|
On January 25 2015 16:47 geript wrote:Show nested quote +On January 25 2015 16:23 jarjarbinks wrote:
Ya I haven't used it this game. That model was based on EOD votes, dead peoples votes, and who dies at night. I first figured that if I could get enough games into the system, I could figure out if say voting for a town that gets lynched makes you any more likely to be scum or not. I was thinking about doing something like seeing how claiming or including votes that switch later affect a person's alignment as well, but those require more digging into every game and would take more time.
I only had 2 games in the system then and 3 now. I've been kind of lazy, mostly because it seems like all the people that play on here were very against it. Statistics are supposed to be used to help support things people normally see in the games. Since my numbers were going against common perception, I figured it was either:
You guys were wrong
OR
My model sucked some big balls.
I went with my model sucked some big balls.
I think if I had time to put in like 30 games to the model I might have something. But that's going to be a while. My 3 game model has it like this (baseline is 2/7 because theres two scum out of 7 people):
Me: 43% mafia... lol
Breshke: 41% mafia
Rsoultin: 31% mafia
Shining: 30% mafia
Everyone else: 29% mafia
This is without a +/- 3% buffer.
But like I said before, I'm not using this at all and for good reason. If I did, I would have still been genuinely confused because of the very large train Day1 and GB voting for me... lol So how much percentage points do everyone get? I ask because I was interested in it at end of game; I mean it did have mafia as the #2,3 and 4 on the to lynch train.
Ya well its obviously due to change. But I guess you know that. There's two parts to it.
First part
If you vote for the lynched townie, it causes a 2% increase, if the lynched townie voted for you, it gives you a 23% increase. If the guy that got killed at night voted for you, then that's a 16% increase.
Second part:
You have to factor in wagons and strength of the votes. So if you look back to the waffle lynch, there was 8 people on waffle, 2 on GB, and 2 on me. All the people who voted waffle get a 2% increase to start with. But since there was such a big wagon on Waffle, I used a factor of (4/12) and multiplied it by the 2% to get the actual increase. I found 4/12 by subtracting 8/12 from 1. I did this because on wagons mafia has less of an incentive to park their vote on the lynched guy. So for day 1 all the waffle voters only got about .7% increase because of their waffle vote.
The less experienced/good the town team is, the less you can use the model. It also helps to have more players in the game.
|
On January 25 2015 16:44 rsoultin wrote:Show nested quote +On January 25 2015 16:23 jarjarbinks wrote:
Ya I haven't used it this game. That model was based on EOD votes, dead peoples votes, and who dies at night. I first figured that if I could get enough games into the system, I could figure out if say voting for a town that gets lynched makes you any more likely to be scum or not. I was thinking about doing something like seeing how claiming or including votes that switch later affect a person's alignment as well, but those require more digging into every game and would take more time.
I only had 2 games in the system then and 3 now. I've been kind of lazy, mostly because it seems like all the people that play on here were very against it. Statistics are supposed to be used to help support things people normally see in the games. Since my numbers were going against common perception, I figured it was either:
You guys were wrong
OR
My model sucked some big balls.
I went with my model sucked some big balls.
I think if I had time to put in like 30 games to the model I might have something. But that's going to be a while. My 3 game model has it like this (baseline is 2/7 because theres two scum out of 7 people):
Me: 43% mafia... lol
Breshke: 41% mafia
Rsoultin: 31% mafia
Shining: 30% mafia
Everyone else: 29% mafia
This is without a +/- 3% buffer.
But like I said before, I'm not using this at all and for good reason. If I did, I would have still been genuinely confused because of the very large train Day1 and GB voting for me... lol Don't know how right your model is there, bro, but if it's dependent on other people's reads, we're doing so well this game it might as well be trash -_- Sorry. Just irritated with myself. Anyway, talk to me about what you think without the stats?
I do feel like most people's reads haven't been the best in this one. Very confused. I didn't get much out of LM's filter either.
|
On January 26 2015 02:12 rsoultin wrote:Show nested quote +On January 26 2015 02:08 jarjarbinks wrote:On January 25 2015 16:47 geript wrote:On January 25 2015 16:23 jarjarbinks wrote:
Ya I haven't used it this game. That model was based on EOD votes, dead peoples votes, and who dies at night. I first figured that if I could get enough games into the system, I could figure out if say voting for a town that gets lynched makes you any more likely to be scum or not. I was thinking about doing something like seeing how claiming or including votes that switch later affect a person's alignment as well, but those require more digging into every game and would take more time.
I only had 2 games in the system then and 3 now. I've been kind of lazy, mostly because it seems like all the people that play on here were very against it. Statistics are supposed to be used to help support things people normally see in the games. Since my numbers were going against common perception, I figured it was either:
You guys were wrong
OR
My model sucked some big balls.
I went with my model sucked some big balls.
I think if I had time to put in like 30 games to the model I might have something. But that's going to be a while. My 3 game model has it like this (baseline is 2/7 because theres two scum out of 7 people):
Me: 43% mafia... lol
Breshke: 41% mafia
Rsoultin: 31% mafia
Shining: 30% mafia
Everyone else: 29% mafia
This is without a +/- 3% buffer.
