Mini #1647: Eine Kleine Nacht-Mord, Game Over

[TellTaleHeart]

Given that lack of anything else, im going to assume that is what you are thinking until you provide anything that could be remotely interpreted as anything else.

There's a track. Not that you care since you're admittedly making shit up.
Post 676 and Post 677

[TellTaleHeart]

I really don't like the day 1 duel between you and Prawn, but think that it's massively unlikely for you to be both scum.

Right, and where did that disappear into? He absolutely swore up and down prawn was scum and he couldn't die fast enough. Apparently it was claimed by the smokey haze of "wah wah, you were going to vote me regardless,

mislynch

."

[LlamaFluff]

Given that lack of anything else, im going to assume that is what you are thinking until you provide anything that could be remotely interpreted as anything else.

There's a track. Not that you care since you're admittedly making shit up.
Post 676 and Post 677

So because I didn't magically track a certain player I am scum?

Look if you want to try and weakly justify a vote you have to be able to do better than that. What part is wrong? The part where I said he is most likely to be VT? The part where I said getting that result might stop onion attacking him if I get mislynched? The part where I said the tell on him I thought was strong but is the type people have a hard time buying?

You are scum. After the claim you back off my lynch, I explicitly state that I am going to be aiming for a VT player, you say you can see that logic, then you revote me because I did exactly what I was saying to do after you get an easy mislynch on a lurker.

Next time don't get greedy and try to pick up a free mislynch.

[Untrod Tripod]

I am vla so plz b patient for me to comment more fully

[onion]

ok back. trying to stay interested in the game. probably do a vote analyais or something. catching up. brb

[onion]

ok, i was inspired by your impossible to read spreadsheet, and made my own.

Number of times each player placed a 2nd-5th vote on someone. AKA the Bandwagon-o-meter.
Tripod: 6
Derangement: 5
sns: 5
Equinox: 4
onion: 3
TTH: 3
CBD: 2
LLama: 2
Prawn: 2
GIF: 1
NJAC: 0

Number of times each player voted for NJAC or GIF. AKA The Scum-o-meter.
Derangement: 3
TTH: 2
Tripod: 2
sns: 2
Llama: 1
GIF: 1
onion: 1
CBD: 0
Equinox: 0
Prawn: 0
NJAC: 0

Ok, so people who happen to be on the top half of both lists are: Derangement, sns and Tripod. they are mathematically scummy. y u on there Derangement?!

So we got the whole fake-llama-track thing on Tripod. i don't particularly like either of them, but we should totally lynch Llama before we lynch Tripod. Llama and Prawn were having at it like fuck for a while, so they probably aren't both scum either. oh the web we weave.

Derangement is on the list, and i really like the guy, he's fantastically pro-town. If i was scum i would have NK'd him. instead, they NK'd GIF, who was pretty useless. i'm forced to consider that maybe Derangement is scum. it would make me sad though.

So right now, lynching Llama is the way to go. he's still terrible, and now killing him gets us info on BOTH. He also expressed early suspicions on GIF, who is town.

[Untrod Tripod]

or maybe your entire set of assumptions is flawed from the outset

[Equinox]

Derangement is on the list, and i really like the guy, he's fantastically pro-town. If i was scum i would have NK'd him. instead, they NK'd GIF, who was pretty useless. i'm forced to consider that maybe Derangement is scum. it would make me sad though.

I agree with Untrod Tripod that it's more likely that your assumptions were flawed. You're assuming that scum are going to be later on a wagon, and you're also assuming -- with your "scum-o-meter" -- that town are less likely to vote town than scum are. The latter assumption is especially dangerous because, surprise, scummy people get voted. Your conclusion that Derangement might be scum just because Derangement was at or near the top of both lists is also likely to be incorrect, and scum could have just not killed Derangement for other reasons. GuyInFreezer said that he was going to reveal something Today, for instance.

Anyway, I was going to read prawneater (and LlamaFluff and snscompt1) but didn't because I'm a master procrastinator. I'll get to that within the next few days or so, but I'm getting slammed at work for the rest of the week; expect better participation after the weekend!

