Which Flight Records Were Deleted — And Whose
The question this page answers. Researchers on X say the public flight history for the aircraft at the centre of this case was pulled down — most loudly that N102DZ, the jet Erika Kirk invited the public to check, was "ERASED in May 2026 from flightradar24 so now NOBODY can check them." If records really were removed, the removals should show a shape: they should cluster on one nationality of aircraft, or one person's travel dates, or one window of time, or one tracking company. Find the shape and you have found who did it and what they were protecting.
So the shape was hunted for directly. Every aircraft-day this investigation has ever asked an archive about was collected — 15,131 logged queries across sixteen case aircraft, two unrelated control aircraft, four independent archives and 917 distinct dates, snapshot 26 August 2026, 17:42 UTC — and the gaps were sorted four ways: by aircraft nationality, by named person, by date, and by tracking site.
None of the four produced a shape. What produced a shape every single time was the archive's own mechanics — and the controls prove it, because two airliners with no connection to this case fail in exactly the same places.
The answer, in one screen
| The hypothesis | What it predicts | What the data shows | Verdict |
|---|---|---|---|
| They deleted the Egyptian jets (SU-BTT, SU-BND, SU-BTU, SU-BTV, SU-BGM) | Egyptian tails suppressed where US tails are not | FlightRadar24 openly serves them today, labelled "Egypt - Air Force", with 56, 14, 56, 54 and 0 history rows — more than most of the US tails | NOT SUPPORTED |
| They deleted Erika Kirk's dates | Overlap dates where she is claimed present go missing more often | Recovered on 57.1% of Erika-claimed dates vs 88.9% where she is not claimed — Fisher exact p = 0.133, not significant, n = 9 on one side | NOT SUPPORTED |
| They deleted Charlie Kirk's dates | Dates he is claimed present go missing more often | 75.0% recovered where he is claimed vs 56.8% where he is not — the difference runs the wrong way, p = 0.329 | NOT SUPPORTED |
| They deleted the event days | A trough in coverage on the day of each speaking event | Day-of-event is the peak, not the trough: 32.6% for the Kirk-side jets against 25–27% either side | NOT SUPPORTED |
| A tracking site took the pages down | Some aircraft pages gone, others fine | 16 aircraft × 5 sites = 80 page checks. Every subject page and every control page returns HTTP 200 in a browser. Zero removals | REFUTED |
| The 2022 record was wiped | 2022 missing for case aircraft only | 1,461 case-aircraft-days asked before 6 Feb 2023 returned 0 tracks — and so did all 28 control days. The archives do not go back that far | ARCHIVE FLOOR |
| Late-2025 was locked down | A block on case aircraft in the weeks after the killing | HTTP 403 on exactly the same 17 dates for all twelve subject tails, and the controls get 403 on the same dates too | ARCHIVE-WIDE |
| The N102DZ history was purged in May 2026 | No N102DZ data in May 2026 | 26 days of primary position trace inside May 2026 are on disk, from an archive FlightRadar24 does not own | REFUTED |
The one thing that did change in the disputed window is not flight data. N102DZ's FAA registration was re-issued on 16 October 2025 — thirty-six days after the assassination — to a Cheyenne, Wyoming holding company, with a further registry action on 23 December 2025. It is the only case-central aircraft whose registration moved after 10 September 2025. That is a records change, in the right window, on exactly the aircraft the erasure claim is about. It is ordinary in itself, and it is documented below rather than argued from.
The five things that looked deleted, and what each one actually is
Every candidate deletion in this investigation collapsed into one of five archive mechanics. Naming them matters, because from the outside all five look identical to suppression, and three of them were published on this site as suppression before the controls were run.
1. The archive floor — 2022 does not exist. The oldest track either free daily archive holds for any aircraft is 6 February 2023 (airplanes.live) and 24 February 2023 (adsb.lol). Ask either of them about 2022 and you get HTTP 404 forever. This is the largest single "gap" in the investigation and it swallows the entire opening year of the following-planes claim.
2. The 403 band — a date range, not an aircraft list. adsb.lol answers HTTP 403 for a run of dates in late 2025. It is the same date range for every aircraft, case or control.
3. Whole-day holes — and both archives have them. airplanes.live lost 2023-09-11 completely; adsb.lol lost 2023-02-06 through 2023-02-10. On each of those days the archive returns nothing for anything, including daily-flying airliners.
