sure
As for what I was saying, let me use an example with 4 players - A, B, C, D.
A is Mafia. Sane Cop.
B is Town. Insane Cop.
C is Town. Naive Cop.
D is Town. Paranoid Cop.
Then we have the players E, F, G and H.
E is Mafia. Naive Cop.
F is Town. Paranoid Cop.
G is Town. Insane Cop.
H is Town. Sane Cop.
A targets F = Innocent
B targets A = Innocent
C targets H = Innocent
D targets F = Guilty
E targets G = Innocent
F targets G = Guilty
G targets C = Guilty
H targets A = Guilty
-B protects A
-D kills F
-H kills A
F dies. Now here, what we do is assume two states - one where the killed player is Innocent (α) and one where the killed player was Guilty (β).
A and D targeted F, and can report their results.
(F) States -
α - A is Sane/Naive. D is Insane/Paranoid.
Because F died, we know that if A is Sane, D must be Paranoid. If A is Naive, D must be Insane.
β - A is Insane/Naive. D is Sane/Paranoid.
Because F died, we know that is A is Insane, D must be Paranoid. If A is Naive, D must be Sane.
We can't go any further with just that information. But in terms of classification, we're left with:
A must be Sane/Insane/Naive
D must be Sane/Insane/Paranoid
And then we go through the rest of the results. We need to keep the α and β scenarios seperate so that we can then begin to eliminate possibilities as to what cop they are. Once we're down to 1, we'll be able to say whether F was Town or Mafia.
We log all the results like that, lynch informative ones or discrepancies, and move to the next one.