The actual preprint article is [1]. It's quite a fun read.
I personally find the argument quite weak, but just like a Jules Verne novel, where the science foundation is a bit shaky, at least the argument is good food for thought, food that in this case I find to be quite delicious.
First here's a summary of the argument: the Pleiades are a constellation of stars that are part of various myths in various parts of the world. The authors claim that all these myths talk about seven sisters, but since nowadays only 6 stars are visible, and 100k years ago 7 stars were visible, ergo the myths can trace back to a common origin story.
Second: what are the Pleiades? They are a cluster (a group of gravitationally bound stars) of about 800 stars, about 450 light years away from us. The vast majority of these starts can only be seen with a telescope.
Now let's see what parts of the author's argument are shaky.
1. How many stars are actually visible? According to [2] (which is listed in the preprint's bibliography):
"modern observers [..] can typically discern only six stars in the cluster. But that's simply a consequence of a light-filled night sky. With sharp eyes and a clear, dark sky, it's possible to spot up to 12 stars in the Pleiades group."
2. Which 2 of the stars are currently seen as 1? It's Atlas and Pleione. However, in the Greek myth of 7 sisters, these two stars are actually the father and the mother, they are not part of the 7 sisters per se. What the Greeks called the Pleiades actually had 9 stars, the 7 sisters and the parents. Assuming the parents are seen as a single star, a simple way to reconcile the commonality of the 7 number in different myths is that other people saw these 2 stars as 1 too, but simply did not count the faintest of the remaining 7. That faintest of the remaining 7 is a binary star Asterope that is on the boundary of the constellation (and likely to be below the horizon as seen in Australia for most of the year), so its inclusion in the constellation is kind of arbitrary
3. Why are the 2 stars not visible as distinct? The angular separation between them is about 5 minutes of arc. Humans can resolve up to 1 arcminute. Here's a little experiment: "e". Take a look at that e. There are roughly 2 holes in it separated by one horizontal lines. Look at that "e" from a distance, and see how far you can resolve the 2 holes in it. On my computer the horizontal line appears to be about 0.5 mm thick, and I can resolve the letter at about 1 meter distance. At that distance 1 minute of arc subtends about 0.3 mm in length (2pi = 6.28 meters divided by 36060). There you have it, the experiment that shows you can resolve about 1 minute of arc.
But if the arcdistance between Pleione and Atlas is 5 arcminutes, why can't we resolve them? The preprint explains that: the faint star (Pleione) is next to the brighter star (Atlas) and we can't distinguish it because of Atlas's glare. It then goes on to say that in such conditions, the minimum distance for people to be able to see Pleione is 3 or 4 minutes. It somehow concludes that because Pleione is at 5 minutes distance, most people can't distinguish it.
4. How about the past? Pleione is a variable star, its magnitude (brighness) changes. Wikipedia lists a magnitude range of 4.8 to 5.5 (currently at 5.05). Is it possible that it was much brighter during the Ancient Greek period? Well, of course it is possible. The preprint states as such: "we cannot discount the possibility that one of the faint stars was much brighter in the past." And then they proceed to just discount that possibility.
Instead they go for the hypothesis that 100k years ago the arcdistance between Pleione and Atlas was higher by 3.4 arcminutes. Now, a very solid principle of numerical calculation is to always be skeptical of extrapolations. Interpolations are fine, but when you see an extrapolation, pay extra attention. Here, the authors extrapolate from the astronomical observations made during the last 50 years to 100k years in the past. 3.4 minutes in 100k years correspond to 0.01 arcseconds in 50 years. Do we have this type of precision? Wikipedia [3] claims we can reach an angular resolution of 0.001 arcseconds using telescope arrays, and it's possible we can do even better now (remember the "imaging" of the black hole about 1 year ago?). But did we have this capability 50 years ago? I'm not so sure.
[1] https://www.dropbox.com/s/np0n4v72bdl37gr/sevensisters.pdf?d...
[2] https://www.space.com/pleiades.html
[3] https://en.wikipedia.org/wiki/Angular_resolution#Telescope_a...