4. The free-tier retention window — 7 days, then 14 days. FlightRadar24 shows a logged-out visitor seven days of history and paywalls the rest. FlightAware shows fourteen. A table that was full on Thursday and empty the next Thursday has not been purged; it has rolled.
5. Coverage, not concealment. A volunteer receiver network only holds what somebody's aerial actually heard. A jet parked at its home stand in Cairo is invisible to a network whose nearest receiver is in another country. The same jet at cruise altitude over the same city is heard from hundreds of kilometres away — which is why the Egyptian tails appear in these archives airborne over Egypt on most of their tracked days and almost never on the ground at Cairo.
None of the five is a person, an agency, or a motive. All five are geometry, retention policy, and volunteer hardware.
How any of this can be said at all: the control test
Nothing on this page may be read as a removal unless an aircraft with no connection to this case failed the same way. That is the whole method, and it is the reason the four pattern tests below can be trusted in either direction.
Two airframes were picked precisely because nobody would suppress them: a Ryanair 737 (hex
4ca7b5) and a Lufthansa A380 (hex 3c6444). They were asked the same questions, on the same
dates, through the same endpoints, by the same code.
| Test | Case aircraft | Ryanair + Lufthansa controls | Reading |
|---|---|---|---|
| Days asked before 6 Feb 2023 | 1,461 asked → 0 tracks | 28 asked → 0 tracks | Archive floor, not removal |
| Days asked from 6 Feb 2023 on | 5,749 asked → 1,516 tracks (26.4%) | 112 asked → 104 tracks (92.9%) | Archives healthy; the gap is about the aircraft |
| The 17 HTTP 403 dates | 403 on all 17, all 12 tails | 403 on the 3 of those dates they were asked | Date-scoped, aircraft-blind |
| 2023-09-11 on airplanes.live | 404 for all 14 tails | 404 for both airliners | Whole-day archive hole |
| Tracking-site pages, in a browser | HTTP 200, 16 of 16 | HTTP 200, 4 of 4 | No page removed from anyone |
The control gap in the third row is the one that needs stating honestly: the case aircraft really are much sparser in these archives than an airliner is — 26.4% against 92.9%. That gap is not evidence of hiding. An airliner flies every day through the densest receiver coverage on earth. A government VIP jet flies a few days a month, much of it outside it. The four tests below exist because that raw gap, on its own, proves nothing at all — the question is whether the shape of the missing days points at anybody.
Test 1 — Is it the Egyptian aircraft?
This was the leading hypothesis, and it is the one the raw numbers most look like they support.
| Fleet | Aircraft-days asked, 2023-02-06 onward | Days with a primary track | Rate |
|---|---|---|---|
| Egyptian SU-BGM, SU-BND, SU-BTT, SU-BTU, SU-BTV | 2,424 | 278 | 11.5% |
| US N102DZ, N40JD, N560TW, N582MM, N872RA, N888KG | 2,763 | 1,047 | 37.9% |
| San Marino T7-ELL | 461 | 90 | 19.5% |
| Controls | 112 | 104 | 92.9% |
The Egyptian fleet is three times sparser than the American one. Three separate checks say that is coverage, not suppression.
First, the tracking sites serve the Egyptian jets openly, today. Probed from a real logged-out browser session on 24 August 2026, FlightRadar24 returns full history for the Egyptian tails and does not hide their operator:
| Tail | HTTP | History rows rendered | Operator as FR24 labels it |
|---|---|---|---|
| SU-BTT | 200 | 56 | Egypt Government / Egypt - Air Force |
| SU-BTU | 200 | 56 | Egypt - Air Force |
| SU-BTV | 200 | 54 | Egypt - Air Force |
| SU-BND | 200 | 14 | Private owner |
| SU-BGM | 200 | 0 — has not flown in 7 days | Egypt Goverment (sic, as printed) |
| N102DZ | 200 | 0 — has not flown in 7 days | Private owner |
| N888KG | 200 | 0 — has not flown in 7 days | Private owner |
| N1KE (control, Nike Gulfstream) | 200 | 0 — has not flown in 7 days | Private owner |
Read that table the way it actually falls. The aircraft showing zero rows are the American ones and an unrelated Nike jet. The aircraft showing full history are Egyptian Air Force. Whatever is happening, nobody is hiding Egyptian government aviation from the public — a site that wanted to would not print "Egypt - Air Force" in the operator field.
