An axiomatic approach to scum hunting

[Flubbernugget]

There's too many scumtells on this site and I can never remember the reasoning behind their validity. I would like to discuss a set of propositions from which all legitimate scum tells should be able to be derived from. For simplicity we will assume some sort of mountainous setup (which is a good enough model for D1, where tells are the most questionable).

1) The respective wincons of the two factions.

2) The town always has a larger population than the mafia.

3) The mafia are informed of their team members. The town are not.

4) Mafia will be lynched if their alignment is proven.

[Flubbernugget]

Tells derived from the above propositions:

1)

At least one mafia must lie to win.

Proof: If the mafia all admit they are scum, they will all be lynched by proposition 4. As such, if the mafia do not lie, they will lose by proposition 1(wincons).

[Flubbernugget]

Expect propositions and proofs to be modified in the OP

[Mulch]

This is bad for towns, it's gonna make us like robots and mechanically when scumhunting is a fluid process.

Not the worst idea, but I think this is an overall negative thing for town play

[Flubbernugget]

A thread like this should never come to the forefront of someone's posting in an actual game but that doesn't mean it's not useful

[Creature]

Town should just note that scum will aim to do whatever town does nowadays.

[Alchemist21]

So what? We're supposed to derive tells from propositions that are either basic facts of mafia games or that are just common sense?

I agree with mulch that scumhunting is too fluid for something like this. Even if a tell is valid at first, scum will adapt and it breaks the tell.

[Lycanfire]

Town should just note that scum will aim to do whatever town does nowadays.

So what? We're supposed to derive tells from propositions that are either basic facts of mafia games or that are just common sense?

I agree with mulch that scumhunting is too fluid for something like this. Even if a tell is valid at first, scum will adapt and it breaks the tell.

forcing scum to play suboptimally makes towns relatively stronger

i consider my scumplay to be too bad to comment much from the scumside. what little i would have to say on the subject is basically the extent of my repertoire.

[Junpei]

Your axiomatic system isn't very formal.

I don't think you can get much traction with just these axioms, I think you need to make more ambitious assumptions to get anywhere (as well as adding a wide range of less ambitious ones).

For instance, the proposition "mafia will respond to a mafia partner's post differently than if that post had come from a townsperson with the same posting history" obviously needs to be admitted into the system. But it requires the lemma that individuals choose actions based on all available knowledge, otherwise mafia could simply ignore the fact that someone is their mafia partner (in principle). So now we're in a framework of bounded rationality, which is immediately problematic because people post from their beds, from the bus, etc, and many posters on this site insight on throwing away logic and using their "gut".

It still may be useful to know what people would do in this framework and how we should react to peoples' behavior if they were rational, but as is obvious now, this system of yours is going to have to assume a lot to get anywhere. You're also going to have to construct abstract objects to represent types of signals, and describe the relationships between these signals. It sounds interesting, I must admit, but I don't have time to mess around with this.