In response to the questions/comments about Electron Cloud -
Eco, yes, you never know where electrons are at any given point in time because they're constantly moving. This is why I used the example of an electron frozen in time.
ZDS, I'm talking about lynching scummy players that are town that one thinks could be scum. Sorry if that wasn't clear - I was referring to "players you aren't sure are town" as "players you're not certain about". Obviously you lynch scum to win the game.
2011: Best Mafia Performance (Individual) - Best Newcomer
2012: Best (False?) Role Claim - Worst Town Performance (Group) - Best Mafia Performance (Group) - Best SK Performance - Best Overall Player
2013: Best Non-SK Neutral Performance
2014: Best Town Performance (Individual) - Best Town Performance (Group) - Most Interesting Role - Best Game - Best Overall Player
2015: Worst Mafia Performance (Group) - Best Read
2016: Best Town Performance (Group) - Best Town Player - Best Overall Player
In response to the questions/comments about Electron Cloud -
Eco, yes, you never know where electrons are at any given point in time because they're constantly moving. This is why I used the example of an electron frozen in time.
Actually, you never know exactly where an electron is because of Heisenberg's Uncertainty Principle and the fact that electrons are not just particles, but also have wave-like properties. It's literally impossible for an electron to be in one specific point, with the orbitals indicateing the area where set of electrons (most likely) are.
2011: Best Mafia Performance (Individual) - Best Newcomer
2012: Best (False?) Role Claim - Worst Town Performance (Group) - Best Mafia Performance (Group) - Best SK Performance - Best Overall Player
2013: Best Non-SK Neutral Performance
2014: Best Town Performance (Individual) - Best Town Performance (Group) - Most Interesting Role - Best Game - Best Overall Player
2015: Worst Mafia Performance (Group) - Best Read
2016: Best Town Performance (Group) - Best Town Player - Best Overall Player
2011: Best Mafia Performance (Individual) - Best Newcomer
2012: Best (False?) Role Claim - Worst Town Performance (Group) - Best Mafia Performance (Group) - Best SK Performance - Best Overall Player
2013: Best Non-SK Neutral Performance
2014: Best Town Performance (Individual) - Best Town Performance (Group) - Most Interesting Role - Best Game - Best Overall Player
2015: Worst Mafia Performance (Group) - Best Read
2016: Best Town Performance (Group) - Best Town Player - Best Overall Player
Interaction and wagon analysis are cornerstones of my play.
You re-evaluate reads by rereading, and when a flip happens rereading will always net you some new bits of information that you haven't discovered before.
The problem is.. rereading takes a lot of time and work, and it's difficult to determine what you ARE looking for when you reread. It's why I prize analysts who constantly reread the game over those who don't.
Sometimes players will anchor in their heads the idea that someone is scum, and pay no attention to pieces of information not (directly or obviously) related to this player. It's a flaw for sure. That being said, there's a difference between players who "constantly reread the game over" and those who are actually able to reevaluate their reads — and I think I'd rather play with someone who doesn't reread (because they don't see the point because they are tunnelling) than with someone whose rereads only feed the tunnel-vision. Of course that's a choice between bad and worse.
I think people do try to take new information, but realize that sometimes it doesn't mean anything. Not every post a scum makes is going to look scummy, but what players need to get better at is reading perspectives. Why would town make post X versus why would scum make post X? If both can fit, what makes the most sense with what you have seen, and almost as imnportantly, can you follow someone's mindset, even if you don't necessarily agree with it.
Don't know if you guys can read this article, but Ari Lax (armlx) just plugged MTGS/mafiascum mafia in it when he was discussing lessons from other games and applying them to Magic.
KoolKoal: Feel free to take this with a grain of salt since self meta isn't particularly helpful, but I think I get scumread mostly for style over substance, but also for a certain lack of substance over style. It's not so much what I AM posting most of the time (though sometimes that can seem bad) but what I'm NOT posting. I've been told I come to non-obvious conclusions a lot, so when I post, quite a bit of the time there's jumps in logic that people can't follow and they think that's scummy. I get that accusation about a lot of questions I ask specifically. People call them "busy work" when the questions are legit etc.
As far as things to ignore, I can't think of anything. I would suggest you focus less on what I'm doing and more on how I'm doing it. That's probably more likely to be accurate. Like I've just said, what I do tends to come off a little weird, but if you look for how I do it, mindset comes into play and maybe you figure out something useful.