[onion]

the assumptions are that scum want to vote for townies, which seems like a pretty good assumption. townies sometimes vote for townies also, which creates noise. when scum vote for townies and don't vote for scum, they show up in vote analysis, and we can use it to find them. to prevent this, scum attempt to vote for scum about as often as a townie votes for scum, keeping them obscured in the noise. but scum voting for scum is anti-scum, and thus pro-town, and so it is in our interests to force scum to do this, and we do that by looking at vote analysis and using it as evidence to hunt scum.

i really wasn't expecting Derangement to show up on that list, and i really wasn't expecting him to not get NK'd. In my mind he's by far the most useful, townish, pro-town player. Because he didn't die, he might be scum. but i'm not really going to to pursue that right now. its a back burner sort of thing.

i am very confused over WHY the scum killed GIF. he was useless. he voted for TTH, then for NJAC. he liked Llama and Tripod, and suspected Prawn, NJAC and sns. did he vaguely hint at having a power role? i don't know why they'd NK someone who was lurker slacker and suspicious when they could have NK'd someone useful, pro-town and active.

[onion]

oh yeah and Tripod, please confirm. please tell us, clearly, if you did or did not visit someone last night.

[Equinox]

Frankly, vote analysis isn't necessarily wrong, and I can see your thought process behind what you're doing. I think I'm just in complete disagreement with how you're doing the analysis, but that's neither here nor there.

As to GuyInFreezer, there's this:

That was an unexpected NK. :o
Can't say I was town-reading GIF, even before his V/LA, so I'm thinking scum took him out because of either his stance on Llama probably not being scum, or due to his Post 632.

Also, I wouldn't write GuyInFreezer off as useless. It's more likely that it's Derangement's hypothesis quoted above, but there's something I want to look at from Day 1 aside from prawneater vs. the world after lolwork.

[onion]

...llama makes a good point in 687.

The purpose of Derangement's plan was to maximize information generated by the track-claim.

If Llama was scum, it is unlikely that he'd fake a positive result. Instead he'd pick someone who he'd be pretty sure wasn't going to visit someone, then fake a negative result on them and hope for the best.

If he followed Derangement's plan, Scum!llama could target someone, possibly out their PR, and then just agree with whatever they said. it IS maximizing information, but its bad in this instance.

So Llama did it right just coming out with the full negative result. Had it been positive, we possibly should have went the other way around, and had him tell us who got targeted. That would give power roles the option to out themselves, or not, and the worst case situation is no worse than the other plan, but it has the chance to do better. let's do it THAT way next time.

Llama, sorry for the harsh words in 679. they were uncalled for in this instance. but you are still scum.

I'm willing to bet that sns and Prawn aren't scum either. the web gets webbier. (one way implication arrows go!)

Scum!Prawn -> Town!sns and Town!Llama -> Town!Tripod
Scum!sns -> Town!Prawn
Scum!Llama -> Town!Prawn

I'm not sure that prawn flipping town would tell us much about sns or Llama. we need to talk about that. someone tell me if Town!Prawn -> Scum!sns or Scum!Llama.

@Equinox, i always want to learn to make the math more powerful, and i've got all the data still here ready to be used. Tell me how you want the analysis built and i'll try my hand at ripping information out of the cold hard fingers of the math.

[prawneater]

Also we have no reason to lynch claimed 1S tracker today.
By keeping him alive, if hes town he'll either get nightkilled or scum waste a valuable night action try to frame the one of the weakest investigative role. If hes scum he still wont be able to lie about it (attempting to refute this ill consider as a scumclaim. I have stuff in mind that I dont want to reveal until tomorrow and id rather not be poked and prodded about this until tomorrow.)

GIF probly got killed because this post kinda looks like a PR claim.

[onion]

ok i can see that, maybe. still seems like a poor reason to choose him above others though.

[petroleumjelly]

Day Two, Vote Count #3

LlamaFluff – 2 – TellTaleHeart, onion
snscompt1 – 2 – Derangement, prawneater
prawneater – 1 – snscompt1
TellTaleHeart – 1 – LlamaFluff

With

9

alive it takes

5

to lynch and

3

to lynch at deadline. Deadline is

March 22, 2015, at 11:30 PM PST.

Countdown

: (expired on 2015-03-22 23:30:00)

Not Voting – 3 – ChannelDelibird, Equinox, Untrod Tripod

[Untrod Tripod]

oh yeah and Tripod, please confirm. please tell us, clearly, if you did or did not visit someone last night.

I did not visit anyone

[Derangement]

So, reading through day 1's Llama vs Prawn again

still

hasn't changed my mind on which of the two I think is scum.
Llama has a nasty habit of wording his interpretation of other people's statements in the way that'll make them look the scummiest, and I find myself often disagreeing with the kind of points he's trying to push.

@Llama: have you re-read the thing you wanted to re-read, and if so, would you like to share any insights you may or may not have gotten out of it?