Second, FlightAware is publishing their current movements. Its activity log on 24 August 2026 held eleven SU-BTT legs between 11 and 19 August 2026 — Cairo to Mersa Matruh and back, ordinary domestic Egyptian government flying, with times and coordinates.
Third, where the Egyptian jets ARE heard says the missing days are physics. Pulling the position data out of every recovered track and bucketing it by region:
| Tail | Days with a track | Includes positions over Egypt | Over Europe | Over North America |
|---|---|---|---|---|
| SU-BTT | 92 | 70 | 30 | 26 |
| SU-BND | 73 | 51 | 15 | 18 |
| SU-BTV | 45 | 37 | 12 | 12 |
| SU-BTU | 55 | 32 | 23 | 23 |
| SU-BGM | 34 | 28 | 5 | 4 |
Egypt is not a hole in the map. These aircraft are heard over Egypt on the clear majority of the days we hold anything for. What is nearly absent is ground presence at their home base. Across all five tails the recovered ground stops are Wilmington, Provo, Paris, Wichita, Lincoln, Minot, Bangor, Farnborough, Corfu, Preveza, Nice, Sofia, Berlin, Rome, Salt Lake City — and, in Egypt, only El Arish, El Nozha and Sudr Cape, once to three times each. Not one recovered ground stop at Cairo International, the field these aircraft live at.
That is exactly the fingerprint of line-of-sight radio. A transponder at 41,000 feet over Cairo reaches a volunteer aerial in Cyprus, Israel or Greece. The same transponder on the ramp at Cairo reaches nothing. The Egyptian gap is a gap in what a hobbyist antenna can hear, and it lands hard on the days these jets never left home.
Verdict: NOT SUPPORTED. The pattern that looked like Egyptian suppression is the pattern of where receivers physically are.
Test 2 — Is it a person? Erika, or Charlie?
This is the sharpest version of the question, because it is the one the erasure claim actually makes. The 73-overlap spreadsheet records, for each alleged following-flight, whether Erika Kirk is claimed present and whether Charlie Kirk is claimed present. If dates were being scrubbed to protect a person, the dates naming that person would come back empty more often than the dates that do not.
They do not.
| Split | Dates where the aircraft was heard | Dates where it was not | Recovery rate | Fisher exact, two-tailed |
|---|---|---|---|---|
| Erika Kirk claimed present | 28 | 21 | 57.1% | p = 0.133 |
| Erika Kirk not claimed | 8 | 1 | 88.9% | — |
| Charlie Kirk claimed present | 9 | 3 | 75.0% | p = 0.329 |
| Charlie Kirk not claimed | 25 | 19 | 56.8% | — |
| SU-BTT (the lead tail) | 22 | 17 | 56.4% | p = 0.547 |
| SU-BND (the second tail) | 11 | 5 | 68.8% | — |
All three are null results, and two of them run backwards. The Charlie-present dates come back better than the rest, not worse. The Erika-present dates do come back worse — but the comparison group is nine rows, the test does not clear significance at any conventional threshold, and the imbalance has an ordinary cause: the not-claimed rows are concentrated in later years the archives cover better.
Verdict: NOT SUPPORTED for either person. There is no person-shaped hole in this dataset.
Test 3 — Is it the dates? The event windows?
If somebody were removing the days that matter, the day of a TPUSA or Charlie Kirk speaking engagement would be the day that goes missing. The sweep queried a ±7-day window around 131 speaking events, which makes this directly measurable: line every queried day up by its offset from the nearest event and read the recovery rate down the column.
| Offset from event | Egyptian SU-* | Kirk / TPUSA N-* | All subject tails |
|---|---|---|---|
| −3 days | 19.6% | 65.2% | 37.2% |
| −2 days | 8.3% | 25.8% | 17.7% |
| −1 day | 9.1% | 27.0% | 18.7% |
| Day of the event | 7.9% | 32.6% | 21.5% |
| +1 day | 6.6% | 31.7% | 20.3% |
| +2 days | 7.5% | 26.7% | 18.1% |
| +3 days | 6.2% | 24.4% | 14.6% |
The event day is the peak on the Kirk side and flat on the Egyptian side. Charlie Kirk's own aircraft are best documented on the days he was speaking, which is what you would expect of jets that actually flew him there. Nothing dips at zero.
Note the second thing that table says, which cuts against this site's own thesis and belongs here anyway: the Egyptian tails show no elevated presence on event days either — 7.9% on the day against 9.1% and 6.6% on the days either side. That is a fact about the following claim, not about deletion, and it is reported in full on Overlap Recovery.
