The UFO Release File

Part One — The Method

How a UFO is graded

Every case in Part Two, and the photographs in Part Three, were graded on the same six-axis battery. The axes grade the file, never the witness and never the night:

• A1 — Provenance & chain of custody. Do the originals survive? Is custody documented end to end? Is what was published what the camera captured?
• A2 — Multi-frame / multi-sensor consistency. Does the configuration repeat across frames — and, the far stronger test, across independent sensors? Repetition inside one device on one night is moderate at best, by construction.
• A3 — Sensor & artifact profile. Do the claimed features survive the known artifact mechanisms of the capturing system? A feature that matches an artifact's signature fails as that feature, whatever the rest of the case does. This axis is graded hard: the artifact explanation has to fit the numbers, not just be available.
• A4 — Geometry & light consistency. Is internal geometry coherent and rigid where claimed? Can scale, range, or kinematics actually be derived — or only asserted?
• A5 — Prosaic & hoax discriminators. Has the case been run against the control pile — balloons and winds aloft, aircraft and flight records, satellites and ephemerides, flares, models and compositing? Surviving the checks earns "unidentified." Failing one closes the case. Never having been checked earns nothing.
• A6 — Witness & record corroboration. Independent witnesses? Contemporaneous documentation? Official records created at the time?

And four rules, printed because they are the whole point:

• A claim is tested as the thing the claimant described — its shape, its anchor, its polarity, its motion. Never as a neighbouring, easier-to-measure shape. We learned this rule on our own photographs, the hard way, in public, in Part Three.
• Artifact explanations are hypotheses, not defaults. They earn their place by fitting the measurements. So does the claim. Both directions earn it.
• The debits print at the same size as the credits. Every grade below carries its strongest counter-evidence beside it.
• A negative result stays open, with its attack log. Nothing in this file is "debunked forever." It is graded, dated, and re-testable — and when a better test exists, it runs.

The base rates, stated once, from the official record: AARO's published resolutions run 51.1% balloons and 35.3% satellites of resolved cases, and every case it has resolved to date is attributable to prosaic objects. Unidentified usually means under-instrumented, not extraordinary — and the battery above is built to say which.

Part Two — The Corpus

33 cases, one battery

The documented photo and video record — the cases where media exists (or authoritatively doesn't), where custody can be traced, and where an official body, a named analyst, or a confessed hoaxer put a verdict on the record. The hoax and solved pile is in by design: a corpus without its fraud pile is propaganda, not forensics.

Unexplained on the record — 8

The contested middle — 15

Identified & hoaxed — the control pile, 10

The re-grade, printed whole. On October 30, 2026, all 33 cases were re-graded under the claimant-first method: 30 grades hold · 1 grade corrected (Calvine — split standing) · 2 cases untestable as described (Tehran, Aguadilla), with three held cases carrying flagged untestable sub-claims (a film claim whose footage was never recovered; a physical-effects claim with no controlled medical record; a "mothership" visual with no media). Every historical closure in the corpus survives the audit — because every closure was earned by a test aimed at the claim as made, closed by an external record. The famous failures are not wrong tests of right claims. They are absent tests wearing verdicts, assumed inputs wearing measurements, and media verdicts wearing event verdicts. Claimant-first grading does not overturn the corpus. It shows the corpus where its own empty chairs are — which is what a standard is for.

Part Three — Our Own Photographs

The photographs of William Alexander Aldred, graded in the open

The event

June 13, 2023, 9:01 PM, downtown Orlando, Florida. William Alexander Aldred photographed a configuration of lights in the eastern sky with his phone — a Samsung Galaxy S23 — over 26 seconds: twelve frames survive (five camera masters with intact EXIF, seven later gallery copies carrying the same content through extra processing). The lights held a configuration, changed colour in sequence, and were gone. The same twelve frames were posted publicly twice — December 14, 2024 and May 4, 2026 — and that twice-posting is itself graded below, as a provenance debit, not managed.

His testimony, labelled as testimony: with his naked eye he saw only the red and green blinking lights. He did not see orbs, a structure, or a ring with his eyes — those appeared only in the photographs. Same object throughout, in his account. He is a lifelong watcher of this subject, and he calls one frame of this set the best UFO picture he has ever seen. Testimony is evidence about the man, not measurement of the sky, and this file keeps it in that layer.

