Osseric, on 19 January 2011 - 11:22 AM, said:
Anyway, not much to go on. This post stood out to me though:
Fener, on 18 January 2011 - 04:53 PM, said:
lolwut? Maths fail. If there are, say, 4 scum, and we lynch an inno today, there is a 4/15 chance that the lynched person is linked to scum. That's only slightly better odds (the trend holds throughout the game), but is offset by the fact that if they were linked to scum, said scum dies a full day-night cycle after, thus getting off a kill, so you're 2 innos down for 1 scum anyway. That's not to say we shouldn't lynch every day - not lynching is only a potentially desirable outcome for town in an endgame situation, but you presented bad logic to justify what should already be our basic strategy anyway, regardless of set-up. Weird.
EDIT - added 'already'
I think you are the math failure here... we get to lynch one guy every day. That guy has a X-in-16 chance of being scum (where X is probably equal to somewhere between 3 and 6, given typical ratios and the mechanics of this game). Every lynch provides what is essentially a second, free lynch. So even if the first guy lynched is an inno, you now have an X-in15 chance of scum being "collared" out of the game tomorrow. That amounts to X/16 + X/15. That is "very close" to doubling your odds of lynching scum with a single lynch (it's actually slightly higher than doubling your odds).
The actual results have nothing to do with the statistical odds of those results, other than showing a likely standard distribution along a parabellum given a large sample size. We only get one sample: this game.

Help




















