I'm guessing that this has to do with an ongoing game.
(11:04:10 PM) senspizzaline: That's actually my bold prediction for the year
(11:04:19 PM) senspizzaline: Miami finishes 2nd in the AFCE.
(11:05:35 PM) jhawk01b: my bold prediction for the year is that whoever wins the NFC West will have a winning record
It is rubbish. Some people just use this instead of a random vote, and at it's least no worse then random. I doubt it's any better then random either, though.
I want us to win just for Yos' inevitable rant alone. -CrashTextDummie
I just looked through Little Italy, and using your parameters, this is my results
Mini 624 -town
Mini 662 -town
Mini 675 -town
Mini 684 -scum Netlava
Mini 698 -town
Open 95 -town
Mini 712 -town
Open 106 -town
Mini 722 -town
Mini 735 -scum houseofcards
Mini 738 -town
Mini 745 -sk
Mini 747 -town, scum geraintm (didn't pick up role pm)
Mini 749 -town
Mini 756 -town, scum alexhans (didn't pick up role pm)
Mini 765 -town
Mini 777- town, scum drakethefake(didn't pick up role pm, replacement confirmed last)
Mini 767 -scum Dourgrim
I wouldn't say that it is in any way an acceptable thing to use in any case at any point in the game. As Battousai nicely listed for us, more often than not, it's a townie that just wasn't paying attention or something. So while, yes, sometimes it can theoretically be smart for mafia to wait to confirm so they can have more time to talk pre-game, it shouldn't be used in a case against someone.
Naughty little fly, why does it cry? Caught in a web, Soon you'll be...
It is quite easy to circumvent the whole 'tell'. I can't see a reason why a mod should ever list, even approximated, the order of the confirming players.
If a mod is allowing pre-game discussion, he should give scum the opportunity to actually have some, without making them suspicious.
In-thread confirmation obviously makes this impossible, but I never saw a good reason for that either.
P(Town last to confirm) = 12.5/18 = 69%
P(mafia last to confirm) = 4.5/18 = 25%
P(SK last to confirm) = 1/18 = 6%
Assuming a 8:3:1 setup
P(random player is town) = 67%
P(random player is mafia) = 25%
P(random player is SK) = 8%
The only thing that can be concluded from this data is that there is a very high probability that the last player who confirms has a role in the game.
Methodology revisions:
1) Don't accidentally skip a game
2) For the games with simultaneous last-to-confirms, count each as .5, although it should be noted that the results are still insignificant at 1
Re Adel's point: "not picking up a role PM" is not a separate behavior from "last to confirm". It's just the mod imposing a [here] undefined time limit for the behavior.
Not picking up a role PM DOES need to be treated differently, Emp. In situations where the PM is not opened, obviously the behaviour cannot by definition have anything to do with alignment. It doesn't really matter in the long run if you include them or not, since they should break entirely randomly, but in the short run they might distort a sample simply due to luck.
Mrrrr? I don't see anywhere in your methodology where you're actually measuring the thing the tell is supposed to catch - did the mod allow scum to talk during confirmations?!?
Mr. Flay wrote:Mrrrr? I don't see anywhere in your methodology where you're actually measuring the thing the tell is supposed to catch - did the mod allow scum to talk during confirmations?!?
That would be step two. Just got to go to all the games in which scum confirmed last and see their scum role/read rules.
Flay: If that mrrrr is directed at me, note what my conclusion actually stated. My [perhaps too facetiously made] point is that this data is inconclusive.
Fonz: Collected data can’t be excluded just because it happens to not fit one hypothesis. For all we know, every player in all 18 games here confirmed 20 seconds after picking up their role, but because of the variation in players checking their PMs and mod’s setting deadlines, this happened too late in 6 instances. You can’t throw out those cases because of factor X, when you have no idea whether factor X was present in the other 15 cases.
Herodotus: If we're ignoring the small sample size in the name of facetiousness, then the only other thing that can be concluded from this data is that there is a very high probability that a player who doesn’t read their role PM has a role in the game.