Just this August, a professor of sociology resigned from Cambridge University and from his fellowship at Jesus College. The details matter here, because the problem at hand is precision. The publisher Palgrave Macmillan confirmed that a book he had repeatedly listed as forthcoming — in his biography, in conference materials, and even in a published essay — had been abandoned years earlier and never existed in press. Archived fundraising records show that a 2010 campaign raised £6,285, not the £5 million he claimed. A statistical analysis commissioned by The Telegraph identified 188 sentences in his doctoral thesis that were identical or near‑identical to a 2009 thesis by another scholar, with the odds against chance overlap estimated at one in 100 billion.

What should give us pause is not the misconduct itself; it’s that four separate systems built to detect exactly this kind of failure did not. The PhD‑granting institution convened a panel that chalked the citation issues up to honest error and never addressed the 188‑sentence overlap. The hiring institution declined to investigate because the work predated his appointment. A major higher‑education outlet shelved a completed investigation for ten months after receiving a letter from a defamation firm. And a police force spent four months investigating the journalist who asked the questions.

What finally worked were the tools anyone can use, such as plagiarism‑detection software, a publisher’s own records, an archived donation page, and a statistical comparison that any competent reader could reproduce. Verification succeeded only where the method was transparent, and the evidence could be checked by someone with no stake in the outcome. That is not a story about elite institutions. It’s a story about what happens anywhere verification is nobody’s job in particular. Yet we, of all fields, should resist any temptation to feel superior. Anomalistics has its own history of unexamined narratives, venues with refereeing thin enough to see through, and foundational cases whose evidentiary basis has never been reconstructed. If verification vacuums are the problem, we have lived with them for a long time.

What’s worth reporting to SSE’s membership is that this has begun to change — and not because of sentiment. For instance, the Koestler registry brought preregistration into experimental parapsychology, the “Transparent Psi Project” built adversarial safeguards into its design from the start, and this year the Journal of Scientific Exploration (JSE) adopted TRACE, a disclosure standard for spontaneous case reports modeled on STROBE, PRISMA, and CARE — a decision made by members of its editorial team, myself included along with incoming Editor-in-Chief Everton Maraldi, and one readers are entitled to weigh accordingly. You can read more about this below. Still, none of these developments prove anything about anomalous phenomena, and that’s precisely their value. Each one makes claims checkable rather than merely persuasive.

Two areas will test this shift most sharply. The first is artificial intelligence (or large language models: LLMs). Frontier scientists are already using LLMs for literature synthesis, case coding, and drafting. The instinct to prohibit such activity is misplaced — these tools perform important tasks. But fluent output isn’t verified output, and our data are often small‑N, ambiguous, and contested in provenance, which is exactly where synthetic plausibility is most dangerous. The rule is the same one we apply to a second rater, i.e., disclose the assistance, and never let an AI model be the only entity that reads a source or output.

The second is replication. Researchers frequently work near the “detection floor,” where a single failed study settles nothing — a reality as true in cardiology and social psychology as it is in anomalistics. However, that argument is meaningless unless it runs both ways. For example, if a handful of “nulls” in empirical psi research can’t bury a small effect, a handful of “hits” can’t establish one. Anyone who invokes statistical power in only one direction isn’t defending a finding but rather defending an ideological preference.

So: preregister. State your analytic plan before you see the data. Make records available where consent permits and say plainly when it doesn’t. Not because critics demand it. Because the alternative is having your record audited by whoever has the strongest motive to embarrass you, on their timeline, in their framing. Oh, and while you’re at it, please re-read William Beaty’s 2024 editorial in JSE: “Journey to Credibility: 13 Rules for Maverick Scientists” https://doi.org/10.31275/20243353). We can’t guarantee what the study of various anomalous phenomena will reveal. However, we can guarantee the integrity of the path we take to find out. That’s how a field earns the right to be heard.