Independent New Hampshire Journalism · Northfield, NHMost of what anyone films in the sky turns out to be ordinary, and that includes what I filmed over Northfield. The federal ledger says the real obstacle is missing data, and the one case that survives every check will be worth more than all the ones that did not.
On the afternoon of October 25, 2026, I looked up over Northfield to log an airplane and saw something beside it I could not name. It was white, elongated, high and far off, and it kept pace with the plane. Then it seemed to stop. It vanished while I was still watching, came back on the other side of the plane, and carried on. I was sure I had seen something I had never seen before.
It was a regional jet. The public flight record for that minute puts GoJet 4527, a Bombardier CRJ550 climbing through about 27,000 feet, on the same line of sight as Condor 114, an Airbus A330 cruising near 39,000 feet and laying the contrail I took for “the plane.” Matched against the frames, that is GSR’s reading of what I filmed, and the full technical account shows every number behind it. The two lines of sight crossed. A smaller, closer jet sliding past a bigger, higher one looks, to an eye locked on one of them, as if the other one stopped and jumped sides. Two parts of my own account did not survive the record: the stop, and the direction I remembered facing.
I wrote the software that took my sighting apart. That is the job it was built for, and I want every reader who sends a clip to know it before they send one. Most of what this engine reports back will be a plane, a satellite, a balloon, a bird or a smudge on the lens. A sky full of explained objects is what gives an unexplained one any value at all.
Full disclosure: I built the UFO Data Analysis Engine described here and shot the clips it was first run on. The measurements, the flight records and the engine’s limits are laid out in GSR’s October 27 technical report, which uses the clips themselves; this column uses the same cases and adds no new measurements. Its first run on readers’ footage is in The UFO Engine’s First Night Shift.
The one I did not see
Seventeen minutes later I walked down the trail, pointed the phone straight up, and shot two more clips. I noticed nothing at the time. Reviewing them in slow motion that night, I found a point of light that blinked in and out of the frame. My reading, and it is only mine, is a scrap of foil carried high on the wind, catching the sun as it tumbles. When I saw it that way, I could not unsee it.
The engine will not say foil. It says a small sunlit thing crossed the frame on a steady path and showed itself about a fifth of the time, on no fixed beat, which rules out a strobe. The same numbers fit a bird’s wings flashing in the sun at a distance, and the first clip holds a bird doing exactly that. At two pixels, with one camera and no range, nobody can tell foil from feathers, and I wrote the program so it would refuse to pretend otherwise. It gives scores and margins. It never gives a verdict.
Foil is not a joke category. The Pentagon’s UAP office lists balloons of every kind among the objects it closes cases on, from large research balloons to reflective foil party balloons, and it notes that many people who report one describe something clearly identifiable as a balloon from the start.
What the federal ledger says
My own guess has been that this engine will find an ordinary answer for 99 percent of what people send. The government’s count does not support a number that high, and the reason is the most useful thing in this column. The All-domain Anomaly Resolution Office received 319 reports between June 2, 2024, and May 30, 2025. It resolved 114, about a third, every one to something ordinary: balloons, satellites, birds, aircraft, drones, one commercial rocket launch and one person in a jet pack. It sent 191, three in five, to an archive because the reports did not carry enough data to judge. It flagged nine, under 3 percent, for further analysis by specialists.
So the honest version of my 99 percent is this. Of the 128 reports the office did not archive for lack of data, it resolved 114, about nine in ten. Most of what stays open stays open because nobody kept the original file, the time, the place and the direction. That is why GSR asks for the file your phone saved, not a screenshot or a repost.
It is also why nobody should feel stupid when the answer comes back ordinary. Civilian pilots, people who look at the sky for a living, tended to put their sightings that year at 45,000 to 60,000 feet, and the office resolved many of those reports as sunlight flaring off satellites. A new modeling tool let its analysts close 238 reports that way. I was fooled by a regional jet over my own house.
The sky is busier than it has ever been, but it is not busy at every hour. The satellites most people see are lit by a sun that has already set on the ground, which is why they crowd the hours after dusk and before dawn and drop into Earth’s shadow as the night goes on, as Sky & Telescope explains. A light that moves like a star in the early evening is worth checking against a satellite pass before it is worth anything else.
When the government had a reason to stay quiet
Some of the old sightings were real secrets. In the mid-1950s Lockheed’s Skunk Works, the company’s secret design shop under chief engineer Kelly Johnson, built the U-2 spy plane, and the CIA paid for it and ran the program with Air Force help. The CIA did not own Skunk Works; it was Lockheed’s. The plane was tested at Groom Lake in Nevada, the site the map called Area 51.
