Maybe you smarter people can help a dullard out here. All of these papers floating around seem to use 2 related sets of "transients", either 5,399 or 107,000. Am I correct that the 5,399 number is a sub-set of the 107,000 which is a sub-set of 268,165 original candidates as described by Villarroel:
External Quote:
The 5,399 catalogue was therefore not created for completeness but for the search for a candidate vanishing star.
Despite multiple threads, multiple attempts and a whole lot of coding by what seem like smart people, no one has yet been able to replicate the exact filtering process, right? The authors have not shared the coding pipeline they used to arrive at these numbers, but have apparently shared some of the resulting data with other "independent researchers".
The above claim is that the 5,399 number was not about "...completeness but the search for
a candidate vanishing star". At face value, I would take that to mean one or maybe a few of the 5,399 transients left after all the filtering and cross matching took place, might be a star that was successfully photographed in the '50s and was/is no longer visible in the sky when later surveys were conducted. Not an unreasonable idea I guess.
But, any quick read of the VASCO website makes it clear that's not what it means. "Vanishing stars" is a euphemism for glinting or flashing objects in Earth orbit during the '50s, but not now. Or as they call them, transients. Further reading shows that the VASCO project strongly suggests that these transients are in fact techno-signatures in Earth orbit. A dog-whistle term for alien UFOs. Something Villarroel has confirmed in interviews.
So, 5399 transients that don't correspond to any known celestial body, are not defects of any kind and might be aliens craft/probes orbiting Earth in the early to mid 20th century. Having established this in the original Solon paper, Villarroel and others then looked for correlations between these transient/techno-signatures and UFO related sightings, stories and dubious UFO nuclear testing lore.
But wait, when they went looking for correlations, Villarroel recommends further filtering of the 5,399 transients:
External Quote:
Anyone using the 5,399 transients is also recommended to add similar additional filtering steps, corresponding to what was done for the 298,165 to ∼ 107,000 objects, to remove potential duplicates and sources that escaped theGaia and PanSTARRS filtering.
Just me, but I would think adding "...additional filtering steps" would further dilute the pool of possible transients to a number less than 5,399. However, it seems the opposite is the case. When looking for correlations, one is to use a much larger 107,000 sample size (bold by me):
External Quote:
Thus, the 107,000-object sample is far more appropriate for statistical inference than the 5,399-object subset used by Watters et al., both because of its substantially larger size, removal of certain false positives and because it avoids cross-matching against a zoo of catalogues across the electromagnetic spectrum with differing and uneven sky coverages.
https://arxiv.org/pdf/2602.15171, Villarroel et al - A Response to Watters et al (2026).
So, if one were to search for correlations with the 5,399 sample, would none be found? It seems that's what happened when Villarroel's former co-author, W. Watters, tried to search for a transient-nuclear test relationship using the smaller number he helped arrive at. Thus, Villarroel let's Watters and others know, that they should use a sample set nearly 20 times larger than the original one.
If the point is to filter out as many known celestial bodies and as much noise as possible from the original POSS 1 plates to find the true transients, than isn't the smaller number more likely to be more accurate? If so, why fool around with the much larger one? IIRC, somewhere back in this thread, the suggestion was made that a larger number was needed to be more statistically accurate, or meaningful, which is what Villarroel suggests in the above quote.
But if there is a finite number of nuclear tests in a giving time period and theoretically a finite number of UFO reports in the same time frame, why can the number of transients being correlated with these finite number be ~20 times larger than the original and presumably more filtered sample? If the 107,000 sample size had failed to arrive at a meaningful correlation, would the suggestion then be to use the larger 298,165 sample that they started with? Are they just using a larger sample to get the results they want and then sending the larger sample to other researchers so they arrive at the same conclusions? The above quotes from Villarroel suggests that further sampling and filtering will change the 5,399 transients into the more useful 107,000 transients.
I guess it may be that after finding the 5,399 transients, they went back and did it again on a bigger sample set and found a lot more transients using the same technique. Maybe?
I'm not getting something.