[Derangement]

@Onion: I personally do not put too much stock into abstract vote statistics analysis, and prefer to focus on more concrete things, like who was voting for whom during the most decisive moments, who sheeps other players often,

why

people say they're voting, and if their actions match those reasons.

Nonetheless, if you think you can glean some insight from looking at the votes history, then at the very least I recommend you keep a few things in mind:

• Different players can be more or less active than the rest.
  Someone may have more votes you consider "scummy" or "town-looking" simply because they have more votes overall;
  Lurkers, on the other hand, might appear very null-ish under this kind of analysis, regardless of their alignment.
  
  
  
• How important each vote was, at the time it was made, and how long it stayed there.
  Early RVS votes, for example, are usually inconsequential, unless they stay there long after people started seriously piling serious votes on the same person (or on someone else!).
  Contrast with a vote for a small-ish wagon, which might have been the tipping point that made more people consider flocking to it.
  
  
  
• Most votes for someone are also unvotes away from someone else.
  Did someone flee from a town-lynch once it seemed inevitable?
  Once we catch one of the scum, did anyone's unvote ever prevent them from getting lynched, or dismantle a wagon?

Goes without saying that all of these things are extremely subjective, and should be interpreted on a case-by-case basis whenever possible, but I'm hoping that they might help you out.

[onion]

well sure, there are plenty of things that are scummy that the vote analysis doesn't catch. However, the only thing a scum hunting tool has to do is to catch scum better than lynching at random. If you have a variety of tools, each of which works better than random, and you use them in combination with common sense and good ol evidence gathering, case building and scumhunting, then you'd end up better off than if you didn't have tools. one tool is the vote analysis.

[Derangement]

Carry on, then.

[onion]

i suppose someone who just voted a huge ton would always have more votes on townies than other people. i could do <votes on townies> / <total votes> to find the percentage of votes they placed on townies. But people who vote more aren't actually good unless they talk more too. there should be a <Votes> / <Posts> percentage to see if someone just votes like crazy. People who's votes to content ratio are scummy, as are people who's town votes to total votes ratio is high. that would mean that to avoid showing up, scum would have to post lots of content per vote, which is something we want, and keep down their votes on townies, which is also something we want. those sound like good analysis!

-=Votes on NJAC or GIF / Total Votes=-
50% GIF: 1/2
33% Derangement: 3/9
33% TTH: 2/6
33% Llama: 1/3
25% Tripod: 2/8
25% sns: 2/8
20% onion: 1/5

-=Votes / Posts=-
25.00% sns: 8/(15+17)
14.81% Equinox: 4/27
09.68% CBD: 3/31
08.11% TTH: 6/74
08.04% Derangement: 9/112
07.84% Prawn: 4/51
07.04% onion: 5/71
06.90% GIF: 2/29
06.52% LLama: 3/46
04.68% Tripod: 8/171

I'm not sure the top one should include 1st votes on people, because we want to encourage scum to stick a 1st vote on someone and stay there pushing the case. that provides content and content we can use to scumhunt.

Are these analysis better than the previous? does it provide more useful information?

[Derangement]

Well, I'm not sure what, if anything, we can glean from (NJAC+GIF)/total votes, for the time being.

If you're working under the assumption scum will vote town more often than town will vote town (is this actually true?), then perhaps once we have more town flips, and/or a bigger sample, someone will stand out.
Right now, I wouldn't trust this statistic even if there were an outlying player: they might just have been unlucky to have vote-swapped between

these

two specific town.
Not knowing anyone else's alignment means we're only looking at a very narrow slice of the votes, when counting which of them are on town.

The second statistic

does

have Bubs/SNS standing out from the rest of us.
I can see it being due to their preference for big but infrequent wall-posts, but which doesn't deter them from voting as often as the more active players.
Of these eight votes, two of them drew my attention: Bubs' temporary 24h vote

(#269)

, and SNS' OMGUS

(#480)

.

I'd personally love to hear from

all

the less-active players more often, so here's hoping day 2 isn't as weird as the first one.

[LlamaFluff]

@TTH - I was looking back at your posts, why did you ever support a NJAC wagon? You really just seem to be more upset with Marquis for lurking, but at best you call him (maybe?) scum for when he was in the very early stages of catching up agreed with a consensus read (which is scummy because?). Then PoE... that apparently never changed over about ten pages?

[petroleumjelly]

ChannelDelibird has been prodded.

[ChannelDelibird]

Ballsacks, etc. Cold is gone; post in a bit once I've resolved a modding commitment elsewhere.