The battery result

The kill-list, checked with real data

• Satellites — FAIL. The camera's field was 2.1 degrees wide. Any satellite in low orbit crosses that field in 1.4 to 4.7 seconds. The cluster persisted, in configuration, with its two main lights locked 44–53 pixels apart, across five camera pointings and 26 seconds — ten times too slow. No catalogued object changes colour green-to-red, emitter by emitter, at about two blinks per second. The famous train that week (a Starlink launch) was 41.9 hours old and dispersing in single file.
• Balloons and lanterns — FAIL. The night's upper-air sounding (Tampa, that evening) and the Orlando airport weather record show westerly winds of 4–7 metres per second at every altitude, veering about 70 degrees between 1 and 2 kilometres up. Anything floating free would have drifted roughly 130 metres during the 26 seconds and visibly deformed. A lantern cannot make green light, and nothing flame-driven toggles at a steady twice-a-second.
• Planets — FAIL. The only planets up were Venus and Mars — in the west. The camera faced east. Fixed stars and planets also do not rearrange their configuration in 26 seconds.
• Aircraft — FAIL, closed October 30, 2026. This was the one open door, and this file printed it as open until the record existed. It now exists: the open ADS-B archive's day file for June 14, 2023 was obtained and enumerated in full — 44,655 aircraft trace files; 24 aircraft placed by measurement within 30 seconds of the window; 51 measured points inside the exact 26 seconds — none in the camera's sector. The one airliner inside the field's bearing for the whole window was 243 kilometres away, its entire wingspan subtending 0.4% of the field's width — a single unresolved point, not three separated emitters. The one low pattern aircraft bracketing the window through a reception gap is excluded on emitter geometry and the colour toggle no aircraft performs, not on the gap. The night's only helicopter tracked due south, in the opposite sector. Whatever flew that minute over Orlando, the traffic record says what it wasn't.

The feature ledger — every structure in the photographs, graded

Every structure that has ever been pointed at in these photographs was formalized in the claimant's own description, pre-registered with the test that would pass it, and measured. The full re-evaluation ran October 30, 2026. Here is the whole ledger, debits first-class:

