Almost nothing in the modern UAP record was seen with an unaided eye. It was seen through a targeting pod, a radar scope, a phone camera or an infrared sensor, and every one of those machines has behaviour of its own. That behaviour lands in the same frame as the object, and it does not come labelled.
This is not an argument that the sensor is always the answer. It is a rule about order of operations: you cannot grade the object until you have graded the instrument, and in the released record the instrument’s contribution is usually documented, sometimes by the government itself. What follows is what this site has found in seven separate files, sorted by which part of the machine did the work.
The display is not the object
An infrared feed shows heat, not light, and it shows it in one of two polarities: white-hot, where warm is bright, or black-hot, where warm is dark. PR38, the 106-second Middle East clip from 2013 that press coverage made famous, runs in black-hot. So the pale field filling the screen is the cooler sky, and the dark spot everyone was looking at is a warm object displayed dark. The long pale streak across the lower frame is, by the same logic, a cooler feature, consistent with a contrail or a band of cloud rather than an exhaust plume. The display registers that the object was warmer than its background. It does not register what the object was.
The polarity can also change mid-clip. PR086, 34 seconds over the East Coast in December 2019, shows an operator switching contrast modes while tracking. When the mode flips, the object inverts: a bright blob becomes a dark blob. A viewer who does not know the switch happened reads that inversion as the object itself changing. It is the display being toggled. GIMBAL prints BLK in its own symbology for the same reason.
The optics draw shapes that are not there
Paused on its best frame, PR38 shows a star with eight sharp arms of alternating length, held inside a tracking reticle. As a description of the screen, that is fair, and it is what the press reported. As a description of the object, it is the lens.
A bright, compact source captured by a camera is smeared into a star pattern by the camera’s own optics, the effect anyone who has photographed a streetlamp at night has seen. The number of points and their relative lengths are set by the geometry of the lens, the aperture and the sensor, not by the source. The proof is in the clip itself: across the frames the star sits at different positions and focus states, and the spike pattern shifts and softens with it. An object does not change its number of arms because the camera refocused.
The mount draws motion that is not there
GIMBAL is the most argued-over case of this kind. A roll-over-nod gimbal’s image-up is the projection of its roll axis, so holding the horizon level costs a derotation, and the measured rotation moves in about five discrete jumps with one long hold, each landing on a visible jolt in the background clouds. That is what a heavy roll gimbal parking and then catching up looks like. The counter-argument, from the pod’s own patent, is that a stepped roll would itself be anomalous, and this site keeps both positions on screen rather than picking.
PR051 is the cleaner illustration because the two readings arrived in the same release. Its uploader titled it “Syrian UAP instant acceleration.” AARO’s description of the identical file records that the sensor panned to hold an area of contrast centred for nineteen seconds, that at the twenty-second mark it stopped tracking, and that the object then rapidly exited the right of frame. While a sensor tracks, the camera cancels the object’s motion, which is why it sits almost still on screen while the background slides past. When the track drops, that cancellation stops, and the object crosses the frame within a second or two whatever its actual speed. The government’s own file and the uploader’s title describe the same two seconds and disagree about what they mean, and the release publishes both without ruling.
The refresh rate manufactures speed
This is the oldest one in the file, and the site has now audited it twice, fifty-two years apart.
In Washington 1952, a plan-position indicator repaints the scope every ten seconds and remembers nothing in between. Join a return that faded in one place to one that bloomed in another, divide by the sweep interval, and you get the impossible numbers that ran in the newspapers. The instrument’s speed machine reproduces the famous 7,000 mph from ordinary subtraction.
In the 2004 Nimitz file, the load-bearing claim is a drop of 28,000 feet to the sea in less than a second, pinned by the Scientific Coalition for UAP Studies to 0.78 seconds, which is one display interval. A screen that updates in sweeps cannot distinguish a real sub-second transit from a track leaving one altitude gate and appearing in another between updates. The number that carries the entire case is a memory of a display.
Neither of these says the object was slow. Both say the number was produced by subtracting two screen states, and that the screen is not a continuous record of the sky.
The numbers that would settle it are the ones removed
The pattern that runs under all of the above: in file after file, the layer that goes missing is the measurement layer.
PR38’s telemetry sits under black redaction blocks. On an un-redacted feed those areas would carry the time, date, sensor mode, magnification, platform altitude, slant range to target and coordinates. Without them the clip carries a picture and not a measurement. GIMBAL never prints RNG at all, because the pod tracked passively, and that absence is the axis of its entire debate. The Nimitz radar recordings have never been released. And PR072, the 17-second phone clip near Karaganda International Airport, arrives with AARO stating that the media was digitally altered before upload and is presented as received, without itemising what changed. The luminous object and its radiating trails sit on the far side of an undocumented edit.
What this does not license
The argument cuts both ways, and this is the part that usually gets dropped.
Every mechanism above explains an appearance. None of them proves the object was ordinary. The briefing on PR051 makes the point plainly: a genuinely fast or genuinely strange object would also slide out of frame when tracking stopped, in exactly the same way, and the clip cannot distinguish the two. The stepped-roll account of GIMBAL is contested by the pod’s own patent. The Washington soundings that were supposed to carry the official inversion answer showed bending of roughly a degree or two, real but short of the textbook trapping gradient, and that awkward number stays on screen in the simulation rather than being tidied away.
Some of this evidence even runs the other direction. An operator flipping contrast modes is doing purposeful work to hold and improve a track, which is a small mark in favour of PR086 being a real, attended observation rather than a stray artefact.
A noisy instrument is not the same as an empty sky. The rule this site works by is narrower than the sceptical reflex and narrower than the believer’s: grade the instrument first, then the object, and say plainly when the instrument’s own record has been altered, as the government did with PR051 and PR072. What survives that order of operations is worth arguing about. What does not survive it was never about the sky.
Sources for this page
- DOW-UAP-PR38, PR086, PR051 and PR072, US Department of War, PURSUE Releases 01 and 02, hosted at war.gov/ufo
- AARO UAP Records, All-domain Anomaly Resolution Office
- CAA technical study of the 1952 Washington radar returns (Borden & Vickers 1953)
- Scientific Coalition for UAP Studies (2019), forensic analysis of the 2004 Nimitz encounter
- The instruments on this site: Gimbal, Tic Tac Kinematics, Washington 1952