Mafiascum.net

Yeah after this game I have no clue
My "read him using other people" tactic didn't work

In post 2174, ScrewTheTells wrote:Curious, how does one town read not-mafia? Because I think if I were town I'd have shot him too.

my readrate on not mafia this game was 100% accurate

In post 2166, SirCakez wrote:?
In a 13p town can have four miselims: 10v3, 8v3, 6v3, 4v3
Here we had three mis-shots - on Lotus, Wheme and NM

I don't consider XyLO to be mislims that town "gets". If town "gets" a mislim, that implies a recoverable position whereas in XyLO a mislim means game over. That's where I got the 2 vs 3 instead of 3 vs 4.

And as I said, even with the single fewer total mislims from a standard 13p, town still wins more often than not if they shoot correctly about half the time.

Hello, this is the newbie-head here. Just wanted to say that this was my first game, and I really enjoyed playing with all of you guys.

In post 125, Netflix and Chill wrote:Assuming no secret alts?

You Cakez dunn Norfolk. The former three have seen me in pretty dead on games. Norfolk lost to town me (we were competing day one wagons and he died)

I was the one who first brought up making you the gunbearer

Oh, I remembered something, Syryana.

In post 2159, Syryana wrote:Assuming town shoots correctly slightly more than half the time (55% of the time, in my sims), town will win 52% of games.

But isn't shooting correctly 50% of the time already WAY above random chance? Because when the gunbearer shoots, there's always at least as many town as scum. Pre-Elim-Lose, there's a higher chance of hitting town (i.e. shooting wrong).

So if you literally gave town a 55% of shooting scum with each shot then this is simulating really effective scum-hunting. The first shot should have 5/12 chance of hitting scum, or 42%. 55% is huge, that';s like a 30% relative increase in scum hunting effectiveness.

In post 2180, ScrewTheTells wrote:But isn't shooting correctly 50% of the time already WAY above random chance? Because when the gunbearer shoots, there's always at least as many town as scum. Pre-Elim-Lose, there's a higher chance of hitting town (i.e. shooting wrong).

So if you literally gave town a 55% of shooting scum with each shot then this is simulating really effective scum-hunting. The first shot should have 5/12 chance of hitting scum, or 42%. 55% is huge, that';s like a 30% relative increase in scum hunting effectiveness.

You're assuming town's chances are always random. In practice, town's chances are AT WORST random.

Town's chances are always better than random. This is the difficulty in simulating mafia; how does one incorporate tactics like scumhunting and townhunting in scenarios? But in all my years playing mafia I've seen town take good shots and bad shots but I've never seen town pick a name out of a hat and yeet it.

If the chances are purely random, scum win an overwhelming 90+% of games. Historical data proves the lie to this; popcorn in past games has had an overwhelming town winrate. The gunbearers on average shoot correctly more than half the time. Hell, I myself hydra'd with Nacho and had a 75% accuracy rate in Open 501. So I took the outliers, averaged them out, and modified the base accuracy until the winrate matched the historical data. Then, using that as a baseline, I added in the vengekills and tweaked the accuracy to see the effects of good and bad scumhunting. Sometimes town all misses, sometimes town all hits. Over a billion iterations (which is what I ran) those sort of outliers iron themselves out.

It is possible I overestimated town's chances slightly, but this setup's swing is more dependent on good play than almost any other anyways due to its (more or less) nightless state and lack of anything resembling a PR game.

On another note though, I will say it was a pleasure having you in this game, both of you! I enjoyed your contributions in the scum PT and impressed by how you used self meta to throw MUSH and the other town off your scent.