12999th poster gets a cookie (cookie thread (Part 7)) (Part 8)

General
#622

what kind of setup :flushed:

1 Like
#623

(see: town of salem role buckets)

#624

Was attempting to do it for a role madness setup.
Didn’t get far enough to establish more than kills, heals, and RBs.

2 Likes
#625

fixed

3 Likes
#626

Kinda want to revive it now.
Main parts that’re hard to code are answers to mechanical information, stuff like, “If no one died and someone was RBed, town kills the RBed player.”

1 Like
#627

Could just…not code that stuff and come back to it later.

1 Like
#628

oh i did this too and gave up because i am so not a data person
i was trying to see the effects of massclaiming on a stripped down version of tos and tos2 i.e. roles had alignments but no abilities

2 Likes
#629

…if you cant tell, this was in my tg days

1 Like
#630

Ok, pulling this up and finding that my method for deciding who dies is kinda wonky in a neat way.
Basically, everyone starts off with a suspicion value of 0.
Every day, town members have a 50% chance to add 0, a 30% chance to lose 1, and a 20% chance to gain 1. Mafia members have a 40% to add 0, a 25% chance to lose 1, and a 35% chance to add 1.
From there, you randomly choose a player with the highest suspicion value and execute them each day.

#631

maybeeeee
hard to know

2 Likes
#632

yeah this is why complexity fucking sucks you’re basically trying to model human thought process

3 Likes
#633

“Where did you get these numbers from, Ash?”
I know that I wanted to make it so that over time, mafia members would generally be more suspicious than town members, but the numbers themselves are very much shots in the dark.

2 Likes
#634

this doesn’t model quizbowlcord games effectively. town suspicion numbers should be going up by an average 30, and mafia suspicion numbers down by an average of 10

3 Likes
#635

[steps up to mic]
“QBCord bad”
[10 trillion billion gazillion applause]

1 Like
#636

even if you used a Person Machine which chose exactly like a person youd fall victim to Culture and Local Skill as shown in this example

1 Like
#637

There’s got to be enough data to somewhat model this.
Like, you’ll have to make significant assumptions, but it’s not a life-or-death thing; I just want to be able to use it to help make sure my games aren’t terribly swingy.

#638

Broad strokes are fine for small-scale stuff.

#639

like a reasonable solution would not assume the entire scum team would bus one member. This is because fol players are not reasonable

#640

utilitarian line of thinking spotted

#641

human behavior can’t be that hard to mathematically model, right? [clueless]