Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[faust]

If I'm town in the same scenario, does it make me more likely scum because it only happens 3% of the time?  Nope.  When I'm town, I'm town.  And yet, regardless of reads and play, ss's argument would be to lynch me as a fake claimant.

The point is that the expected number of correct lynches when you lynch you in that situation is 0.97. That's like 2-4 times the expected number of correct lynches when lynching someone else in that situation. 3% of the time, it ends up being a bad idea, but simply because it is a good idea 97% of the time, it's worth doing every time unless there is an extremely good reason to believe otherwise.

So could you just figure out the percentages for every player in a given game and then lynch by number?

That's pretty much how the game is played, though we usually call it "scum points".

[ashersky]

If I'm town in the same scenario, does it make me more likely scum because it only happens 3% of the time?  Nope.  When I'm town, I'm town.  And yet, regardless of reads and play, ss's argument would be to lynch me as a fake claimant.

The point is that the expected number of correct lynches when you lynch you in that situation is 0.97. That's like 2-4 times the expected number of correct lynches when lynching someone else in that situation. 3% of the time, it ends up being a bad idea, but simply because it is a good idea 97% of the time, it's worth doing every time unless there is an extremely good reason to believe otherwise.

So could you just figure out the percentages for every player in a given game and then lynch by number?

That's pretty much how the game is played, though we usually call it "scum points".

I know you are joking, but that's basically the opposite of what I said, right?

To use the ss/Awaclus (SAwaspawn) method, in a game of asher9++, full claim from the beginning, run the percentages for each player's claim, lynch the least likely until you win.  Right?  Why even play?  Just have a script do it for you.

[yuma]

If I'm town in the same scenario, does it make me more likely scum because it only happens 3% of the time?  Nope.  When I'm town, I'm town.  And yet, regardless of reads and play, ss's argument would be to lynch me as a fake claimant.

The point is that the expected number of correct lynches when you lynch you in that situation is 0.97. That's like 2-4 times the expected number of correct lynches when lynching someone else in that situation. 3% of the time, it ends up being a bad idea, but simply because it is a good idea 97% of the time, it's worth doing every time unless there is an extremely good reason to believe otherwise.

So could you just figure out the percentages for every player in a given game and then lynch by number?

That's pretty much how the game is played, though we usually call it "scum points".

I know you are joking, but that's basically the opposite of what I said, right?

To use the ss/Awaclus (SAwaspawn) method, in a game of asher9++, full claim from the beginning, run the percentages for each player's claim, lynch the least likely until you win.  Right?  Why even play?  Just have a script do it for you.

Because the more optimally way to play is what we did here... Wait to claim until later when the math is more meaningful because it is backed up by reality--confirmed flips--and then combine that math with the thousand of posts that were made for an accurate read on the game.

People keep reflecting on to ss's run of this game to show their point. But that game in fact proved both points. At least for me. Cause in that game I ignored both  my gut read and math (because I was using bad math). This time we combined them intelligently and it worked out great.

[ashersky]

If I'm town in the same scenario, does it make me more likely scum because it only happens 3% of the time?  Nope.  When I'm town, I'm town.  And yet, regardless of reads and play, ss's argument would be to lynch me as a fake claimant.

The point is that the expected number of correct lynches when you lynch you in that situation is 0.97. That's like 2-4 times the expected number of correct lynches when lynching someone else in that situation. 3% of the time, it ends up being a bad idea, but simply because it is a good idea 97% of the time, it's worth doing every time unless there is an extremely good reason to believe otherwise.

So could you just figure out the percentages for every player in a given game and then lynch by number?

That's pretty much how the game is played, though we usually call it "scum points".

I know you are joking, but that's basically the opposite of what I said, right?

To use the ss/Awaclus (SAwaspawn) method, in a game of asher9++, full claim from the beginning, run the percentages for each player's claim, lynch the least likely until you win.  Right?  Why even play?  Just have a script do it for you.

Because the more optimally way to play is what we did here... Wait to claim until later when the math is more meaningful because it is backed up by reality--confirmed flips--and then combine that math with the thousand of posts that were made for an accurate read on the game.

People keep reflecting on to ss's run of this game to show their point. But that game in fact proved both points. At least for me. Cause in that game I ignored both  my gut read and math (because I was using bad math). This time we combined them intelligently and it worked out great.

Then, would a better way to use the SAwaspawn Method be X number of random lynches to confirm flips, then a forced mass claim, then lynches by percentages?

I'm just trying to prove SAwaspawn's point that reality, scum hunting, thoughts, feelings, or humans really don't matter.

[yuma]

I'm just trying to prove SAwaspawn's point that reality, scum hunting, thoughts, feelings, or humans really don't matter.

