ZUG!!! RUN!!! You can steal my identity and I’ll go hide on my 5000 alts
This is acknowledged and is not in fact a design flaw, but a design feature
The entire point of the setup is to heavily encourage people to mechanically solve and then cry when it gets them absolutely nowhere
Don’t Ask Don’t Tell (About the Wolf Detector 6000)
notblackorwhite criminal record:
- Attempted murder
- Homophobia
Might be conspiracy to commit murder or something I don’t know. I bet I could pass the bar exam if I tried
I feel like the bar exam can’t be that hard there are some dumb lawyers out there
My worst trait is that I unironically believe I could bullshit the bar exam with a bit of prep despite knowing in my head it’s probably not true
REAL
uh, no, I’d sooner ban chromium lol
I did neither of those things. You can’t prove it. In fact, my sentiment analysis bot has detected you’re both a liar and MEAN with a p-value of 0.001.
I feel honored to be mentioned here
hi pigeon
highlighting tools?
I’m talking about american law tho we got idiots over here test has gotta be ez
that being said I feel extremely bad for that person in the first post holy shit
What I’m claiming is that, such a tell, one that is able to be found through statistical analysis but is not easily visibly apparent by just reading their posts, very, very likely does not exist
That word choice tell you were talking about having -
Well, I looked into that MU tab a bit more. And that data doesn’t look to be normalized at all.
My word variance as Mafia is “2526”
My word variance as Town is “3953”
A 56% difference!
But then you look at the number of posts I’ve made as town and mafia, 2080 as town and 1176 as mafia, and realize that those variance numbers are on the same order of magnitude as the post count numbers.
Direct comparison of unnormalized data alert!
Direct comparison of unnormalized data alert!
Direct comparison of unnormalized data alert!
Normalizing those variances by the number of posts and you get a variance of
2.15 as mafia
1.90 as mafia
Okay, that 56% difference is now down to 13% - the large majority of the discrepancy has already been normalized away.
And then you realize that number of words made per post as mafia and town isn’t accounted for - so if I instead normalize it by number of words posted by each alignment: 64238 as mafia, and 99176 as town
My word variance goes to:
0.0393 as mafia
0.0398 as town
That’s now a 1.36% difference
And is not at all statistically significant.
And that’s not even accounting for other confounding variables like how long you live in a game, the average volume of the game as a whole for your town games vs mafia games, etc.
But just normalizing the data, not even worrying about confounding variables, and the “tell” is gone.
It’s all improper normalization, over-parameterization, overfitting, p-hacking, data dredging, improperly conducted statistics.
Your word variance tell does not exist.
It’s just tea leaves and boogeymen.
society
I wonder what Hydreigon’s word variance is like
wait I just realized that Laz939 is the Hydreigon of Omega Strikers
what