Some responses from Multi Czar to my bug report on the matter has provided some interesting consequences. First, that this is not just a whole bunch of whacky luck that's being observed; this is what the code is "supposed" to be doing. And, perhaps much more interesting, that "supposed" is a very key word. He said it wasn't a bug, but the way he phrased it indicated that it was more of a harmless unintentional oddity in the code, rather than an intentional design feature. When asked about that, he had the following to say:
Here's some data I've pulled from the wiki and my own recent multi-uses (incidentally, it's annoying as shit that a copy-paste from a spreadsheet results in such ill-formatting in code tags). And I petition we consider "medior" to be a word. Sounds cool. Otherwise, the trinket breakdown in the spreadsheet is based upon the prevalence of their appearances, with the type of worthless item noted on the side. For some uses the 2 least prevalent ones appeared in equal numbers, and I placed them according to whatever whim struck me at the time.Multi Czar wrote: It's not intentional for the worthless trinkets, but has something to do with the code for item drops in general. Since it doesn't affect play it's not really a bug, just a sort of odd functionality.
EDIT: I'm not entirely sure what this might mean for item drop mechanics, and if what it does mean is actually detectable through anything other than the 31337 scrolls. But if anyone has any theories or data they'd like to add for 31337 scrolls, or item drops in general, and how to test them, that'd be much appreciated. Maybe a small impact on items that can drop multiple times from a monster? Or even on getting multiple (but distinct) drops from a monster in general?