Thirteen Cops

[mith]

Way too much randomness to get anything meaningful, I suspect. There's a very fine between completely solved and too random in all-Cop games.

To justify this type of setup, I think you would need to prove both that there is no breaking strategy which guarantees EV>1/2, and that there is enough information to be meaningful at endgame.

[mith]

I remembered another old game today; wasn't an all-Cop game. For the game in question, every player had 4 attributes (one of which was their role, the others were physical traits; there were 3 or 4 options for each). The Detectives in the game could investigate a body each night (there was a NPC N0 death, scum had to pick a killer), and would get one piece of information about the killer. Scum Detectives (I was the only one in this particular game) could plant evidence on a body instead, since they knew whodunit. It was an interesting way to introduce probabilistic evidence.

In a 3:10 with 4 traits and a N0, you'd be looking at 6 possible results for the real cops (4 real plus 2 planted by the non-killing scum), plus fake results from all scum. So not much to go on day 1. The other advantage to this type of setup as far as controlling information is that scum control who goes out each night, and could either send the same killer repeatedly until lynch (minimizing exposure for the others), or you could mix it up to confuse the results more. Might still be solvable somehow, of course. Probably more interesting in a larger setup with more scum, to mitigate the effect of the day 1 lynch determining the result.