Kevin Randle was browsing YouTube when he found a video claiming that a surgical nurse named Margaret Ellen Dodd had been stationed at Roswell in 1947, tasked with photographing the bodies of recovered alien beings. The story had a specific texture: Dodd had confided her account only once, in 1985, to an Albuquerque UFO researcher. The researcher wrote it up. The government confiscated the report when he died in 1993. But not before he made eleven copies and distributed them.

For anyone who has spent decades working the Roswell case, that kind of detail — granular, atmospheric, resistant to verification — is a familiar pattern. Randle, who has spent the better part of forty years investigating the 1947 incident and publishing extensively on it, did not recognize the name. That, in itself, was a signal.

The story being told here is not about whether something crashed at Roswell. It is about a specific and recurring problem: how fabricated or dramatized UFO claims, now amplified by AI-assisted YouTube channels, feed into a comment ecosystem where fiction and testimony become indistinguishable — and how researchers who actually know the record can still trace the seams.

What the record shows

Randle's account of his investigation — published July 25, 2026 on his long-running blog, A Different Perspective — is worth following closely, because it maps exactly how this kind of claim gets tested.

The first check is the '47 RAAF yearbook, assembled at Roswell Army Air Field during the summer of 1947 by Walter Haut, the base's public information officer. Margaret Ellen Dodd does not appear in it. She does not appear in the morning reports — the administrative documents that would log any nurse's arrival at, departure from, or absence at the base. For an officer and lieutenant, as the video described her, that absence is hard to explain away.

Randle emailed Tom Carey, a Roswell researcher based at the Roswell UFO Museum, who had been working the case for decades and had an extensive personal archive of local records. Carey replied the same day. He had searched the '47 yearbook and found one Dodd — "Pfc. Willy Dodd," listed in Squadron S, composed entirely of Black airmen. Dodd's account described an officer, so Carey set that aside. He found a "Harry Dodd" listed in a photocopy of the 1947 Roswell City Directory as an orderly at Saint Mary's Hospital. A search across New Mexico and Arizona records produced no connection between any of those names and a surgical nurse matching the claimed profile.

The closest Carey found was an "M.E. Heid" in his copy of the 1993/1994 Roswell White Pages. But Randle's original research had indicated the nurse in question died in 1991 in Tucson, Arizona — a detail that matched neither the name nor the record Carey located. Dead ends all the way down.

Carey did one more thing. He forwarded Randle's inquiry to Tony Bragalia, another researcher with long ties to the Roswell investigation. Bragalia pulled the channel description for "Contact Redacted," the YouTube account that had posted the video. What he found was unambiguous.

The channel's own disclaimer — highlighted by Bragalia and quoted in full in Randle's post — reads: "This channel features exploratory interpretations and speculation about unexplained phenomena, presented through investigative-style storytelling rather than established fact. The perspectives and visual representations are dramatized or creatively constructed for storytelling purposes. Some visuals may be generated, enhanced, altered, or reconstructed using automated or generative artificial intelligence tools to create a more immersive viewing experience. The content presented here should not be considered factual and is intended solely for commentary and entertainment."

That is the channel's own language. It is not skeptical analysis applied from the outside. The content creator explicitly disclaimed any factual basis for the material.

Randle's reaction was notably restrained, and more telling for it. "This," he wrote, "should end any acceptance of the tale as real."

"should end any acceptance of the tale as real."

It has not.

The comment section problem

What prompted Randle to write the post at all was not the video itself — he expected that kind of content to circulate. It was the comments. Viewers were treating the Dodd story as authentic history, grafting it onto other Roswell claims, cross-referencing it with accounts they had absorbed from elsewhere. Glenn Dennis came up repeatedly.

Dennis is a name that carries real weight in Roswell circles — and real baggage. A former mortician who claimed involvement with the aftermath of the 1947 crash, Dennis told investigators he had been warned off by military personnel at the base hospital and that a nurse he knew had described an examination of non-human remains. He named the nurse, eventually, as Naomi Self, sometimes spelled Selff.

The problem, documented at length by Randle and others over many years, is that the nurse does not exist. Vic Golubic, a researcher who pressed Dennis on the matter, obtained a roster of approximately 125,000 nurses who had served in the Army. Naomi Self appeared on none of it.

When confronted with that, Dennis changed course. He said the name he had provided was not her real name. Randle notes, pointedly, that Dennis had told him nothing of the kind when the account was first given — and told nothing similar to investigators Karl Pflock or Philip Klass, whose work in the early 1990s would have circulated the caveat if it had existed. A retroactive disclaimer, Randle's account suggests, is not the same as a consistent position.

Dennis also claimed the nurse had been killed in an aircraft accident that took four other nurses with her. In the early 1990s, both Randle and Don Berliner — who later co-authored a book on Roswell with Stan Friedman — searched for any evidence of such a crash. They found nothing.

The pattern here is not subtle. A claim is made. It is investigated. The underlying evidence fails to materialize. The claimant adjusts. The adjusted version is accepted by a new audience that hasn't tracked the adjustments.

The Dodd story follows the same arc at higher speed. A fabricated account, posted with an AI-assisted production aesthetic and explicit entertainment disclaimers the algorithm buries below the fold, gets shared without the disclaimer. Viewers encountering it cold have no reason to know it was never presented as fact in the first place.

Why the gaps matter

Several questions that Randle raises don't resolve cleanly, and they are worth naming.

If the Albuquerque researcher made eleven copies of his report and distributed them before his death in 1993, who received them? The Roswell case attracted an active network of researchers through the late 1980s and 1990s. Stan Friedman — perhaps the most prominent name in that network, and a researcher whose 1992 book with Don Berliner would logically have incorporated any credible witness account predating the case's wider public attention — apparently received no copy. Randle flags this specifically. It is a reasonable question. A pre-1990 account, if genuine, would have been exactly the kind of primary material Friedman sought.

There is also the hospital itself. Randle notes that the base hospital at Roswell in 1947 was not the consolidated brick structure built in the 1950s. It was a series of dispersed buildings. A nurse who had genuinely been assigned there, he observes, would have known that. The video's account, he implies, does not reflect that physical reality.

Neither detail proves fabrication by itself. But together with the record gaps — no yearbook listing, no morning reports, no independent corroboration from any of the eleven supposed copies, and a content creator who told readers upfront not to treat the material as factual — the evidentiary picture is not ambiguous.

What would change it? A copy of the Albuquerque researcher's report, in the hands of a named researcher who could authenticate it, would demand serious attention. So would any documentation placing Margaret Ellen Dodd at Roswell — a service record, a personnel file, anything that predated the YouTube video. Absent that, Randle's conclusion is the only defensible one.

The broader problem won't close with this story. Contact Redacted had posted three other videos in a similar vein at the time Randle wrote his post. The comment sections on those videos, presumably, are filling the same way. The researchers who know the record are aging. The YouTube algorithm is not.

If any of the eleven purported copies of that 1985 report surface in a named researcher's archive — with provenance and chain of custody — the picture changes. Until then, this is a story about how fiction becomes folklore, and how long it takes someone who actually knows the files to catch up.