Mafia Ratio Optimization

Game Discussion
#1

A buddy in another community shared this undergraduate paper on Mafia optimization and probabilities.

Mafia Modeling

The paper just talks about mountainous games with single-opportunity-at-EoD voting and ERIT (Elim Random In Tie)

The relatively intuitive proof of increase of mafia win probability on odd total player numbers, as well as the final optimal ratio equations are both very interesting IMO. e.g. 11/2 is not what people would expect to be optimal for a 13er.

The fact that they specify the recurrence equation on players makes for a good formula IMO, though. Makes it easier to adapt to games with roles too.

4 Likes
#2

given my experience with mountainous games of different size, yeah it is lol

1 Like
#3

what

#4

Hm I’d seen a general rule of T=3M+1.

1 Like
#5

7 = 3(2)+1
genius

2 Likes
#6

I think that rule exists to account for a few things

  1. Games are rarely mountainous

  2. Social interactions are impossible to accurately model

If the game is modeled mathematically then pretty much every single standard mountainous setup ever has been scumsided

3 Likes
#7

I’ve played 9p mountainous before (once, a long time ago)

town won

1 Like
#8

flashbacks to Random_Strategy

#9

nice

#10

I don’t like 3M+1 because it assumes very particular things about the setup

1 Like
#11

What I’ve learnt here is that you’re all bad at the game of standard mafia yet you think you’re all geniuses who can outsmart the other people at the table. Newsflash: you probably can’t outwit the other people at the table because the other people at the table are also smart.

2 Likes
#12

ratio

9 Likes
#13

+cope+seethe

2 Likes
#14

It’s a decent starting point for most setups, where you can then adjust each side as needed

1 Like
#15

Ultimately what you really want to do is make it so the mafia have to lynch half of the mechanical non-clears without letting either side sweep too quickly

2 Likes
#16

it’s more efficient to just design what you want to do then set parameters because it creates less friction in the process

1 Like
#17

Interestingly, I find that role madness setups tend to need to veer FAR away from it too. I’m partial to the following optimization model, which is not everyone’s favourite:

4/1
5/2
9/3
12/4
15-(1+)-5

1 Like
#18

i’m glad i’m not the only one who saw this thread then immediately thought this

2 Likes
#19

counterpoint
i modeled a 5v4 to be townsided because people could never be wrong on the internet

1 Like
#20

Time to make a 2-7 game townsided.

1 Like