The CIA’s own history of the program, written by agency historians in 1992 and released with its Area 51 references in 2013 through a Freedom of Information Act request by George Washington University’s National Security Archive, says the U-2’s high-altitude testing set off a tremendous increase in UFO reports. Airliners then flew far below it. At dusk an airline crew could be in darkness while a U-2 overhead still caught the sun. According to the history’s second chapter, Air Force Project Blue Book investigators checked sightings against U-2 flight logs and could not tell the letter writers what they had seen. The history’s estimate is that U-2 and later OXCART flights accounted for more than half of all UFO reports in the late 1950s and most of the 1960s. That is the agency’s own figure, and UFO researcher Stanton Friedman disputed it when it surfaced.
The lesson for a records newsroom is plain. An unexplained light can be a real government secret, and the cure for that is neither belief nor mockery. It is records. New Hampshire has its own example: the Air Force’s record card for the 1965 Exeter sightings concludes in one word, UNIDENTIFIED, and GSR walked through that file line by line.
How to film something worth checking
The engine learned more from my mistakes than from anything that went right. Keep the phone at 1x if you can, or note how far you zoomed: my first clip was shot at about 5x, which made every angle the engine measured five times too large until the flight record caught it, and the same zoom once led it to score an airliner as a balloon. Get a tree line, a roof or a ridge into the frame for a few seconds, because blank sky gives the software nothing to measure the camera’s own movement against. Say which way you were facing, and expect to be checked; I got mine wrong. If someone is with you, have them film from twenty or thirty steps away. One camera gives angles, not distance. Two cameras with times and GPS turn a dot into a size and a speed, and that is the thing most likely to move a clip out of the archive pile.
Why a believer built a debunking machine
I believe there is something to the UFO and ghost phenomena, and I will say plainly that the belief does not rest on any clip I have analyzed. I have studied UFO and paranormal cases since I was ten years old, and New Hampshire has given that study plenty of material, from Betty and Barney Hill on down. That is exactly why the software has to be stricter than I am. A believer who cannot be talked out of a jet is no use to anyone, least of all to the case that might someday be real.
So here is what you get if you send footage. The engine fingerprints the file, checks the time, place and camera, replays the public flight record, measures what moved and how, and scores every ordinary explanation with the margin printed beside it. If the answer is a balloon, GSR will say balloon. The terms for what gets published, and what never does, are spelled out on GSR’s footage page, and the drop box is at Something You Can’t Explain.
Some of the answers will sting. Mine did. But a single clip that survives every check this engine and the federal record can throw at it would be worth more than a million that did not, and the only way to find that clip is to clear the sky of everything else first. Most nights the answer will be ordinary. Mine was a jet. Send yours.
— Dexter Dow, Granite State Report
Fact check
| # | Claim | Status | Source |
|---|---|---|---|
| 1 | The editor’s eyewitness account of Oct. 25, 2026: an elongated white object keeping pace with a plane, seeming to stop, vanishing and reappearing on the other side | ATTRIBUTED | Dexter Dow, written account of Oct. 25, 2026, as published in GSR’s Oct. 27 technical report |
| 2 | GoJet 4527 (CRJ550) climbing through about 27,000 ft and Condor 114 (A330-900) near 39,000 ft were on one line of sight from the camera; their lines of sight crossed; GSR reads the white object as GoJet 4527 beside Condor 114’s contrail | VERIFIED | GSR, Oct. 27, 2026 technical report, fact-check rows 11 and 31–33 (adsb.lol archive traces); the identification is GSR’s inference from the record and the frames, as that report states |
| 3 | Two parts of the editor’s account did not survive the record: the stop and the remembered east-facing direction | VERIFIED | GSR, Oct. 27, 2026 technical report, flight-record section and fact-check rows 25 and 37 |
| 4 | Seventeen minutes later, two clips shot pointing up showed a blinking point; visible about a fifth of the time, on a steady path, with irregular gaps | VERIFIED | GSR, Oct. 27, 2026 technical report, fact-check row 44 (33 of 170 frames, R² 0.963) |
| 5 | The foil reading is the editor’s own; the engine’s measurements do not distinguish foil from a bird’s wings, and the first clip shows a bird whose wings flash in the sun | ATTRIBUTED | Dexter Dow, note of Oct. 27, 2026; GSR technical report, rows 51, 52 and 58 |
| 6 | AARO resolves many cases as balloons ranging from latex or reflective foil party balloons to large research balloons, and reporters often describe their observations as clearly identifiable as balloons in the initial report | VERIFIED | AARO, FY2025 Consolidated Annual Report on UAP, section IV.C |
