1/8 Nod (@nizzaneela) has updated his paper demolishing Pekar et al's claim of 'two introductions'.
v3 is more accessible, same brutal conclusion: the supposed support for two spillovers is an artefact of rigged hypothesis testing.
arxiv.org/html/2502.20076v3
2/8 Quick recap: Pekar et al claimed a Bayes Factor of 60 for two introductions.
After three major errors were found (including a bad cut-and-paste!), it dropped to 4, just at the noise level border.
3/8 Nod later showed that the 4 is still wrong.
The one-introduction model was tested against stricter conditions than the two-introduction model, which is basic statistical malpractice.
Under the same conditions, the Bayes factor is <0.25.
4/8 When you do the proper Bayes factor calculation intended by the authors, the model supports 1 introduction!
This is the kind of sloppiness that gets published in Science and then paraded around as settled fact by Hotez, Rasmussen, Daszak, Gronvall and the usual suspects.
5/8 Remember the backstage drama when these errors were first caught?
The correction saga is worth reading:
gillesdemaneuf.medium.com/backstage-story-the-oct-2023-correction-to-pekar-et-al-e167080e957d
6/8 Bad science, terrible peer review, aggressive PR campaign.
The zoonosis narrative gets the red carpet treatment while proper scrutiny gets you labeled a conspiracy theorist.
Standard operating procedure.
@thackerpd
7/8 Michael Weissman @mbw61567742 has also made some important contributions to fixing the Pekar et al paper, by nothing else than using the model the author intended.
See this podcast with James M. Robins, an epidemiologist and biostatistician at Harvard University:
econjwatch.org/podcast/michael-weissman-on-lab-leak-and-science
Gilles Demaneuf
@gdemaneuf
Pointy Head. Opinions, analyses and views expressed are purely mine and should not in any way be characterised as representing any institution or company.