Verdict: NOT SUPPORTED. No date-shaped hole.
Test 4 — Is it a tracking company?
Five tracking sites, sixteen case aircraft, four unrelated controls. Eighty page checks.
Asked by a script, three of the five sites answer HTTP 403 — FlightRadar24, RadarBox and Planespotters — for every aircraft, subject and control alike, including FlightRadar24's own home page. That 403 is a bot filter. This site published it as a documented removal and had to retract it; the full account is on What A 403 Actually Means.
Asked from a real logged-out browser, the picture is unanimous:
| Site | Subject aircraft served | Control aircraft served | Any removal found |
|---|---|---|---|
| FlightRadar24 | 16 / 16 (HTTP 200, identity block intact) | 4 / 4 | None |
| FlightAware | 16 / 16 | 5 / 5 | None |
| ADS-B Exchange | 16 / 16 | 5 / 5 | None |
| RadarBox | Script-blocked — tells us nothing either way | Script-blocked | Untestable |
| Planespotters | Script-blocked — tells us nothing either way | Script-blocked | Untestable |
Not one page in this investigation has been removed from any tracking site. The per-aircraft
provenance record records removed_from: [] for all sixteen tails.
Verdict: REFUTED. No company-shaped hole.
What the empty tables actually are: the retention window, measured
The strongest-looking evidence for deletion was never a 404. It was watching a page that used to show flights stop showing them. That happened, it is on record in the Internet Archive, and it has an exact explanation.
N888KG is the clean case, because the Internet Archive happened to capture its FlightRadar24 page fourteen times across five months:
| Snapshot date | Flight rows | Oldest flight shown | Age of that flight |
|---|---|---|---|
| 11 Sep 2025 (×3 captures) | 2 | 10 Sep 2025 | 1 day |
| 12 Sep 2025 | 2 | 10 Sep 2025 | 2 days |
| 13 Sep 2025 | 2 | 10 Sep 2025 | 3 days |
| 18 Sep 2025 (×4 captures) | 0 | — | 10 Sep is now 8 days old |
| 19–20 Sep 2025 | 0 | — | — |
| 2 Nov 2025 | 0 | — | — |
| 28 Jan 2026 | 0 | — | — |
The rows vanish on the first capture after the 10 September flight crosses seven days old. Not on the day of the killing. Not on the day the story broke. On day eight.
Test that against every FlightRadar24 capture on disk, across five aircraft and six years:
Oldest flight visible in any archived FlightRadar24 free page: 7 days. Zero exceptions in 22 captures spanning 2020 to 2026, Egyptian and American tails alike.
Oldest flight visible in any archived FlightAware free page: 14 days. Zero exceptions in 16 captures spanning 2011 to 2016.
Two different companies, two different hard ceilings, never once exceeded. FlightRadar24's own page prints the reason next to the empty table: "More than 7 days of history is available with an upgrade to Silver (90 days)…" — and paywall=1 on every one of the twenty rows probed, subject and control alike.
This is the finding that dissolves the headline claim. A free FlightRadar24 page for N102DZ never held more than seven days of history at any point in its existence. There was no multi-year log on that page in April 2026 for anybody to erase in May 2026. The multi-year tally the 73-overlap spreadsheet was built from did not come from that page and could not have.
And the data for the disputed month is on disk anyway
The claim is that N102DZ's history was erased in May 2026 so that nobody could check it.
This investigation holds 26 days of primary ADS-B position trace for N102DZ inside May 2026 — 2, 3, 4, 5, 6, 7, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 20, 21, 22, 23, 24, 25, 27, 28, 29 and 31 May — pulled from globe.airplanes.live, a volunteer network FlightRadar24 does not own, does not supply, and cannot edit. Coverage continues either side: 25, 26, 29 and 30 April before it, and 1 through 5 June after.
Whatever did or did not happen at FlightRadar24 in May 2026, the aircraft's movements that month are recoverable by any member of the public, and this site has them.
The one thing that genuinely changed in the window
The sweep did turn up a real, dated, documentary change in the disputed period — and it is not flight data.