2011: Best Mafia Performance (Individual) - Best Newcomer
2012: Best (False?) Role Claim - Worst Town Performance (Group) - Best Mafia Performance (Group) - Best SK Performance - Best Overall Player
2013: Best Non-SK Neutral Performance
2014: Best Town Performance (Individual) - Best Town Performance (Group) - Most Interesting Role - Best Game - Best Overall Player
2015: Worst Mafia Performance (Group) - Best Read
2016: Best Town Performance (Group) - Best Town Player - Best Overall Player
2011: Best Mafia Performance (Individual) - Best Newcomer
2012: Best (False?) Role Claim - Worst Town Performance (Group) - Best Mafia Performance (Group) - Best SK Performance - Best Overall Player
2013: Best Non-SK Neutral Performance
2014: Best Town Performance (Individual) - Best Town Performance (Group) - Most Interesting Role - Best Game - Best Overall Player
2015: Worst Mafia Performance (Group) - Best Read
2016: Best Town Performance (Group) - Best Town Player - Best Overall Player
Todd Anderson wrote last week about learning from other games and applying to Magic. While he tended to focus on in-game procedures, I found it interesting because a lot of I have found works on the testing and deck selection side lines up with what I learned playing another game: Werewolf. For those unfamiliar with the game, you have a small group of people with the knowledge of who the rest of their group is (wolves), mix them in with a larger group (townsfolk), and slowly have the wolves eliminate other players while the group as a whole democratically tries to determine who the informed minority is and eliminate them.
The first big lesson was that when the numbers are good and you disagree with them, odds are you are just wrong. A big part of Werewolf is that wolves also participate in the process of voting to find wolves, allowing them to skew the process. There is a town role that lets a player eliminate people, similar to how to wolves do, called the vigilante. In practice it was used as a tactical role to manage numbers, but after a while, people crunched the numbers and realized it had the single highest correlation to town wins of any role in the game.
Even beyond stats, the logic is pretty obvious once laid out. Assuming the game is balanced from the start, removing a random player should help both sides equally. Even assuming true randomness, the vigilante isn't going to remove themselves from the game, so their choice is always slightly biased towards removing a wolf. There's also basic math of the game ending after N people are eliminated, and using the vigilante repeatedly limits the number of eliminations the wolves control and adds to the number of town-controlled ones.
That doesn't stop people from getting really mad about you playing the role optimally. The typical complaint is that faster games lead to less accumulated information and worse voting selections or that the random removal of a powerful town role is a big issue.
Sound familiar?
This logic has definitely died down over the years in Magic, but there is still a strong bias towards midrange and control decks by some players. You hear the same arguments: a fear of losing to specific answers, of not having a plan for long games.
A fear of having less control over the game, regardless of whether it means you win more.
Bias definitely has a place in close decisions, but don't let bias let you make the wrong choice in decisions that aren't close.
Hoping that this doesn't get me a slap on the wrist, but what he did indeed say is worthy of some discussion.
So, I just took a glance at my signature, and I'm realizing that I've won quite a few Mafia awards.
I think the two that I still want to win before I quit are "Most Entertaining Player" (which means I of course have to turn the ham up to 11) and "Best Host" (which would require me to be a lot more attentive to games I'm running). I don't suspect I'll either of them this year - so I'll just have to try harder next.
2011: Best Mafia Performance (Individual) - Best Newcomer
2012: Best (False?) Role Claim - Worst Town Performance (Group) - Best Mafia Performance (Group) - Best SK Performance - Best Overall Player
2013: Best Non-SK Neutral Performance
2014: Best Town Performance (Individual) - Best Town Performance (Group) - Most Interesting Role - Best Game - Best Overall Player
2015: Worst Mafia Performance (Group) - Best Read
2016: Best Town Performance (Group) - Best Town Player - Best Overall Player
Any chance that data could be put into a more manageable format? Like a spreadsheet with columns for game size, each town and scum power role and who won? Very nice to see that for Basics (and Newb games as well) the win rate is almost dead on 50/50.
Also I see you got yourself a neat new sig quote. From a high quality game no less!
These are the cores of my scumhunting toolkit, and I strive to hone them whenever possible. I find it useful to approach the game from multiple angles, and combining together can help you form a coherent picture of the game.
For mindset analysis, you have to put yourself in the mindset of the person making that post. When looking at a post, ask yourself, "What does someone have to be thinking in order to make this post?" What are they trying to accomplish with that post? Are they trying to genuinely solve the game, and making a post to do precisely that, or are they faking it, and trying to go through the motions? Is the question that they're asking trying to help them figure out a specific player's alignment, or was the question asked to make the player look like they're scumhunting, and therefore making them look more town?
