a modbot would go so hard
2 Likes
i’d drop like a 100 bucks tip to whoever wants to dev a modbot with automatic turbo capabilities, is roughly the market price of how much i’d pay to have auto turbos on fol
What is auto turbo?
Discourse has webhooks so honestly it might not be that hard to write a simple one that counts votes. The difficult part would probably be creating a plugin to integrate an interface for it directly into Discourse (or just creating an external interface otherwise).
turbo that is run by a modbot vs. a host. like MU modbot
Wtf is turbo
basically just game with super short day/night phases, think like 10 mins / 5 mins phases or somethin similarly shorrt
1 Like
tools R bad since not everyone can use them and youre putting yourself at a disadvantage if you arent or something
(too lazy to read rn but this seems like an interesting topic- might actually post a wallpost)
ngl a modbot type thing would be pretty fun to write
1 Like
i forgor how this works properly but cant you do like word2vec or something to analyzle all the words a person used when town vs when mafia + like weigh it with appearance of words and when and a bunch of other statistical tomfoolery
though yeah mafia is a permanent conversation where player meta is very important as non meta stuff
i know in one of the games i hosted wolfteams could vote on behalf of their partner
(votes were done in dms and then publicly stated and you needed permission)
2 Likes
Earlier in this thread there was debate on whether it’s possible to use big data like that to accurately predict alignments
It’s not completely known whether it’s possible, but it hasn’t been done before (and some people like Lumi highly doubt it could work)
Oh its been done.
A logistic regression classifier fit on a combination of these feature sets achieved an average precision of 0.39 (chance = 0.26) and an AUROC of 0.68 on 5000+ word documents. On 50+ word documents, an average precision of 0.29 (chance = 0.23) and an AUROC of 0.59 was achieved
2 Likes
Might be overfitted; but a 25% chance of being wolf is strong (i still havent read the paper)
1 Like
is that not just, like… rand?
2 Likes
wait yeah 
3 wolves /12 all players (i.e 9v3) =25%
I think the paper explained more but i still havent read it
1 Like
stupid fucking programmers making robots to analyze text just to say any given player is mafia 25% of the time
5 Likes
I too can say a player is mafia with roughly 25% accuracy
2 Likes
ok actually nevermind maybe I cannot do that. that may be overstating my capabilities
3 Likes