Huh, that is interesting.
So if I understand from other posts correctly you have a simulator that drops items (I'm guessing for an individual or does it simulate a team?) And you track that to make sure the percentages are where you want them as well as the correct type.
Then you also have live metrics telling you in broad terms how closely for the entire population those numbers are matching up?
So even with those tools the bug, which my friend and I have experienced though my GF has not can only be reproduced at scale?
I actually find that kind of cool.
I know you're busy but, have their been similar bugs you've found and in general terms do you know why tools for finding them didn't work and you needed to rely on reports on a massive scale to track them down?