| 7 | AARO received 319 reports June 2, 2024–May 30, 2025; resolved 114, all to prosaic objects including one commercial rocket launch and one manned jet pack; 191 to its active archive; nine flagged for further analysis; three recommended for closure pending review; two merger actions | VERIFIED | AARO, FY2025 report, executive summary and section III.A |
| 8 | “About a third,” “three in five,” “under 3 percent” and “about nine in ten”: 114/319 = 35.7%; 191/319 = 59.9%; 9/319 = 2.8%; 319 − 191 = 128 not archived, 114/128 = 89.1% | VERIFIED | Arithmetic on row 7 |
| 9 | Civilian pilots’ altitude estimates tended to run 45,000–60,000 ft, and AARO resolved many of those reports as satellite flaring; a new three-dimensional modeling capability let analysts resolve 238 reported UAP as satellite flaring | VERIFIED | AARO, FY2025 report, sections IV.B and VI and executive summary |
| 10 | Satellites are seen after sunset or before sunrise, when the ground is dark and they are sunlit, and can slip into Earth’s shadow as night deepens | VERIFIED | J. Kelly Beatty, Sky & Telescope, “Take a Sat-Seeing Tour” |
| 11 | Skunk Works was a Lockheed division formed by chief engineer Kelly Johnson; Skunk Works engineers created the U-2; its formal name became Lockheed’s Advanced Development Programs | VERIFIED | Lockheed Martin, “Missions Impossible: The Skunk Works Story,” April 7, 2016 |
| 12 | The CIA financed the U-2 (a check to Kelly Johnson at Lockheed in February 1955, then a contract for 20 aircraft); the plane was tested at Groom Lake, Area 51; the CIA history by Pedlow and Welzenbach was written in 1992 and released with Area 51 references in 2013 under a FOIA request by the National Security Archive | VERIFIED | National Security Archive, Electronic Briefing Book No. 434, Aug. 15, 2013 |
| 13 | U-2 high-altitude testing led to a tremendous increase in UFO reports; airliners then flew at 10,000–20,000 ft; at dusk an airliner at 20,000 ft could be in darkness while a U-2 caught the sun; Blue Book investigators checked sightings against U-2 flight logs and could not reveal the cause to letter writers; U-2 and later OXCART flights accounted for more than half of UFO reports in the late 1950s and most of the 1960s | VERIFIED | Pedlow and Welzenbach, The CIA and Overhead Reconnaissance, chapter 2, pp. 72–73, via the National Security Archive. The “more than half” figure is the CIA history’s own estimate |
| 14 | Stanton Friedman disputed the CIA history’s U-2 claim | ATTRIBUTED | NBC News (Alan Boyle), Aug. 2013, citing Friedman’s interview with the Associated Press |
| 15 | The Air Force record card for the 1965 Exeter sightings concludes UNIDENTIFIED | VERIFIED | GSR, Oct. 11, 2026, citing Project Blue Book, National Archives catalog 302543878 |
| 16 | The first clip was shot at about 5x zoom, inferred from the flight record and the frame, so the engine’s 1x angles were about five times too large; before zoom became an input, a 5x clip led the engine to score an airliner as a balloon; sky-only frames leave camera motion unmeasurable; two clips with GPS and timestamps turn angles into distance | VERIFIED | GSR, Oct. 27, 2026 technical report, fact-check rows 34, 35, 48 and 57 and the section on the engine’s limits |
| 17 | The editor believes in UFO and ghost phenomena for reasons other than evidence from his own analysis, and has studied UFO and paranormal cases since age ten | ATTRIBUTED | Dexter Dow, statement to the newsroom, Oct. 27, 2026 |
| 18 | The 114 closed cases by object type: 60 balloons, 44 satellites, 4 aircraft, 2 UAS, 2 birds, 1 missile/rocket, 1 jet pack; balloons and satellites total 104 | VERIFIED | AARO, FY2025 report, Figure 1 (“AARO’s Closed Cases by Identified Object Type”), read from the chart; sum checked |
| 19 | The lines of sight of GoJet 4527 and Condor 114 crossed 14.8 seconds into the clip | VERIFIED | GSR, Oct. 27, 2026 technical report, fact-check row 33 and the flight-path figure reproduced here |
| 20 | Lockheed Martin’s video was published for the U-2 program’s 60th anniversary in September 2015 | VERIFIED | Lockheed Martin YouTube channel, video description, Sept. 14, 2015 |
| 21 | The National UFO Reporting Center asks people not to report Starlink satellite trains | VERIFIED | GSR, Oct. 11, 2026, fact-check row 11, citing NUFORC’s Report a UFO page |
Send the original file your phone saved, with the date, time, place and the direction you were facing. If the answer is ordinary, GSR will say so.
granitestatereport.com/unexplained · granitestatereport@gmail.com
Editor’s note. The Northfield cases, their measurements and the aircraft identification come from GSR’s October 27, 2026 technical report and the flight-record traces cited there; the identification is GSR’s inference from the record and the frames, and the foil reading is the editor’s own. Federal case figures are AARO’s fiscal 2025 annual report. The U-2 history is the CIA’s 1992 history as released through the National Security Archive. This is a signed column; the opinions are the editor’s. Corrections: Granite State Report corrects verified errors promptly and appends a note identifying what changed and when.