That's the key with mindset analysis. You have to look at a post that is alignment indicative (many throw-away posts aren't) and ask yourself, "What is this post trying to do?" Is the post trying to actively affect the game? Does that post move the game forwards?
Mindset analysis is really the key to scumhunting as town. If you're good at mindset analysis, then you will become a very feared scumhunter. Mindset is a bit more useful to generate town reads - like, say I'm pushing Iso, and detail lots of posts as to why he is scum and should be lynched. Look at my posts - am I town and genuinely believe that Iso is scum and should be lynched, or am I scum trying to drive a mislynch on Iso and I know that he will flip town?
Conviction is one of the strongest towntells, in my experience - just need to be careful if the player you're townreading due to it is town and has genuinely believed like they caught scum, or if they're scum and can genuinely bus a buddy because they know that they're scum. (I did this in the Invitational Day 2 with my scumbuddy Wheat, as I knew that aggressively pushing him, even going through the motions to defend townies like Seppel because I wanted Wheat dead, would make the game a lot easier).
If you can do mindset analysis properly, it should pretty much be the only tool you need to catch scum, as you can quite easily establish a "block" of townreads that are town and the scum will can never mislynch. Process of elimination takes care of the rest of the scum.
Like, if you were following GoldenEye, the reason why we were able to rout out the scum so quickly was because we had a strong "block" of town - TappingStones, me, Bloscovi, Silvercrys, not_a_gimmick - that the scum could never infiltrate. Mechanics and PoE made it really easy to identify the scum.
I'm not good at mindset analysis, so that's why I rely on the other two tools in my scumhunting "toolkit" - vote analysis, and interaction analysis - to root out scum.
Vote analysis has honestly gotten a little less effective here because it's been popularized so much, but I still think it's a useful tool. The key to vote analysis is to look at the votecounts and try and determine if there are scum on it. The heuristics to keep in mind are that in a five man wagon, there is likely to be at least one scum on it - especially in a small game - and on a mislynch, the chances that there is scum on the wagon rises to approximately 95%. With scum lynches, it's a bit tricker, but you can decently count on one of the scum bussing, and the other scum to be off the wagon.
We really haven't had many games solved by vote analysis for.. some reason, so let me pull up Twinborn Mafia for some example vote counts. Here's the Day 1 lynch of Twinborn:
Votecount is skewed a bit since I had to vote myself to ensure a lynch at deadline, but say that it's Day 3 of this game, and you want to take a look at wagons. Mindreaver, Hunger, tomsloger, and I flipped, so it would look something like this:
That's three names to pull out of a hat. Tordeck, Huntzilla, and killjoy. Due to my wagon being the Day 1 mislynch wagon, you can be 95% sure that there is scum on the wagon. Keep in mind: This is just one part of your scumhunting "toolkit". You can use mindset analysis to fill in the rest! So let's say that you think killjoy is town because his posts have been actively trying to solve the game. You now have Huntzilla or Tordeck to look at for potential scum.
Spoilers: Huntzilla is town, and Tordeck is scum. For some reason the town decided to lynch the triple voter because.. why, but yeah. That's generally how small game vote analysis works - you fill in the alignments of people you /know/ are town, you use other methods to determine how other people are town, and PoE out the scum. Vote analysis generally works better, IMO, if the town is losing, since you have more town flipped. Interaction analysis works better if the town is winning (you have more scum flipped), which I will get to now.
Interaction analysis is basically where you see scum flipped, and you pull up their posts and see how they interacted with other players. This is somewhat complex, and interaction analysis is mostly honed off of pattern recognition - you learn to recognize the common scum posts and see if they clear someone or not. The heuristic here is "If flipped scum doesn't interact with a player, it could be because they don't want to fake it". Generally, the more interactions flipped scum has with a specific player, the more likely that the specific player is town.
An easy way to keep track of interactions is the interaction chart. I do this by inputting a list of all of the players in code tags, and then go down the flipped scum's interactions. If they passively mention somebody, then it's a weak interaction and is a dash (-) in the chart. If they voted somebody or directly talked to them, then it's a major interaction and is a line (|). Like, do you remember Final Fantasy Mafia II? Take a look at this interaction analysis post from Wheat.
