Hm. Well, that's pretty gg for us if Lain is adamant it's me.
@Rag:
Previous iterations of JD's Rumbles has had increasing methods by which the antagonist could subvert the linear progression of solving. Your analysis, while helpful yes, has mostly been about a wide-effect ability or singular-effect ability. You even posit to the rest of us about what other methods could exist when, given previous designs, it's hard to anticipate what the mod has designed.
To that end, perhaps Zinger was right about it being you (as opposed to me) and trusting Lain fully. I mean, he is dead after all. There's no telling exactly how results were effected until we get to see the design file.
Vote: Rag
there are lots of easy to imagine mechanics that can't be modelled by either "nightwide sabotage once per game" or "single target sabotage nightly", and even the night-wide sabotage isn't ruled out becuase if it say, randomized all results, then it would still be possible. (although would be ruled out with something like a 75% chance by my results)
if the sabotage occurs when the villain is targeted or when the villain's target is targeted my model fails. if the villain has five shots that they can use freely at any time (5 in one night or 2-2-1 or whatever else) then my model fails. There are any number of things i wouldn't be able to imagine that could cause my models to fail.
Its not foolproof, its just looking at a few possible scenarios (default setup, nightwide sabotage, single-target nightly sabotage) and seeing if its possible for you/lain to be scum under them. Within those possible scenarios, its much more likely that you're scum. We can't conclude what the probability of one of those scenarios matching this game is, but we can safely say that the probability is non-zero, and so the method is useful, to some unknown degree
insofar as we can't predict the setup, there's not really any winning strategy, and so there's not much for us to do other than try to predict the setup anyways.