Powerless mafia, mafia win

[aage]

Bernie Sanders has been replaced by Razorvich, on account of not having opened his role pm yet and last posting on the forums three weeks ago.

[blacky365]

Bernie Sanders has been replaced by Razorvich, on account of not having opened his role pm yet and last posting on the forums three weeks ago.

taking the abov into account, i recant my earlier statement that he would be more likely town... he could be anything.
I still stand by my comments re JFM though (unless I hear otherwise).

[dgz345]

I vote for lynching Razorvich before he gets here. (y)

[aage]

jfm10 has been replaced by Metsfanmax, on account of not responding to my PMs.

[dgz345]

i guess town odds will go down a bit now.

[Ragian]

You think mets is such a terrible player?

[dgz345]

You think mets is such a terrible player?

haha no. nothing about the players. its just that the lead we have gathered will probably be thrown out.

the thing is that when they are back now (both) none would want to lynch them.

but if both of them died and one of them flipped scum we would have a huge advantage.

We got the "no lynch" pattern from this day.

if both of them comeback the only correct thing would be to vote no lynch. (to get to the odd number of players by night kill)

[Razorvich]

I vote for lynching Razorvich before he gets here. (y)

too late I'm here

With no power roles, I suppose this game comes down to tripping people up on what they say, will be very interesting to see this pan out.
After reading the thread to get up to speed, this is the post that stood out for me, even though a few posts later Blacky attempts to qualify his statement.

The point of this game was to try to give rookie players, such as myself, a chance at having a game with no pr’s.

So I vote no lynch... it will be night 1 correct? and then someone will die. Am I correct here?

[Razorvich]

that should read IF I vote NO LYNCH

[dgz345]

UNVOTE

[dgz345]

So I vote no lynch... it will be night 1 correct? and then someone will die. Am I correct here?

Yes that was correct. but we dont want to kill 1 person.

We want to get to an odd number of players so there needs to die 3 or 1 person until next day.

therefor if mets doesnt come here we need to lynch one today. and if he gets here we vote no lynch. and only get the 1 night kill.

[DirtyDishSoap]

I'll just put this out. if one of them shows up as scum. their scum buddy is blacky, strike or DDS.

Could you be more subtle with cssting shade? What makes you think either of the 3 are scum?

[Pikanchion]

Until Metsfanmax posts it is better to lynch somebody than not to lynch somebody.

Unvote: No Lynch
Vote: blacky365

blacky365 has frequently made poor arguments (and stood by them after scrutiny), and clearly skimmed at times also.

[Metsfanmax]

Why is no lynch the way to go? It's something math based, I'm sure, but I suck at math.

I know the math and I agree to it but I don't know why day 1?

Later on in the game you can learn from NK choices therefore can force scum to make suboptimal lynches.

This is in a vanilla game I mean.. there aren't any PRs to try to save for a night action

There's a reason I explained this all before the game started, the benefit from doing this is applied every single lynch after we no lynch so the earlier the better.

Town never benefits in any way by lynching on an even number of players compared to the next odd number down. Think of it this way, if one were to treat the No Lynch day and night that follows as having never happened then compared to a game where the Town lynched every night:
1. The game lasts the same number of days and nights
2. The game features the same number of lynches
3. The game will have the same number of night kills
However!
4. Town will have a higher probability of lynching correctly every single night.

Whether we lynch or not today, Town gets three lynches before game over if Mafia don't get lynched, we have a choice between those lynches being at 20%, 25%, and 33.3%, or being at 22.2%, 28.6%, and 40%. Verifying this is exceptionally simple. Why would you ever pick the former unless you're scum?

I cannot be bothered arguing this any further (which is the main reason I brought this up before the game started — this is objectively the correct way to do things in this setup, and I'd rather risk having made the correct argument for Town to win and end up quietly dropping it in the case I roll Mafia, than end up making the correct argument only later as Town with people questioning my motives). So basically from this point on I will voting exclusively No Lynch or for anybody who argues for lynching until the total number of players is odd during a day phase.

Unvote: Skoffin
Vote: No Lynch

—

Also aage: Town has only a 29.84% chance of wining, not 35.21% (the T column is Town players, not Total players).