The scum in that game were KoolKoal, KamikazeArchon, Pizzaguy, Megiddo, Wheat, and vezok.
If you take a look at all of the major interactions (lots of lines) then you'll see that they were mostly all townies! Chris, DCIII, Iso, RelmArrowny, and Sepiriel were all town and could be cleared. Meanwhile, all the scum had weak interactions with Wheat, especially Pizzaguy. Of course, this doesn't tell the full story - you need to actually /look/ at the interactions in context - but I've seen the chart be right more often than not. If you see, I accompanied the chart with me actually looking at the interactions, and doing that I was able to clear many of the high performing townies and scumread Megiddo for having lots of "softball interactions" with Wheat. (Of course, that game ended up being one of my worst performances ever, but that post was a shining light in a sea of dreck).
I know a game designed by you is guaranteed to be good, since you had a keen eye for balance when you were reviewing Ace Attorney.
It's the announcement sticky at the top of the forum.
EDIT: oh ***** that was my 5,000th post.
Eco, yes, you never know where electrons are at any given point in time because they're constantly moving. This is why I used the example of an electron frozen in time.
ZDS, I'm talking about lynching scummy players that are town that one thinks could be scum. Sorry if that wasn't clear - I was referring to "players you aren't sure are town" as "players you're not certain about". Obviously you lynch scum to win the game.
{мы, тьма}
2012: Best (False?) Role Claim - Worst Town Performance (Group) - Best Mafia Performance (Group) - Best SK Performance - Best Overall Player
2013: Best Non-SK Neutral Performance
2014: Best Town Performance (Individual) - Best Town Performance (Group) - Most Interesting Role - Best Game - Best Overall Player
2015: Worst Mafia Performance (Group) - Best Read
2016: Best Town Performance (Group) - Best Town Player - Best Overall Player
Actually, you never know exactly where an electron is because of Heisenberg's Uncertainty Principle and the fact that electrons are not just particles, but also have wave-like properties. It's literally impossible for an electron to be in one specific point, with the orbitals indicateing the area where set of electrons (most likely) are.
{мы, тьма}
2012: Best (False?) Role Claim - Worst Town Performance (Group) - Best Mafia Performance (Group) - Best SK Performance - Best Overall Player
2013: Best Non-SK Neutral Performance
2014: Best Town Performance (Individual) - Best Town Performance (Group) - Most Interesting Role - Best Game - Best Overall Player
2015: Worst Mafia Performance (Group) - Best Read
2016: Best Town Performance (Group) - Best Town Player - Best Overall Player
And you say that, yet, I rarely see people analyze interactions, wagons, game flow, etc. effectively as new information comes to light.
{мы, тьма}
2012: Best (False?) Role Claim - Worst Town Performance (Group) - Best Mafia Performance (Group) - Best SK Performance - Best Overall Player
2013: Best Non-SK Neutral Performance
2014: Best Town Performance (Individual) - Best Town Performance (Group) - Most Interesting Role - Best Game - Best Overall Player
2015: Worst Mafia Performance (Group) - Best Read
2016: Best Town Performance (Group) - Best Town Player - Best Overall Player
You re-evaluate reads by rereading, and when a flip happens rereading will always net you some new bits of information that you haven't discovered before.
The problem is.. rereading takes a lot of time and work, and it's difficult to determine what you ARE looking for when you reread. It's why I prize analysts who constantly reread the game over those who don't.
I think people do try to take new information, but realize that sometimes it doesn't mean anything. Not every post a scum makes is going to look scummy, but what players need to get better at is reading perspectives. Why would town make post X versus why would scum make post X? If both can fit, what makes the most sense with what you have seen, and almost as imnportantly, can you follow someone's mindset, even if you don't necessarily agree with it.
The GJ way path to no lynching:
Also reminder that we still need one more person for a Diplomacy game if anyone is interested
/in 6/7
Eco
KillJoy
Iso
Proph
GJ
ced
Is there anything I should be doing? Anyone needs adding or removing or something?
I'm home for a few days.
Edit: oh seems I'm too late for that Diplomacy game. Sounds like fun though.
http://www.starcitygames.com/article/33175_New-Spoilers-And-Awkward-Formats.html
{мы, тьма}
2012: Best (False?) Role Claim - Worst Town Performance (Group) - Best Mafia Performance (Group) - Best SK Performance - Best Overall Player
2013: Best Non-SK Neutral Performance
2014: Best Town Performance (Individual) - Best Town Performance (Group) - Most Interesting Role - Best Game - Best Overall Player
2015: Worst Mafia Performance (Group) - Best Read
2016: Best Town Performance (Group) - Best Town Player - Best Overall Player
Or, like, pay for it or whatever
Someone clearly doesn't play Commander.
