No, but I should be able to look at doing such once I have the first post here updated. Do you mean "number of replacements in a game vs which faction wins", or "whether a slot is replaced vs whether it won"?
This should be required reading for...everyone for anything, really.
Town player replacements are at 37.2%. Town Newbie player replacements at 44.9%. Scum player replacements at 46.5%. Scum Newbie player replacements at 64.6%.
Oh my, that is a big gap. 9.3% for all players, and 19.7% for newbies. But newbtown replace out often enough to make risking a policy lynch on replacing out a bit dangerous.
It seems that people are misinterpreting the fact that newbie scum are very likely to replace out as thinking that if a newbie replaces out, they are likely to be scum.
There have been 668 newbies over 128 games, so an average of 5.21875 newbies/game.
On average (assuming random selection of roles), there are 5.21875*(2/9) = 1.16 scum-newb/game.
On average (assuming random selection of roles), there are 5.21875*(7/9) = 4.06 town-newb/game.
Using the slot as a basis:
P(newb-scum replaces out) * #newb-scum = #newb-scum replacing out per game
0.528 * 1.16 = 1.345
Following the same thing for newb-town, we get
0.368 * 4.06 = 1.49
So if a newb replaces out, we get a ~47.4% chance that they are scum and a ~52.6% chance that they are town.
47.4% is a lot higher than the regular 2/9 chance of hitting scum when lynching randomly. In fact, if you have odds of 47% of hitting scum, I'd say it warrants a PL.
think of it this way,
presuming D1 and 2 vs 9 and the accuracy of your calculations, if in these games studied...
a town lynched randomly, they'd lynch scum correctly maybe a little more than 20% of the time on average
if town instead lynched on the basis of replacements, that rate skyrockets to nearly 50%
that's what makes something a scumtell: they indicate scumhood at a rate better than chance
In post 60, TierShift wrote:47.4% is a lot higher than the regular 2/9 chance of hitting scum when lynching randomly. In fact, if you have odds of 47% of hitting scum, I'd say it warrants a PL.
if it's based on a scumread, it's not a PL
but I agree to a big extent; it seems that if you found someone scummy before and then they replaced out, the strength of your read should in general increase
In my experience as scum on the website I used to play mafia on a while back, there would usually be one person (maybe two if the scum team was big enough) on the scum team who did pretty much all the work for the team while the others just sent in whatever night actions the one person told them to, or in some cases the one person would send in their teammates' night actions for them.
I don't know if such a thing happens or has happened on this site, but if so, it doesn't surprise me that scum replacement rates might be quite high as lesser scum just ride on the success (or lack thereof) of their de facto leader.
In post 51, Toomai wrote:No, but I should be able to look at doing such once I have the first post here updated. Do you mean "number of replacements in a game vs which faction wins", or "whether a slot is replaced vs whether it won"?
The first one. I've seen it bandied about a lot that "replacements help Scum" but I wonder if that's really true.
It might also be interesting to see whether the alignment of the replaced slot had anything to do with it (do Scum replacements weaken that faction? or strengthen it (perhaps because they're usually more experienced)?) but that seems like it might be a little too much work.
I would suspect that contrary to that claim, replacements help whichever faction gets the replace. It would be neat to see the stats.
Usually a replace causes a slot to turn from lurky-unmotivated into active-motivated, and can completely thwart "kill the strongest townies" strategies when the "weak" townies transform into competent scumhunters.
I think there're probably a lot of factors that go into it, like the occasional tendency to lynch replaced newbs as Scum or the improvement of many newb slots by Wisdom.
In post 51, Toomai wrote:No, but I should be able to look at doing such once I have the first post here updated. Do you mean "number of replacements in a game vs which faction wins", or "whether a slot is replaced vs whether it won"?
The first one. I've seen it bandied about a lot that "replacements help Scum" but I wonder if that's really true.
Here you go: (images are same as in first post, so they auto-update every now and then)
In post 67, goodmorning wrote:It might also be interesting to see whether the alignment of the replaced slot had anything to do with it (do Scum replacements weaken that faction? or strengthen it (perhaps because they're usually more experienced)?) but that seems like it might be a little too much work.
Yeah I don't know what I could do to get that.
This should be required reading for...everyone for anything, really.
Hmmm. Not really enough data for most to make a really accurate general assessment but it does sort of look like the initial proposition has some teeth.
I have a new stat coming in a few minutes (I'm just posting so it's not at the bottom of a page) - replacements sorted by role. It's pretty interesting.
This should be required reading for...everyone for anything, really.