I am not a fan of this reasoning. The social effects of the lynches themselves is not included: as others have pointed out, it ignores the usefulness of establishing voting patterns earlier rather than later. And since the magnitude of this effect is pretty hard to establish, one should be very skeptical of Pika's analysis. Even aside from that the utility of lynching itself is not established or factored into this probabilistic analysis. In a game with only one mafia player, which is his setup, if you lynch the mafia player at any time you end the game. So there has to be some information in this calculation that relates to the accumulated effects from lynch attempts -- is there some point at which we've lynched enough times for the expected likelihood of having lynched the mafia player grows over 50%? If not, should we just not lynch at all until there are 3 players left? This is completely neglected, yet Pika's posts make it sound like a given that we should have a D2 lynch.

I don't want to litigate this endlessly but since it seems virtually every player has decided to just go along with Pika's analysis it seems like a rather important point.

[Metsfanmax]

Until Metsfanmax posts it is better to lynch somebody than not to lynch somebody.

Unvote: No Lynch
Vote: blacky365

blacky365 has frequently made poor arguments (and stood by them after scrutiny), and clearly skimmed at times also.

Your analysis isn't actually about the number of players remaining being even or odd; that's merely an instrumental effect of the underlying reasoning about deferring a lynch until there are fewer players left, all else equal. Suppose we followed your pattern of lynching on the days where there are an odd number of players remaining. We'd have two paths to endgame as a result of me being modkilled:

D1 No Lynch: 8 (D2) -> 7 (D3) -> 5 (D4) -> 3 (D5)
D1 Lynch: 7 (D2) -> 5 (D3) -> 3 (D4)

In both cases town lynched three times, and the same number of players remained when the lynches occurred. So why is it better to lynch somebody today (in the scenario before I posted)? At best it's neutral, and at worst it's net negative since it cuts down the amount of discussion we can have.

[dgz345]

Vote no lynch

[dgz345]

@mets

I don't know where you got your numbers from but there is 10 players and 2 of em are mafia.

[Pikanchion]

I am not a fan of this reasoning. The social effects of the lynches themselves is not included: as others have pointed out, it ignores the usefulness of establishing voting patterns earlier rather than later. And since the magnitude of this effect is pretty hard to establish, one should be very skeptical of Pika's analysis.

How could lynching earlier ever possibly provide more information or be more informative than lynching later if the same number of lynches will occur in either scenario? A voting pattern is only useful to Town if Town (either in whole or in part) can act in a way that utilises this information. We have no power roles, so the lynch is the only time we can make any use of it. You appear to be suggesting that we should lynch earlier so that on a later day, a day when we are No Lynching anyway, we have that information to help us. How could that information ever possibly help us at that specific moment?

Even aside from that the utility of lynching itself is not established or factored into this probabilistic analysis. In a game with only one mafia player, which is his setup, if you lynch the mafia player at any time you end the game. So there has to be some information in this calculation that relates to the accumulated effects from lynch attempts -- is there some point at which we've lynched enough times for the expected likelihood of having lynched the mafia player grows over 50%? If not, should we just not lynch at all until there are 3 players left? This is completely neglected, yet Pika's posts make it sound like a given that we should have a D2 lynch.

You clearly have no grasp of probability if you think multiple No Lynches could ever be the optimal strategy for Town in this setup. Lynching at 7 players cannot prevent Town from lynching at 5 players, so why let the Mafia make 2 kills (by No Lynching twice) when you could have Mafia make one and Town make the other? What is gained by this additional day of doing nothing? How can Town make use of these crucial voting patterns in doing so?

Your analysis isn't actually about the number of players remaining being even or odd; that's merely an instrumental effect of the underlying reasoning about deferring a lynch until there are fewer players left, all else equal. Suppose we followed your pattern of lynching on the days where there are an odd number of players remaining. We'd have two paths to endgame as a result of me being modkilled:

D1 No Lynch: 8 (D2) -> 7 (D3) -> 5 (D4) -> 3 (D5)
D1 Lynch: 7 (D2) -> 5 (D3) -> 3 (D4)

In both cases town lynched three times, and the same number of players remained when the lynches occurred. So why is it better to lynch somebody today (in the scenario before I posted)? At best it's neutral, and at worst it's net negative since it cuts down the amount of discussion we can have.

