and descent rate of any flare in the inventory.

AARO’s own placeholder for that remainder is “Unrecognized Technology (Pending)”, and it immediately disarms it: the assessment rests solely on narrative testimony and on the elimination of competing hypotheses, and is unsubstantiated by technical data or physical evidence. An agency writing that sentence about its own leading hypothesis is doing the job properly. Nothing in this file supports reading that 40 percent as anything beyond unexplained.

The pictures are not pictures

Ten images accompany this event, published as “Digital Renderings”, and two short films published as “Digital Recreations”. They look like photographs of a desert at night with orbs in it. They are not.

Witness 2 states it in the file: they are AI-generated images, made from prompts the witness supplied, containing only generic landscape, with the orbs sized for legibility rather than to scale, and produced two and a half years after the events. The notional map carries the same declaration for its own imagery, naming the tool.

So the visual record of this case is: nothing. Not one photograph, not one frame of film, not one sensor track of the phenomena. The imagery that exists is illustration of testimony, generated after the fact, and treating it as evidence of an object would be the single easiest mistake to make with this file. It scores zero on our optics axis for that reason, and it would score zero if there were a hundred renderings instead of ten.

The renderings are worth publishing. They are worth publishing labelled.

One round red light, one square white light

The most striking thing in the file is a detail nobody would invent twice.

Witness 1, following a car at night, notices its tail lights are wrong: “one taillight was red and one was white… one was round and one was square.” The car then leaves the road, hovers, raises no dust, and later drifts past in the air with its lights off.

Witness 5, at about 04:30, sees two lights a metre off the ground over the pavement: “One was white and square, the second was round and red.” Through night vision the lights go out and a thin line remains, about four feet wide, horizontal to the earth. The sketch is in the file.

That is the same object, described by two agents in almost the same words, and it matches the map’s “Dark Kite” — one red light and one white light on a thin dark shape. It is the strongest internal corroboration in the whole package.

It is also where we have to stop. The narratives do not say whether those two agents were partners in the same vehicle. Two accounts from one team are one channel, not two. AARO withheld the pairings, so the correspondence stands as consistency, not as independent confirmation, and the multi-channel axis is scored accordingly.

Why this scores where it does

Documentation scores 5. It is the best-documented case in this corpus: a signed agency analysis with its reasoning exposed, five first-person narratives published unedited — one witness calls contrails “chemtrails”, another reaches for “portals” and then carefully says he is not claiming that — a map, sketches, and a stated split between what is explained and what is not. Nobody had to pry this out.

Radar scores 1 and optics 0. Radar and ADS-B covered the airspace and were used in the analysis, but no instrument detected the phenomena; a negative return is an argument, not a measurement. Physical trace scores 1: Witness 5 reports fresh, sharp, symmetrical markings in the ground and floating filament, and neither was collected or analysed.

So a case with a government file of exceptional quality lands in class C, material gaps, on an overall of 2.2. That is not a contradiction. It is the thesis of this database stated in one line: the quality of the paperwork and the quality of the evidence are different quantities, and this subject keeps mistaking the first for the second.

What would settle it

Four things, in order of how much they would move the score.

The video. Witness 2 refers twice to footage of one of the encounters, and to audio of the agents’ reactions. AARO’s analysis of Incident 1 states that the reporting agents collected no video, photographic or technical data, and that statement is scoped to that incident — but a recording referred to by a witness exists somewhere, and it has not been published. It is the only item that could move optics off zero.

The radar and ADS-B extract. The 60/40 split is the load-bearing number in the entire file, and it is asserted rather than shown. The underlying tracks would let anyone check it.

The pairings and the geometry. Which agents were together, and where each team stood. Without them six witnesses cannot be resolved into a count of independent channels, and no line of sight can be reconstructed.

The ground. Somebody photographed markings, or did not. If a site was worked the next morning, the result belongs in the file.

None of these require declassifying a location. All four are the difference between a case that reads well and a case that can be checked.