• The ring around the back lights — ARTIFACT, claimant-tested. The annulus is a saturation shell: the camera's red channel clips far wider than green or blue, and the "ring" is where red saturates around the core. This pass tested the ring's placement — a ring belonging to a craft would sit on the craft; a saturation shell sits on the light. The shell's centre is on the light to within 2–4 pixels in both frames. It is the light's own glow. Dead as a physical ring.
• The orbs — ARTIFACT at tested strength. The two discs at deep zoom are the pair's lights, defocused. Their brightness profile sits inside the family of ordinary defocused discs from the same camera — control discs are, if anything, more rim-bright. We also tested for surface markings, structure at a fixed angle, the way markings on a real sphere would sit, with an instrument validated on a planted synthetic: the amber disc has real internal texture, but so do plain defocused control discs from the same phone, at the same strength. Nothing measured separates them from bokeh. The white disc's interior is clipped flat white by the sensor — it contains no information at all, and no test can change that.
• The squares in the dark sky — RENDERING, decided on positive evidence. The terraced, axis-aligned texture in the dark fields is really rendered in the files — and its lattice is phase-locked to the camera file's own 8-pixel compression grid, it appears identically in green and orange control photographs, it is absent over the same camera's dark sky, and it grows every time the file is re-saved. It is the pipeline, not the sky.
• The red structure / "the white stick" — UNTESTED, and we say so. The claim: the red light is a machine — "the red thing with the black, with the white stick." We built two different instruments to test for an attached protrusion at the strength these files could hold. Both failed their validation: a protrusion at half the light's peak brightness hides inside the light's own bloom, and the mathematics absorbs it. An instrument that cannot see its own planted control does not get to print a negative. So this element stands untested at file level — not refuted, not supported. (The downward black element of the same description was tested directly: in all 24 comparable bright objects, the column below the light is never darker than its flanks. There is no dark appendage in these files.)
• The dark dash under the lights — UNEXPLAINED, and better measured than before. In one frame (and the :58-second master) a short, defined dark dash hangs under the light cluster. It was found by measurement, then seen unprompted by five of six blind graders shown the frame with no briefing. This pass ran a census of all twelve frames: the dash-class reading concentrates in those two frames — it is not a family-wide rendering habit of this camera. No artifact account fits it. It doesn't prove anything by itself; it is a small, real, unexplained mark, and it stays on the books as one.
• The dimmed underside of the triangle's lowest light — UNEXPLAINED, sharpened. In all three frames that resolve the triangle, the lowest light's glow is dimmed on its underside — up to 72% below its own surrounding glow level. This pass tested whether the dimming follows the neighbouring lights, the gap between the glows, or a fixed direction on the object itself. It follows none of them cleanly; it sits on the underside in every frame. A real, repeatable asymmetry that no tested geometry owns.
• The tornado from the center — CONTESTED, and the honest version is longer than the headline. The claim, in the claimant's own words from the original post: "It looks like a tornado coming from the center." This department then spent a morning grading it — and graded the wrong shape first, in print: the early attacks tested a bright column (not found) and a circulation cell (a blind panel that caught a planted spiral 6 of 6 caught the frame 0 of 6 — not found), and a later attack tested a helix, a winding corkscrew, and found its handedness dead to the fourth decimal. The claimant then corrected the claim: not a helix — a funnel. A black vortex cone, descending from the center of the three lights, widening as it drops. The funnel test was built for exactly that shape — dark cones anchored at the triangle's measured center, spin scored separately from darkness, a planted cone it had to catch. It caught the plant: recovered one size-step away, winding read perfectly. Then its own validation gate failed — the scrambled-noise null of this frame family manufactures cone-shaped darkness larger than a genuine funnel's, so the instrument cannot certify a verdict in either direction. The result stands as the discipline requires: unresolved by instrument — neither found nor excluded — and OPEN, with its re-opener named (a repaired null, pre-registered before it measures). What the voided instrument saw, printed as description and not verdict: at every developed funnel size, in every resolved frame, the cone darkness reads an order of magnitude below the planted funnel's. The man's own dark-streak pointer, meanwhile, scored a documented win in the first attack: the streak he pointed at was really there — five of six blind graders saw it — even though it was not a tornado.
• The colour toggle — REAL, and it kills the easy answers. Across the video family, the two main lights alternate green and red with a dwell of 0.7–1.0 seconds per state, and one transition frame catches them in mixed states — one green, one red, same instant. The colour change lives in the lights themselves, not in the camera. This single behaviour is why satellites, lanterns, planets, and aircraft fail: none of them do this.
• The triangle — THE STRONGEST ITEM IN THE FILE. Three lights, rigid configuration, resolved in three frames. In the two frames captured about a second apart, the triangle's side-ratios agree to 0.4% (1.479 against 1.473). The formal claim test is graded insufficient frames — three frames cannot certify rigidity the way thirty could — and that insufficiency is printed, not smoothed over. But nothing in the artifact family produces a second frame that matches the first to four parts in a thousand.

The verdict

What would change the grade

Printed in advance, so nobody has to wonder what we're waiting for:

• The funnel re-test with a repaired instrument. The null that voided the first funnel attack has named repairs. When the repair is pre-registered and run, a supported cone upgrades the tornado line from open to measured; a null prints at the same size. (The flight-records item that used to stand at the top of this list is closed — see the kill-list above. It changed the grade: A5 rose. In this file, in public, in both directions — that's the deal.)
• The descent test on the time-ordered video material — the stills' only sequence failed it, on an order that isn't time; the video is the named re-opener.
• Any second witness, instrument record, or independent capture of that minute over Orlando.

What this file is not

It is not proof of extraterrestrials. It is not a claim that William Alexander Aldred met anyone, rode in anything, or knows what this was. His testimony about his own life and experiences is his, is labelled testimony wherever it appears, and is never used as measurement. The internal working dossier behind this file — the raw frames, the full battery, the attack logs — stays with the Department; what is published here is everything the battery earned, including every debit.

The famous cases got graded. His photographs got graded harder — because they're his, and because a grading machine that goes easy on its owner isn't a machine, it's a mirror.

That's the release. Grade everything. Print the debits. Stay in the sky.

— Premium Marketing Solutions · October 30, 2026