This is plainly incorrect, your "potentially net positive" scenario allows the Mafia team to make an additional kill over my scenario, the Mafia kill will always be a Town player. It is objectively, quantifiably worse.

—

Unvote: blacky365
Vote: No Lynch

[dgz345]

With AFK Lynch:
D1(10) No lynch (Kills 2 AFK and 1 Night Kill --> D2(7) --> Lynch + Night KillD3(5) (3 town + 2 scum) (MyLo Chance to mislynch 60%)

1 AFK D1 No lynch:
D1(10) No lynch (Kills 1 AFK and 1 Night Kill --> D2(8) --> Lynch + Night Kill D3(6) (4 town + 2 scum) (MyLo Chance to mislynch 67%)

With 1 AFK and Lynch:
D1(10) Lynch (Also kills 1 AFK and 1 Night Kill --> D2(7) --> Lynch + Night Kill D3(5) (3 town + 2 scum) (MyLo Chance to mislynch 60%)

No AFK D1 Lynch:
D1(10) Lynch + Night Kill --> D2(8) Lynch + Night Kill --> D2(6) (4 town + 2 scum) (MyLo Chance to mislynch 67%)

No AFK D1 No lynch:
D1(10) No lynch + Night Kill --> D2(9) Lynch + Night Kill --> D3(7) Lynch + Night Kill --> D4(5) (3 town + 2 scum) (MyLo Chance to mislynch 60% )

[dgz345]

Hate that I can't edit posts ;;__;;

[Metsfanmax]

@mets

I don't know where you got your numbers from but there is 10 players and 2 of em are mafia.

Yes, but Pika's post made the argument that the same reasoning applied even if there were multiple mafia players, and at the time he made the post aage made it sound like we were most likely to have one mafia player, not two, in a 10-player game, so it just makes it easier to analyze in the lens of having a single mafia player.

How could lynching earlier ever possibly provide more information or be more informative than lynching later if the same number of lynches will occur in either scenario? A voting pattern is only useful to Town if Town (either in whole or in part) can act in a way that utilises this information. We have no power roles, so the lynch is the only time we can make any use of it. You appear to be suggesting that we should lynch earlier so that on a later day, a day when we are No Lynching anyway, we have that information to help us. How could that information ever possibly help us at that specific moment?

I think that D1 lynches in a game with town PRs are usually net negative because of the chance that you mislynch the town cop. But in a vanilla game, everything changes. D1 lynches are incredibly interesting in a vanilla game because no one has anything to cling to, no way to generate misinformation based on roles or previous events. It's pure scumhunting ability matched up against pure ability to hide in plain sight. The longer we talk (as long as there's some structure to how we do it), the more information we can extract from people and the likelier we are to catch scum in a lie. Remember there is no day deadline, so we have as long as we want to figure out something interesting to do. I think that we should -- D1 is possibly the most important day we will have in this game, if for no other reason than we decide whether to go with Pika's strategy or not, and there should be substantive discussion associated with that.

Along those lines, we don't need to actually carry through a lynch for useful information to be generated by voting patterns. For example, suppose we agreed to work together to establish a tentative lynch candidate on days with even numbered players left (kind of like a straw poll), and require everyone to pick a candidate or face a policy lynch, and continue until we actually reach a majority proposed lynch, then agree to just not actually follow through with it (everyone officially votes No Lynch at that point). Then we've held people's feet to the fire and gotten some interesting voting patterns on the record. Because that voting pattern was on the record when the mafia selects their night kill we have altered the decisions they make, and we will learn something from who turns up dead based on what their vote would have been. If we just end the even numbered-player days as soon as possible, as you have basically proposed, we lose the possibility to have two times as many votes! So your arguments earlier about how D1 (and D2) just invites WIFOM instead of constructive dialogue is apparently because you didn't think very hard about whether there was in fact a way we could make the day constructive with a little structure (or you intentionally chose not to because you're scum).