{мы, тьма}
2012: Best (False?) Role Claim - Worst Town Performance (Group) - Best Mafia Performance (Group) - Best SK Performance - Best Overall Player
2013: Best Non-SK Neutral Performance
2014: Best Town Performance (Individual) - Best Town Performance (Group) - Most Interesting Role - Best Game - Best Overall Player
2015: Worst Mafia Performance (Group) - Best Read
2016: Best Town Performance (Group) - Best Town Player - Best Overall Player
The first big lesson was that when the numbers are good and you disagree with them, odds are you are just wrong. A big part of Werewolf is that wolves also participate in the process of voting to find wolves, allowing them to skew the process. There is a town role that lets a player eliminate people, similar to how to wolves do, called the vigilante. In practice it was used as a tactical role to manage numbers, but after a while, people crunched the numbers and realized it had the single highest correlation to town wins of any role in the game.
Even beyond stats, the logic is pretty obvious once laid out. Assuming the game is balanced from the start, removing a random player should help both sides equally. Even assuming true randomness, the vigilante isn't going to remove themselves from the game, so their choice is always slightly biased towards removing a wolf. There's also basic math of the game ending after N people are eliminated, and using the vigilante repeatedly limits the number of eliminations the wolves control and adds to the number of town-controlled ones.
That doesn't stop people from getting really mad about you playing the role optimally. The typical complaint is that faster games lead to less accumulated information and worse voting selections or that the random removal of a powerful town role is a big issue.
Sound familiar?
This logic has definitely died down over the years in Magic, but there is still a strong bias towards midrange and control decks by some players. You hear the same arguments: a fear of losing to specific answers, of not having a plan for long games.
A fear of having less control over the game, regardless of whether it means you win more.
Bias definitely has a place in close decisions, but don't let bias let you make the wrong choice in decisions that aren't close.
Hoping that this doesn't get me a slap on the wrist, but what he did indeed say is worthy of some discussion.
I think the two that I still want to win before I quit are "Most Entertaining Player" (which means I of course have to turn the ham up to 11) and "Best Host" (which would require me to be a lot more attentive to games I'm running). I don't suspect I'll either of them this year - so I'll just have to try harder next.
{мы, тьма}
2012: Best (False?) Role Claim - Worst Town Performance (Group) - Best Mafia Performance (Group) - Best SK Performance - Best Overall Player
2013: Best Non-SK Neutral Performance
2014: Best Town Performance (Individual) - Best Town Performance (Group) - Most Interesting Role - Best Game - Best Overall Player
2015: Worst Mafia Performance (Group) - Best Read
2016: Best Town Performance (Group) - Best Town Player - Best Overall Player
Also I see you got yourself a neat new sig quote. From a high quality game no less!
1) Behavioral/mindset analysis
2) Vote count analysis
3) Interaction analysis
These are the cores of my scumhunting toolkit, and I strive to hone them whenever possible. I find it useful to approach the game from multiple angles, and combining together can help you form a coherent picture of the game.
For mindset analysis, you have to put yourself in the mindset of the person making that post. When looking at a post, ask yourself, "What does someone have to be thinking in order to make this post?" What are they trying to accomplish with that post? Are they trying to genuinely solve the game, and making a post to do precisely that, or are they faking it, and trying to go through the motions? Is the question that they're asking trying to help them figure out a specific player's alignment, or was the question asked to make the player look like they're scumhunting, and therefore making them look more town?
That's the key with mindset analysis. You have to look at a post that is alignment indicative (many throw-away posts aren't) and ask yourself, "What is this post trying to do?" Is the post trying to actively affect the game? Does that post move the game forwards?
Mindset analysis is really the key to scumhunting as town. If you're good at mindset analysis, then you will become a very feared scumhunter. Mindset is a bit more useful to generate town reads - like, say I'm pushing Iso, and detail lots of posts as to why he is scum and should be lynched. Look at my posts - am I town and genuinely believe that Iso is scum and should be lynched, or am I scum trying to drive a mislynch on Iso and I know that he will flip town?