Pulled from the FAA Releasable Aircraft Database on 24 August 2026, all ten US-registered tails this investigation tracks, sorted by last registry action:
| Tail | Registrant of record | Certificate issued | Last registry action |
|---|---|---|---|
| N55906 | OPOLINER LARRY — Winter Park, FL | 2005-05-12 | 2023-03-17 |
| N1098L | LASAI AVIATION II LLC — Manassas, VA | 2023-05-15 | 2023-05-23 |
| N2100L | LASAI AVIATION II LLC — Manassas, VA | 2023-07-10 | 2023-07-10 |
| N872RA | A&E HAWKER LP — Santa Ana, CA | 2019-05-16 | 2023-08-04 |
| N888KG | N888KG LLC — Lehi, UT | 2022-07-26 | 2023-08-18 |
| N560TW | T W LEWIS-AVIATION LLC — Scottsdale, AZ | 2014-03-07 | 2023-10-13 |
| N582MM | DOJA CONTRACTING LLC — Mesa, AZ | 2024-01-25 | 2024-01-25 |
| N40JD | NUP AVIATION LLC — Princeton, IL | 2024-08-22 | 2024-08-22 |
| N102DZ | ELDERBERRY CHESTNUT LLC — Cheyenne, WY | 2025-10-16 | 2025-12-23 |
| N59906 | ALPHA WHISKEY 2 LLC — Brandon, MS | 2026-04-04 | 2026-04-04 |
N102DZ is the only case-central aircraft whose registration moved after 10 September 2025. The certificate was issued thirty-six days after the assassination, to a Wyoming holding company, and a further registry action followed on 23 December 2025 — months before the reported May 2026 "erasure", and on precisely the tail number the erasure claim names.
State what this is and what it is not. It is a public federal record of a registration change, in the right window, on the right aircraft. It is not a deletion, and it is not evidence of one. Aircraft are re-registered constantly; ownership passes; a registrant is frequently a holding LLC and a holding company is not a person, so this site does not go behind one to name individuals. It is also worth saying that the community tracking databases still list the previous name for this tail — which is one prosaic way a researcher could look at two sources, see two different owners, and reach for the word "erased".
What makes it worth publishing is that it is the only dated change anywhere in the fleet that falls in the disputed window at all. Every other candidate on this page dissolved into archive mechanics. This one is a real record, and it is left standing as a question rather than an answer.
What is still genuinely gone
Sorting the gaps did not make them go away. Four things remain unrecoverable, and the honest version of this page has to lead with them as prominently as the refutations.
1. All of 2022, from the daily archives. The following-planes claim opens with alleged overlaps in October, November and December 2022 — New York, Chicago, Boston, Wichita, Wilmington, St Louis. Neither free daily archive can be asked about any of them. The only free route into 2022 is the ADS-B Exchange monthly sample, which holds one day per month, the 1st. It can fix a position on 1 October 2022. It can say nothing whatever about 5 October 2022. Twenty of the eighty-five alleged overlaps sit under this floor and cannot be tested by anyone, for or against.
2. Whose feet were on the aircraft. This is the one that matters most and the one no archive anywhere will ever fix. A trace proves presence. It never proves purpose and it never proves occupancy. Twenty-six days of N102DZ tracks in May 2026 do not place any named person on any leg. Erika Kirk's itinerary is the missing document, and no backup produces it — which is precisely why she was able to issue the invitation, and precisely why taking her up on it does not settle anything either way.
3. Twenty-two alleged overlap dates the archives simply never heard. Inside the covered period, twenty-two aircraft-and-date pairs return nothing from either network after the GitHub off-site backup was also asked and also came back empty. That is a coverage fact, not a finding. Parked with the transponder off, outside receiver coverage, and the claimed date is simply wrong all look identical from here — and several of these rows are already marked inaccurate in the spreadsheet's own audit column.
4. What any of these pages displayed in May 2026. A page that loads today cannot testify about a table's contents eight months ago. "The page was taken down" is refuted. A narrower claim — that history was specifically purged and later restored — is beyond every method used here, in either direction, and no page on this site may assert it.
What would actually prove a deletion
So that this page is falsifiable rather than merely reassuring, here is what would overturn it. Any one of these would be a finding:
- A tracking-site page that loads for a control aircraft and 404s for a case aircraft, in a real browser, on the same day. Eighty checks so far; zero.
- A dated Internet Archive capture showing rows, and a later capture of the same page showing fewer rows for flights still inside that site's retention window. The N888KG series looked like this until the seven-day arithmetic was done.
- An aircraft-day present in one volunteer archive and absent from another, where the same archive holds a daily-flying control on that date. Ten such cases exist; every one is explained by a whole-day archive hole that hit the controls too.