Since the rest of your comments are constructed on top of this fallacy that nothing useful happens on the non-lynch days, I won't respond to the rest.

[Metsfanmax]

Well again, kills from this mechanic would be favourable for Town in all likelihood, so I'm all for it. If the six of us who have posted all voted No Lynch now we would kill four players, and if that wasn't a win, another would be killed by the remaining mafia player(s), leaving us with five players remaining. This should result in a 42.67% win rate for Town overall (2 Mafia survive 33.33% of the time resulting in a 13.33% win rate, 1 Mafia survives 53.33% of the time resulting in a 46.67% win rate, no mafia survive 13.33% of the time resulting in a 100% win rate, the aggregate of these results [0.33*0.13+0.53*0.47+0.13*1=~0.43] is a 42.67% Town win rate).

I don't buy these EV calculations that aage linked to (so I also don't buy this calculation). Yes, you end up with an EV (by their definition) of 2/15 = 13.33% in the case with 3 town and 2 mafia remaining (for example), but those calculations are done from the perspective of the lynch occurring randomly, and the lynches are not random. Let's simplify out correlations between votes related to activity on previous days by assuming we start fresh on D1 with 3 town and 2 mafia, and simplify even further by assume that town acts blindly in choosing a lynch candidate (rather than some form of scumhunting taking place), so that effectively they are rolling a die. A key concern is that mafia is not acting blindly since they have information town does not (they know the alignment of every player; a town player only knows the alignment of themselves). So it is simply wrong to suggest that a mafia player is equally likely to get lynched compared to a town player, because a mafia player will presumably never vote for either of the two mafia players. If the two mafia players act in concert to target a particular townie, and the three town players still effectively act randomly, then the end result is a much greater chance of lynching the player the two mafia selected (only one of the three town has to unintentionally defect) than the result of lynching one of the two mafia players (all three town have to choose correctly). So the EV calculation they present is actually an upper bound for town's winning chances, and the real likelihood may be substantially lower.

[Pikanchion]

I think that D1 lynches in a game with town PRs are usually net negative because of the chance that you mislynch the town cop. But in a vanilla game, everything changes. D1 lynches are incredibly interesting in a vanilla game because no one has anything to cling to, no way to generate misinformation based on roles or previous events. It's pure scumhunting ability matched up against pure ability to hide in plain sight. The longer we talk (as long as there's some structure to how we do it), the more information we can extract from people and the likelier we are to catch scum in a lie. Remember there is no day deadline, so we have as long as we want to figure out something interesting to do. I think that we should -- D1 is possibly the most important day we will have in this game, if for no other reason than we decide whether to go with Pika's strategy or not, and there should be substantive discussion associated with that.

Along those lines, we don't need to actually carry through a lynch for useful information to be generated by voting patterns. For example, suppose we agreed to work together to establish a tentative lynch candidate on days with even numbered players left (kind of like a straw poll), and require everyone to pick a candidate or face a policy lynch, and continue until we actually reach a majority proposed lynch, then agree to just not actually follow through with it (everyone officially votes No Lynch at that point). Then we've held people's feet to the fire and gotten some interesting voting patterns on the record. Because that voting pattern was on the record when the mafia selects their night kill we have altered the decisions they make, and we will learn something from who turns up dead based on what their vote would have been. If we just end the even numbered-player days as soon as possible, as you have basically proposed, we lose the possibility to have two times as many votes! So your arguments earlier about how D1 (and D2) just invites WIFOM instead of constructive dialogue is apparently because you didn't think very hard about whether there was in fact a way we could make the day constructive with a little structure (or you intentionally chose not to because you're scum).

And again, how is this any better than doing the same on the following day? All Mafia have to do in order to reduce your entire "system" to nothing is choose their kill randomly. This is WIFOM hell.

We could have a thousand votes unofficial lynches per day under any system, the entire idea is folly.

—

Since the rest of your comments are constructed on top of this fallacy that nothing useful happens on the non-lynch days, I won't respond to the rest.

Liar.