But like I said before, I'm not using this at all and for good reason. If I did, I would have still been genuinely confused because of the very large train Day1 and GB voting for me... lol So how much percentage points do everyone get? I ask because I was interested in it at end of game; I mean it did have mafia as the #2,3 and 4 on the to lynch train. Ya well its obviously due to change. But I guess you know that. There's two parts to it. First part If you vote for the lynched townie, it causes a 2% increase, if the lynched townie voted for you, it gives you a 23% increase. If the guy that got killed at night voted for you, then that's a 16% increase. Second part: You have to factor in wagons and strength of the votes. So if you look back to the waffle lynch, there was 8 people on waffle, 2 on GB, and 2 on me. All the people who voted waffle get a 2% increase to start with. But since there was such a big wagon on Waffle, I used a factor of (4/12) and multiplied it by the 2% to get the actual increase. I found 4/12 by subtracting 8/12 from 1. I did this because on wagons mafia has less of an incentive to park their vote on the lynched guy. So for day 1 all the waffle voters only got about .7% increase because of their waffle vote. The less experienced/good the town team is, the less you can use the model. It also helps to have more players in the game. JJB, help me out here. I'm OMGUSing Damdred. Are my points valid or am I being a stubborn ass again? >< I know you're town, and I respect your intelligence, so can you double-check me real quick?
Hold on I'm catching up. I'm assuming you have an argument on him in this thread somewhere?
|
On January 25 2015 16:49 geript wrote:
To be clear, the reason why it was bad was that you never explained how people got to X%. You know. Like:
Voting for a dead towny adds X percentage
Being voted for by a dead towny adds Y percentage
etc.
Ya I could see that. I think it was because I knew I needed more datapoints and didn't want people to be hung up on certain %'s that change when I get more data. I could see how people would have taken it for a "he's just making things up" kind of deal.
|
On January 26 2015 02:15 rsoultin wrote:
EBWOP:
o.0 can you tell me where you got your percentage increases from? the 23% and 16%? just as a point of curiosity
Regressions. Took old games and found the change in %'s from what the %'s should have been if the person voting for town was killed etc. And that 23% is beyond what the maximum % increase would be. The highest it would be in this game is 23% times 12/13 which is 21.2%. That would be if the lynched guy parks a solo vote on someone he thinks is mafia while no one else believes him. Probably will never happen.
Sorry if its confusing. It would be easier to explain it if you saw the model.
Honestly I feel like my model is useless in this game though. It's not going to help us find scum.
|
On January 25 2015 06:43 Damdred wrote:
I'm wrong more often than not, LM is a lower activity player. And I could be tunneled, and i'm probably wrong about Geript.
I think this quote shows when/why he began switching.
|
Idk rsoultin. Damdred was the first Breshke vote. He didn't need to do that. I could see him seeing that DP had some good points and that LM wasn't mafia and decide to switch to the other guy people were suspicious of. Half survival half change in point of view. Town wants to survive too, not just mafia.
I guess if you think that breshke is town, then it would make more sense if Damdred was mafia to switch to him. There was the beginnings of a train on Damdred. The thing I worry about most with Damdred is the GB kill day 1. I could see a team with damdred/trf on it wanting GB dead. But I could see a team of Trf + two people who haven't played with geript and DP much wanting GB dead as well...
The more posts I get from Dp and Geript, the more I think they are town. Unless they are both mafia, they are 100% sure each other is town too. I might park my ideas that they are town for a while, unless something crazy happens.
I think the NK will help us get a better idea of what is going on.
Honestly I've gotten more suspicious of you (Rsoultin) as the game has gone on. The problem is I can see the way you are acting being due to u as a person, which makes me hesitate on it. I could see you having this strong/long a filter with you as mafia though...
Shining is probably the biggest question mark in my mind.
|
On January 26 2015 02:46 rsoultin wrote:Show nested quote +On January 26 2015 02:40 jarjarbinks wrote:On January 25 2015 06:43 Damdred wrote:
I'm wrong more often than not, LM is a lower activity player. And I could be tunneled, and i'm probably wrong about Geript. I think this quote shows when/why he began switching. Yeah, that was his explanation for the vote. Did you read the quotes I posted and my comments? That's what I need the sanity check on.
Ya idk your argument didn't really convince me of anything. You have quotes there, but idk if they are as implicating as you believe they are. I'll go double check.
|
Sorry computer issues.
Double check has me seeing confusion on the Damdred flip, but I didn't get the same end result as you did. It is weird to me that Damdred would flip and still say that LM is a good lynch when LM still had a ton of traction. I think that if he knew that LM would flip town, as soon as he flipped to Breshke he would have not said anything later about LM being a good lynch. I would think he would want it to look like he realized LM was town from DP's argument and switched on to someone else because of the argument so he's slightly cleared of suspicion when LM flips town.
Actually the hard-bus trfel part of the whole rsoultin=mafia thing makes little sense to me. I don't see you taking such a big risk for no reason. The only reason why I felt you were suspicious was because some of your play has paralleled his (lol). It seemed to me like Trfel had already planned to defend me and get on LM. Like pretty much when the game started. He always defended me from the get go and went after LM. Before I thought that meant he was bussing LM, obviously that wasn't the case. You have done the same thing in this game. The reason why I didn't feel like it was worth bringing up is because I saw that you as a person would probably defend me in the game and that LM was a relatively easy guy to take for mafia until EOD yesterday. Your resistance to change EOD yesterday does follow Trfel's planned actions. It could be coincidence though, but I guess I am allowed to be suspicious lol
|
I could definitely see that.
How does that statement presuppose that the main wagon is town? I missed that part.
In your mind, You me bresh shining are all town. Is that still true? That would make dp or geript 1 mafia and damdred the other?
|
hmm ya I don't know. I think what happened here is that damdred pointed out an observation he has seen in the game that isn't 100% true and you not only called him out for it you also took it to the farthest conclusion.
When I read that I'm thinking he's not basing that observation on the vote/flip of LM. Its more on the basic knowledge of scum vs town making deadlines.
I could see myself as mafia waking up if I was pushing one of the wagons and wanted to see if it worked. I could also see myself as town doing the same thing if I thought I was right. That means I disagree with his statement in defining DP as town.
|
|
|
|
|
|
EPT
