Of all the "events" they find, none of them are independently detected in a second image. Old photographic plate date is filled with spurious artifacts from defects (tares, scanning defects, holes in the emulsion). And if you look at the events this team are pointing to, they don't have the same profiles as normal stars. The team has no substantial evidence that they are real. The mundane explanation that they are defects has not been eliminated. It is really shitty "science".
https://arxiv.org/abs/2402.00497
The correlation with nuclear testing could be something as simple as the weather being good in California and Nevada at the same time. They never checked whether the whole dataset of plates is correlated. But given the team's shitty work, I don't really trust their correlation statistics, they haven't provided the data for anyone to check.
If you look at the teams earlier paper you can see they admit these plates are filled with defects. See the giant flying ass shaped thing, which they admit is a defect. How do they know some are real transients and some are just defects? They don't. None of the transients is detected in following exposures. Also these magic events cluster together, which can simply be explained by faulty plates.