Even aside from that the utility of lynching itself is not established or factored into this probabilistic analysis. In a game with only one mafia player, which is his setup, if you lynch the mafia player at any time you end the game. So there has to be some information in this calculation that relates to the accumulated effects from lynch attempts -- is there some point at which we've lynched enough times for the expected likelihood of having lynched the mafia player grows over 50%? If not, should we just not lynch at all until there are 3 players left? This is completely neglected, yet Pika's posts make it sound like a given that we should have a D2 lynch.

You clearly have no grasp of probability if you think multiple No Lynches could ever be the optimal strategy for Town in this setup. Lynching at 7 players cannot prevent Town from lynching at 5 players, so why let the Mafia make 2 kills (by No Lynching twice) when you could have Mafia make one and Town make the other? What is gained by this additional day of doing nothing? How can Town make use of these crucial voting patterns in doing so?

Your analysis isn't actually about the number of players remaining being even or odd; that's merely an instrumental effect of the underlying reasoning about deferring a lynch until there are fewer players left, all else equal. Suppose we followed your pattern of lynching on the days where there are an odd number of players remaining. We'd have two paths to endgame as a result of me being modkilled:

D1 No Lynch: 8 (D2) -> 7 (D3) -> 5 (D4) -> 3 (D5)
D1 Lynch: 7 (D2) -> 5 (D3) -> 3 (D4)

In both cases town lynched three times, and the same number of players remained when the lynches occurred. So why is it better to lynch somebody today (in the scenario before I posted)? At best it's neutral, and at worst it's net negative since it cuts down the amount of discussion we can have.

This is plainly incorrect, your "potentially net positive" scenario allows the Mafia team to make an additional kill over my scenario, the Mafia kill will always be a Town player. It is objectively, quantifiably worse

—

So the EV calculation they present is actually an upper bound for town's winning chances, and the real likelihood may be substantially lower.

Regardless of the validity of that statement, do you actually have a point or are you merely trying to discredit me for the sake of it?

[Metsfanmax]

And again, how is this any better than doing the same on the following day?

They're not exclusive; we would do both. The D1 vote potentially has additional value because it still requires players to take a stand on the alignment of other players.

All Mafia have to do in order to reduce your entire "system" to nothing is choose their kill randomly. This is WIFOM hell.

That would be fine with me. If Mafia have to do something other than what they would regularly prefer to do because town took some action D1 instead of sitting around twiddling their thumbs, I consider that an obvious choice if you're town.

We could have a thousand votes unofficial lynches per day under any system, the entire idea is folly.

It is folly if you treat it like folly, sure. But if we decided to take it seriously and we constructed useful discussions or votes we could have then perhaps it could lead somewhere. Rejecting it out of hand without even being willing to discuss the possibilities does not make sense to me. Scum hunting requires discussion. I don't think this is considered a strange concept in the mafia world.

Regardless of the validity of that statement, do you actually have a point or are you merely trying to discredit me for the sake of it?

I am trying to discredit your argumentation, not for its own sake but because I think you're attempting to take us down the wrong path using poor math. You can maybe use this sort of reasoning while trying to construct balanced setups but a real game is much more complex than merely the expected value calculation.

[BuJaber]

I don't think blacky is more guilty than dgz. Both are trying to draw conclusions from the two players that didn't post but their reasoning is what's interesting here.

Dgz has made the entire game about 2 players based on a "hunch". He jumped the gun and made it harder for anyone to defend these players without risking being mislynched, and what's worse is that if he meant what he said and he was actually town then he ruined the trap by not letting scum post and associate themselves with these two before calling them out on it.

It sounds to me more like forcing an associative to push more lynches away from himself, and by extension, a partner, than someone actually following a hunch and trying to find a partner for his top scum suspect.

Blacky's reasoning is more naive than nefarious. It's a common trap for newbies in particular to give more weight to real-life explanations of why someone is town or scum than is logically reasonable. I believe blacky is much better than he gives himself credit for but he has already shown a slight weakness in logical thinking in his arguing with dgz. And I think dgz is using that to win the irrelevant part of their argumemt while the relevant part goes unignored or unprotested.

If we're lynching someone day 1 I'm lynching dgz NOT blacky.