Conviction is one of the strongest towntells, in my experience - just need to be careful if the player you're townreading due to it is town and has genuinely believed like they caught scum, or if they're scum and can genuinely bus a buddy because they know that they're scum. (I did this in the Invitational Day 2 with my scumbuddy Wheat, as I knew that aggressively pushing him, even going through the motions to defend townies like Seppel because I wanted Wheat dead, would make the game a lot easier).
If you can do mindset analysis properly, it should pretty much be the only tool you need to catch scum, as you can quite easily establish a "block" of townreads that are town and the scum will can never mislynch. Process of elimination takes care of the rest of the scum.
Like, if you were following GoldenEye, the reason why we were able to rout out the scum so quickly was because we had a strong "block" of town - TappingStones, me, Bloscovi, Silvercrys, not_a_gimmick - that the scum could never infiltrate. Mechanics and PoE made it really easy to identify the scum.
I'm not good at mindset analysis, so that's why I rely on the other two tools in my scumhunting "toolkit" - vote analysis, and interaction analysis - to root out scum.
Vote analysis has honestly gotten a little less effective here because it's been popularized so much, but I still think it's a useful tool. The key to vote analysis is to look at the votecounts and try and determine if there are scum on it. The heuristics to keep in mind are that in a five man wagon, there is likely to be at least one scum on it - especially in a small game - and on a mislynch, the chances that there is scum on the wagon rises to approximately 95%. With scum lynches, it's a bit tricker, but you can decently count on one of the scum bussing, and the other scum to be off the wagon.
We really haven't had many games solved by vote analysis for.. some reason, so let me pull up Twinborn Mafia for some example vote counts. Here's the Day 1 lynch of Twinborn:
Prophylaxis (7) - Mindreaver, tomsloger, Hunger, Tordeck, Huntzilla, Prophylaxis, killjoy
Votecount is skewed a bit since I had to vote myself to ensure a lynch at deadline, but say that it's Day 3 of this game, and you want to take a look at wagons. Mindreaver, Hunger, tomsloger, and I flipped, so it would look something like this:
Prophylaxis (7) - Mindreaver, tomsloger, Hunger, Tordeck, Huntzilla, Prophylaxis, killjoy
That's three names to pull out of a hat. Tordeck, Huntzilla, and killjoy. Due to my wagon being the Day 1 mislynch wagon, you can be 95% sure that there is scum on the wagon. Keep in mind: This is just one part of your scumhunting "toolkit". You can use mindset analysis to fill in the rest! So let's say that you think killjoy is town because his posts have been actively trying to solve the game. You now have Huntzilla or Tordeck to look at for potential scum.
Spoilers: Huntzilla is town, and Tordeck is scum. For some reason the town decided to lynch the triple voter because.. why, but yeah. That's generally how small game vote analysis works - you fill in the alignments of people you /know/ are town, you use other methods to determine how other people are town, and PoE out the scum. Vote analysis generally works better, IMO, if the town is losing, since you have more town flipped. Interaction analysis works better if the town is winning (you have more scum flipped), which I will get to now.
Interaction analysis is basically where you see scum flipped, and you pull up their posts and see how they interacted with other players. This is somewhat complex, and interaction analysis is mostly honed off of pattern recognition - you learn to recognize the common scum posts and see if they clear someone or not. The heuristic here is "If flipped scum doesn't interact with a player, it could be because they don't want to fake it". Generally, the more interactions flipped scum has with a specific player, the more likely that the specific player is town.
An easy way to keep track of interactions is the interaction chart. I do this by inputting a list of all of the players in code tags, and then go down the flipped scum's interactions. If they passively mention somebody, then it's a weak interaction and is a dash (-) in the chart. If they voted somebody or directly talked to them, then it's a major interaction and is a line (|). Like, do you remember Final Fantasy Mafia II? Take a look at this interaction analysis post from Wheat.
The scum in that game were KoolKoal, KamikazeArchon, Pizzaguy, Megiddo, Wheat, and vezok.
If you take a look at all of the major interactions (lots of lines) then you'll see that they were mostly all townies! Chris, DCIII, Iso, RelmArrowny, and Sepiriel were all town and could be cleared. Meanwhile, all the scum had weak interactions with Wheat, especially Pizzaguy. Of course, this doesn't tell the full story - you need to actually /look/ at the interactions in context - but I've seen the chart be right more often than not. If you see, I accompanied the chart with me actually looking at the interactions, and doing that I was able to clear many of the high performing townies and scumread Megiddo for having lots of "softball interactions" with Wheat. (Of course, that game ended up being one of my worst performances ever, but that post was a shining light in a sea of dreck).