- An FAA LADD or PIA privacy filing for one of these tails. FlightRadar24 still prints
N102DZ's real Mode S address
A00C85, matching the FAA record exactly, so no PIA substitution is in force. A LADD blocking request would be the ordinary, lawful, documented way an owner removes a tail from public tracking, and it would be visible. - A paid-tier record for the disputed window. FlightRadar24 Silver shows 90 days, Gold and Business more. This whole investigation ran on free sources. A single paid subscription would reach into windows no method here can, and that is the cheapest open lead on the page.
Counterpoints
"You only tested free tiers, so of course you found retention limits." True, and stated. A paid FlightRadar24 or FlightAware tier sees further back and might show something this method cannot. What the free-tier finding does settle is narrower and still decisive: the specific claim was that a public page was erased so the public could not check it, and that public page never held more than seven days to begin with.
"Two volunteer networks that both use the same software could both be missing the same thing." Fair, and it is why the controls carry the weight rather than the agreement between the two archives. It is also why the archives were cross-checked: on the aircraft-days both hold, the two independent networks agree to the second on timestamps and to four decimal places on position.
"An absence of evidence for suppression is not evidence of absence." Correct, and this page does not claim otherwise. It claims something smaller and checkable: four specific suppression hypotheses were stated in advance, tested against controls, and none of them survived. Somebody may yet find the shape. It is not in these 15,131 queries.
"The sweep is still running, so these numbers will move." They will. An automated job was still fetching event windows when this snapshot was taken at 17:42 UTC on 26 August 2026. The counts will grow; the four null results are structural and are not expected to.
Reproduce every line of this page
Every number above comes from files committed in this repository. Nothing was retrieved by hand and nothing requires an account.
# the raw evidence: one .meta.json per query, recording the exact URL, HTTP status,
# byte count and UTC retrieval time - misses included, on purpose
ls site/docs/Planes/*/data/recovered/*.meta.json | wc -l
# the archive control test - Ryanair and Lufthansa on the same dates, same endpoints
cat site/docs/Planes/following/apis/public_open_source/data/recovery/archive_control_probe.json
# the tracking-site page test - 5 sites x (5 subjects + 5 controls)
cat site/docs/Planes/following/apis/public_open_source/data/recovery/page_control_probe.json
# the browser pass, which is the only way to ask FlightRadar24 anything
cat site/docs/Planes/following/apis/browser_capture/captures/fr24_page_availability_2026-08-24.tsv
cat site/docs/Planes/following/apis/browser_capture/captures/fr24_n102dz_removal_test_2026-08-24b.tsv
# the archived tracking pages the 7-day and 14-day windows were measured from
cat site/docs/Planes/following/apis/public_open_source/data/recovery/wayback_flight_rows.json
# the per-aircraft provenance record - removed_from is empty for all sixteen tails
cat site/docs/Planes/following/apis/public_open_source/data/recovery/fleet_recovery_matrix.json
# the FAA registry pull behind the registration table
cat site/docs/Planes/following/apis/government/data/faa_registry/tracked_tails_faa_master.json
# re-run the sweep itself
cd site/docs/Planes/following/apis/public_open_source/code
node recover_erased.js --tail N102DZ
python3 fetch_event_windows.py --run --side following --window 7
Related pages
- Flight Data Recovery — the full sweep, source by source
- What A 403 Actually Means — the control test written out as a method
- Per-Aircraft Recovery Status — one section per tail
- Overlap Recovery — all 85 alleged overlaps against primary data
- Erika Flight Logs Erased — the claim this page was built to test
- N102DZ — the aircraft at the centre of the erasure claim
- Getting The Data — the four data lanes and what each one can reach
Where this page fits
This is the page that says which flight records were actually removed, and the answer is that after 15,131 archive queries and eighty tracking-site checks, none were — the gaps sort into an archive floor, a paywall window and volunteer receiver coverage, and every one of them fails the same way for a Ryanair 737. That matters beyond aviation, because the same discipline decides which of this investigation's other suppression claims hold: the 403 that was published as a removal and retracted is the cautionary case, while a named witness saying agents phoned back to confirm he had erased his own 4K footage is a human on both ends of the line and survives every control a status code cannot. Between those two poles sit warrants sealed into 2026, the ATF fragment that could neither be identified nor excluded, and the Egyptian-registered aircraft themselves — which are still openly published by every tracking site on earth, and are therefore an argument that has to be won on where they flew rather than on who hid them.