Well I don't think they (I guess they can speak for themselves) would agree that is their point and neither would I, nor do I agree with it.

What I am seeing is a lot of talking past each other...

[ashersky]

I'm just trying to prove SAwaspawn's point that reality, scum hunting, thoughts, feelings, or humans really don't matter.

Well I don't think they (I guess they can speak for themselves) would agree that is their point and neither would I, nor do I agree with it.

What I am seeing is a lot of talking past each other...

Dunno, Awaclus just said a 97% probability is enough to ignore all other factors.

If a person can specifically pinpoint the exact probability of a claim being a lie, why wouldn't you just do that and be done with it?  Assume a game of 12 players, and Awaclus and silverspawn do the math and come up with:

Player 1: 85.256%
Player 2: 82.14356%
Player 3: 73.1234%
Player 4: 66.232093%
Player 5: 65.4342111%
Player 6: 62.2313%
Player 7: 55.25162%
Player 8: 34.65235462%
Player 9: 31.2524673%
Player 10: 22.46562342%
Player 11: 9.243536%
Player 12: 1.3573%

Where the percentages are the chance they are lying.  Just do that at the start of each game, lynch from the top of the list, probably hit all scum before running out of mislynches, right?

That's what I understand to be their point: If the math is correct, reads and feelings do not matter.

[Awaclus]

Player 1: 85.256%
Player 2: 82.14356%
Player 3: 73.1234%
Player 4: 66.232093%
Player 5: 65.4342111%
Player 6: 62.2313%
Player 7: 55.25162%
Player 8: 34.65235462%
Player 9: 31.2524673%
Player 10: 22.46562342%
Player 11: 9.243536%
Player 12: 1.3573%

Where the percentages are the chance they are lying.  Just do that at the start of each game, lynch from the top of the list, probably hit all scum before running out of mislynches, right?

That's what I understand to be their point: If the math is correct, reads and feelings do not matter.

Well, if those were the percentages, then you could pretty much just do that. The problem is that you won't get percentages like that. Instead, the vast majority of the time, you will get percentages that are close enough to the 1/4 that every player already has anyway that it would only be a very marginal factor in determining who is the best lynch.

Reads and feelings do matter. If someone has, say, a 40% chance of lying but a very consistent town narrative while someone else has a 10% chance of lying but has done something that questionably shows that he doesn't have the information that he's claiming to have or has information that he has claimed not to have or that he's playing against the win con he has claimed to have, it might be correct to lynch the latter player if the former player's town narrative makes him ~1/2 times as likely to be lying as he otherwise would and the latter player's scumminess makes him ~3 times as likely to be lying.

However, if someone has a 40% chance of lying while someone else has a 3% chance of lying, you would have to read the latter person as being 13 times as likely to be scum as the former person to just make up for the mathematical odds. That's, like, the kind of read you can only have when a claimed Mason has voted for the other claimed Mason, or other similar practically-confirmed-scum situation. And sure, I would lynch the claimed Masons in that case, but I wouldn't lynch them only because one of them has been slightly scummier than the 40% chance of lying person.

[yuma]

How many versions of this game have been played?

Adventure Time - Town won - Cop, 1-shot Doctor, Universal Backup (CCE)
Homeland Mafia - Mafia won - 1-shot Vigilante, 1-shot Cop, Innocent Child, Universal Backup (VCMM)
Holiday Mafia - Town won - 1-shot Cop, 1-shot Doctor, Doctor, Universal Backup (CDD)?
Pony-Mafia - Town won - Innocent Child, Roleblocker, Universal Backup (MMB)
Fanficton - Mafia won - Roleblocker, Doctor, Universal Backup (BD)
Scout Mafia - Town won - Roleblocker, 1-shot Vig, Masons (BVM)

4 town wins to 2 scum. Any others that I missed?

[iguanaiguana]

Am I crazy? When did ss vote for his partner?

huh... looks like I didn't do that after all. I thought I had. However, chairs voted for me here. And I did say I found him scummy at some point, pretty sure.

You know what, didn't Chairs post in his QT that he forgot you were both masons for most of that day?

[Axxle]

So I just want to note that I TOTALLY FORGOT SS WAS MY PARTNER at the time I voted him (D1).

[iguanaiguana]

 

[2.71828.....]

[post deleted because it h=may not have been the pbest thing to post]

[gkrieg13]

[post deleted because it h=may not have been the pbest thing to post]

Are you drunk right now?

[2.71828.....]

but srlsy.  guys.  thicker skiins please.  I hate having civiility issues in my games.  it is just a game.

also, math is cool.  and math was correct....this game.  mafia is not about the odds though.  it is about the gut.  although maybe thatis bywh my lynch rate is so awful

[2.71828.....]

