EV calculation: The best breaking strategy is to assume that the town is directing guesses (and assigning each guess to a player who isn't included in the scumteam they're guessing at); this effectively means that the scum's guess opportunities can help out the town (because two guesses that would turn out to be wrong will be "assigned" to scum, thus meaning that a townie doesn't have to waste their guess on it). If all scumhunting is done at random, the chance the game result is a town win on day 1 is 2/9 = 6/27; and that it's a town win on day 2 is 7/9 × 7/21 = 7/27.
That gives a 13/27 chance that town will win by one or the other method (and a 14/27 chance of a scum win). Even if town doesn't try to direct guesses, something that could be quite hard in practice, the EV isn't that much lower (it's now 2/9 + (7/9 × 5/19) = (38 + 35)/171 = 73/171). One thing I'm a little worried about is that these EVs seem a bit high given that scum only get one kill, but the setup hopefully isn't horribly townsided anyway. Another potential problem is the lack of information for town after D1. Still, I thought I'd post this anyway in case anyone's really interested in it (if there is a lot of interest, I can run it) or in case it gives someone ideas.
I think this should be alright without associatives? it isnt rly efficient for town to entirely control the choices bc if scum gets into the towbloc they are completely fucked.