As interesting these kind of projects are, one should understand that the final, averaged results can, at the very best, only be a broad guideline, not a hard and fast rule. Just because someone can split a certain, uneven double star with a given aperture doesn't mean that everyone else can, too, even if they know it's possible. It depends on a myriad factors, the biggest of which is the observer.
So what the project does is not using observers to rank double stars, it's using double stars to rank observers*, even if the reverse is what was originally intended.
*) And, possibly to a large extent, also the quality of the equipment and the skies of the various observers.
Very nice comments, Thomas. I greatly like the idea of ranking observers.
Whether or not the observers agree......
It could of course, if there are enough observations, give some idea of what a miscellaneous collection of observers might be able to see in the way of unequal pairs, obviously (as always) allowing for the seeing conditions having to co-operate with the tougher ones. And, of course, allowing for false positives with some observations, or even for too much optimism as an on-going feature of the occasional observer. But perhaps it will average out. Who knows? One day the piles of data will have an attempt at analysis imposed on them, and the result will be ???????
I think Wilfried Knapp's work on a rule-of-thumb for such objects was better thought out, and the discussions of it here on CN led I'd say to some good thinking on the matter of likely "fuzzy limits" for unequal pairs. It's worth reading Wilfried's article summarising his investigations that was published online in the JDSO vol 14 no 1 January 2018.
Various folk have worked on this matter. My own preference for a starting point is still Treanor's 1946 article, which begins with the logically coherent notion from diffraction theory that the limit relates to the Rayleigh criterion. Obviously, there will be significant differences according to the varying delta-M values of the doubles observed. Near-equal doubles are better discussed in relation to Couteau and Taylor's ideas. Very unequal, say 2-mags or more, Treanor; and in between Wilfrid has some interesting suggestions. And, complicating it all, the role of CO - central obstruction - in its varying size.
I suspect this topic, resolution limits of unequal doubles, will still be a topic of debate and discussion twenty years from now.