Student Mafia V - Page 122

On January 28 2015 07:10 geript wrote:

Nah, Shining was WAAAAY easier than DP. Like objectively, DP's probably easier. But Shining's posts this game were so trash compared to his newbie game. Like the thought process in between was different. Like I had 5 good separate reasons for wanting to lynch him.

On January 28 2015 07:11 DarthPunk wrote:


There were also a lot of points of time in which I was trying to lynch him. So i think the distance was enough that we still would have won if I flipped scum.

On January 28 2015 07:14 Blazinghand wrote:


I thought you guys played great! I was just noting that town was totally capable of winning right to the end

On January 28 2015 07:15 Half the Sky wrote:


Generally speaking there is a town bias in most observer QTs I've been through

I wanna talk about another subject of mine near and dear to my heart: vote count analysis! VCA is a strong tool to have in the toolbox of any good mafia player. It gives you the following advantages:

(1) It's just plain good at catching mafia when done right. Votes are among the most concrete data you'll have in a game of mafia, and while the innate subjectivity of interpreting those votes makes it an inexact science, it's still a great starting point for trying to find mafia.
(2) You can do at least a rudimentary vote count analysis even without reading the thread all the way through. This makes it especially handy as a "reset button" of sorts for you; as I've noted already, I advocate giving yourself breaks from the game as town, even at the expense of being fully caught up, so if you had a busy day or two and you don't have time to get up to speed, you can still contribute to the discussion and maybe light a creative spark in someone else.
(3) Very few people seem to bother doing VCA, despite how useful it is. This gives your town game an extra dimension compared to most other players on TL, which means you'll be a more valuable teammate, because you can approach the same situation everyone else is seeing through a different, more refined lens.
(4) On a personal meta point, VCA is something you can still do as mafia. You just have to make deliberate errors in interpreting the actions of the other players. Good town players will see through what you're doing, but if you seem sincere enough and you ask decent-sounding questions (read: you can competently play mafiaside), you can still do VCA and be okay. This hurts your town game a little bit, in that you can't get chump town reads, but it gives you a dimension of apparent productiveness as mafia which can help you succeed there.

So what are we looking for in a vote count analysis? I'm going to link you to one I did in Imperial Mafia, because it's an example that pretty clearly demonstrates how VCA works with minimal side-reading required. (I summed up the relevant filter activity in the VCA post, so you don't even need to read anything in the thread, just the post.)

Click Me

So in it I identified Koshi as town, and geript, Superbia and ritoky as suspects. I was wrong about geript (he was lynched d2 and flipped town), but right about Koshi (shot n1), Superbia (lynched d3) and ritoky (lynched d4). If you read a little further I started to get on the right track about Artanis (survived to endgame as mafia); I'm not gonna say I would have caught him, because he played a hell of a game, but I think I might have correctly identified him as a person of interest if I had been able to do a full VCA instead of the partial one listed there. (Yes, it's partial.)

So what are we looking for in a VCA? I always start with outliers: people whose votes ended up on neither of the main wagons. Outliers are fantastic for VCA, because there's really only two conclusions to be taken from them after correcting for irl activity, and both are alignment-indicative:

(1) This person was on an outlier. This person made a case for said outlier and appeared to push it reasonably consistently, engaged the target to build the case and encouraged others to vote for the target, including engaging prospective voters with questions and arguments for lynching the target. Essentially, this person <i>tried</i> to make their target a main wagon, and simply failed to get traction for whatever reason. This almost always means the person is town. Koshi is a good example of this from the Imperial analysis. If you read his filter, he started to suspect Vivax was mafia, and then he questioned Vivax about a few things. He votes Vivax and then is very vocal about trying to kill Vivax. Vivax almost becomes a main wagon but ends up falling off at the end in favor of geript and, later, batsnacks.

(2) This person was on an outlier. This person may have made a case at some point for their target, but they didn't push the case very hard. They didn't really engage their target much, they didn't engage prospective voters to get them on board, they weren't very loud about getting their target killed. Essentially, this person can't be said to have tried very hard to get their target killed. This makes them statistically significantly more likely to be mafia than random, by a wide margin. I won't say it's a >50% thing without actual data to back it up, but my intuition from my past games says that this is pretty close to the best mafia tell in the game. Palmar phrases it like this - "The guy who's off doing his own thing is almost always mafia" - and I think he's pretty much on-point. This described ritoky and especially Superbia from Imperial; both of them did apparently have irl issues conflating things, but even when they were around during d1, they didn't push their reads much and it was clear they weren't terribly invested in getting their alleged targets killed.

Some donkeys - a subset of which are scum, but surprisingly most of which are town - will try to tell you that this is WIFOM, because if mafia know that being on an outlier is more often than not revealing for them, they'll just not be on an outlier. Resist this. It's not a bad argument, and it's why I would probably err toward town if I'm truly undecided on whether or not someone pushed their target, but it doesn't excuse the players under discussion if they didn't push their target.

The next thing we want to look for is why people jumped on the main wagon. The catch here is that we're not often going to find mafia this way, because mafia and townies do a lot of the same things when they're getting on a main wagon. They both tunnel on the main wagon, they both vote the main wagon and go afk, they both sheep a townread onto the main wagon, and they both lynch townies... a lot. So what we're looking for here is not whether or not someone is mafia but whether or not someone is town. You want to look for the people who were actively trying to figure out the alignment of the person that got lynched, who were pushing to get them lynched, and then read their arguments to see for yourself whether or not they make sense. If you find that a player did the first two things, and that their arguments seem sincere, then you put them in your do-not-lynch pile. If you don't, you put them in your lynch pile until further notice.

Looking at the VCA from Imperial again, we can start with the batsnacks lynch. JAT and I both made cases for batsnacks in the final 1.5 hrs of d1, and we both were actively urging people to vote for batsnacks and asking batsnacks questions about what he was trying to accomplish. We were both town. The other four on the wagon were DrHelvetica (town), sicklucker (town), Artanis[XP] (mafia) and Damdred (mafia). DrH had basically cleared himself d1 because he built a monstrosity of a case on TheChyz - he ended up following onto the batsnacks wagon because his train didn't go anywhere, but had he stayed on Chyz we would have towncleared him by the criteria I listed above, and he only switched because he knew he wasn't getting the lynch he wanted that day. sicklucker was a point of contention for most of the next night; if you read my questioning of him throughout the night you'll see that I was ultimately able to correctly townread him before I died, and if you keep reading, he pulled some silly cop shenanigans d2 that cleared him.

On January 28 2015 07:15 geript wrote:

I'll explain why it's a bad check. Cop play is tough, but generally you want to think of the cop as a vigilante in some regards. Checks toward inactives are huge. Like LM was the best N1 check because there's literally so little to read him off of and he's lynch bait. Breshke wouldn't have been a bad check.

On January 28 2015 07:24 LoneMeow wrote:
On the positive side of things, I now have a very reliable heuristic on finding scum: Anyone who town reads or defends me is very likely scum... :D

On January 28 2015 07:19 LoneMeow wrote:
rsoultin, sorry about the yelling. I was "kind of" frustrated since it seemed like you were not even bothering to read what I was posting and I was almost certain you were actually town and not intentionally pushing mislynch.