[post deleted because it h=may not have been the pbest thing to post]

Are you drunk right now?

What?  yes

[Axxle]

E for excellent mod

[faust]

E for excellent mod

Would have been even better had he been drunk the whole time.

[silverspawn]

Am I crazy? When did ss vote for his partner?

huh... looks like I didn't do that after all. I thought I had. However, chairs voted for me here. And I did say I found him scummy at some point, pretty sure.

You know what, didn't Chairs post in his QT that he forgot you were both masons for most of that day?

I honestly think (though I don't remember it exactly) that I'd have voted for him had he not voted for me by accident, though. pretty sure I wanted for one of us to vote the other from the start. so I'm still guilty!

[2.71828.....]

Sorry, I am way behind in everything.  Ash, I promise I will get an mvp and do that while historical record thing for this game.

Eventually

[yuma]

Sorry, I am way behind in everything.  Ash, I promise I will get an mvp and do that while historical record thing for this game.

Eventually

How about a sneak preview of the mod QT? I always enjoy reading about a mod berating town for being silly...

[2.71828.....]

Sorry, I am way behind in everything.  Ash, I promise I will get an mvp and do that while historical record thing for this game.

Eventually

How about a sneak preview of the mod QT? I always enjoy reading about a mod berating town for being silly...

It also has all the links to all other QTs contained within it
http://www.quicktopic.com/52/H/54aEurt5rfUFq

[XerxesPraelor]

I assume the game is done now - I'm pretty sure silverspawn is right here.

To address ashersky's point, the SAwaspawn method of asher9++ is bad because you can do better than it by using scum tells etc and general scumminess. But you shouldn't throw the percentages away - you just take the percentages as a starting point, and then work from there. If there were no day, and each player just sent their votes in to the mod, the SAwaspawn method would be close to being the best method (there's still some WIFOM, but it's closer). The thing is, though, what people do during the day changes the subjective probability that they are scum, and you have to take that stuff into accound.

[liopoil]

I haven't read everything, but I am leaning towards witherweaver when he says that the arithmatic is not relevent. Example:

There are two potential town power roles, one which has a 50-50 shot of being in the game, another which only has a 3% chance of being in the game. Someone gets to L-1 and claims the 3% role. This, in a vacuum, is meaningless. The only way to use the 3% figure quantitatively would be to take into account the whole setup and do a truly infeasible calculation. It certainly doesn't mean that there is a 97% chance they are lying. No, the only practical way to use it is to (depending on context, but probably) consider it as a 'very scummy' thing. Possibly this means lynching them for sure, but again, it completely depends.

I may have just constructed a strawman for what the other people are arguing, but the main point is that you have to be really careful with all the conditionals and givens, and usually that means you don't get anything meaningful.

[liopoil]

Wait, what? Don't lynch claimed masons until lylo or they are scum for suresies. Is that really what this was all about? Dice Mafia all over again... Arch and I lost that one because we didn't consider trying to get Masons mislynched.

[scott_pilgrim]

The probability should be the base chance that you're looking at.  If there's a 97% chance you're scum, independently of reads, then that's where you start from when you take reads into account.  I hope everyone can agree on that, since it's basically true just by definition.  The question is how big of difference reads should play.  I don't think pretty much any "soft" evidence could ever persuade me not to lynch someone who has a 97% "base" chance of being scum, because it's nearly impossible for someone to prove they're town.  I could conceivably be convinced to lynch someone who has a 3% "base" chance of being scum, but they'd have to come pretty close to just admitting they're scum.

What silverspawn is arguing (and I agree) is that people in general put way too much weight in reads.  It's way too easy to talk yourself into something, people are full of biases and really a lot of the time it's basically just a shot in the dark.  My intuition says that people should be looking at differences of like +/-20% based on reads, not like +/-100% which is basically what most people do, from what I can tell (just ignore probabilistic evidence or treat it like it's insignificant in comparison to their reads).  Unfortunately there's no way to support this claim (or the opposite), without testing it over thousands of games, which is completely impractical.

Wait, what? Don't lynch claimed masons until lylo or they are scum for suresies. Is that really what this was all about? Dice Mafia all over again... Arch and I lost that one because we didn't consider trying to get Masons mislynched.

Were you there for the asher++ game where scum claimed masons and won because town didn't do/trust probability?  There was like a 3% chance they were town, but people talked themselves into it because...I don't know why really.  I wasn't paying much attention to the game, but it was so painful to watch everyone ignore the overwhelming evidence.  Like at that point they're basically confirmed scum, but people don't care because it "feels" like they're town.