Mafia-style Go

[Kirby]

I think it'll work. I'm in

[prokofiev]

a) Players rotate choosing moves (voting for moves just complicates things and would take way too long)

b) Every N (say for 9x9 with 8 players, i.e. 4 per side, use N=1; for 19x19 and 8 players, use N=3 maybe) cycles through each player moving, everyone votes someone off

How do you ensure people have to make moves for both sides? Especially when there's an even number of players? Although I agree about how long voting for each move will take, I think we lose a lot of information without it. Perhaps to compensate we have to encourage a lot of table-talk: everyone tries to persuade whomever is moving next of their preferred move...

I don't think it's fair to hold the first vote until we've seen everyone make a move for both sides.

It's a good idea for this to happen for 13x13 or 19x19 games. With an even number of players, you'd just have 1-8 go and then 2-8 go again, and then 1 (or some similar scheme) if N is at least 2.

For 9x9, the game is going to be short anyways. There's still information in just a move for one side. Also, mafia-style games should have some degree of randomness in terms of who you're voting off. If there's too much information (a long list of votes on every move) it seems it'll either be too obvious what's going on or people will have to vote for moves that are very subtly good/not good.

(That said, anything is fun to try and I don't mean to suggest the other way won't work, just to put out a complete proposal with several ideas that seem desirable to me.)

[Kirby]

Yes, I should have said that I agree the situation on the board definitely needs to count for something.

Ah, gotta love miscommunication.

[Joaz Banbeck]

You have incentive to play well enough to help your team to win. You have incentive to play poorly enough not to get voted off.

Kirby is right, I think. But has anybody followed this to its logical conclusion? You can't help your team if you get voted off. Therefore, Staying in outweighs playing good moves for your team. You don't have to play good moves, because the other team - by the same logic - will not be playing good moves either.
You simply have to outlast others, then when you can't get voted off, you start playing good moves. ( I fear that there will probably be plenty of them to be played. )

The most important goal is NOT to make good moves, it is to determine the affiliation of the strongest players, and get rid of them if they are your opponent.

The best strategy is to play mediocre moves to disguise your true affiliation. Indeed, it may be safest to make random moves. Your affiliation cannot be determined if you publicly base your moves on some random public event.
An example of that is this: suppose the play is in the lower left quadrant. I announce that I am basing my move on the score of the Lakers vs Spurs game. I further announce that my X-coordinate will be the last digit of the Lakers' score, and my Y-ccordinate will be the last digit of the Spur's score. ( Some collision algorithm is needed, but that doesn't change the overall logic ) When the game is over, I know my move. So does every one else. I clearly have made no input into it whatsoever. My affiliation cannot be determined at all.

[prokofiev]

You have incentive to play well enough to help your team to win. You have incentive to play poorly enough not to get voted off.

Kirby is right, I think. But has anybody followed this to its logical conclusion? You can't help your team if you get voted off. Therefore, Staying in outweighs playing good moves for your team. You don't have to play good moves, because the other team - by the same logic - will not be playing good moves either.
You simply have to outlast others, then when you can't get voted off, you start playing good moves. ( I fear that there will probably be plenty of them to be played. )

The most important goal is NOT to make good moves, it is to determine the affiliation of the strongest players, and get rid of them if they are your opponent.

The best strategy is to play mediocre moves to disguise your true affiliation. Indeed, it may be safest to make random moves. Your affiliation cannot be determined if you publicly base your moves on some random public event.
An example of that is this: suppose the play is in the lower left quadrant. I announce that I am basing my move on the score of the Lakers vs Spurs game. I further announce that my X-coordinate will be the last digit of the Lakers' score, and my Y-ccordinate will be the last digit of the Spur's score. ( Some collision algorithm is needed, but that doesn't change the overall logic ) When the game is over, I know my move. So does every one else. I clearly have made no input into it whatsoever. My affiliation cannot be determined at all.

This is a fine observation. The implication of this seems to be that if the game knocks players off quickly enough to likely reach a state with only players on one side left, then you should play randomly, as you suggest, as you'll have plenty of time when only your team remains to trounce the opposition.

Two solutions to this: 1) have the game reach a steady state of 1 vs 1 at some point (as suggested above by MountainGo and favored by myself) or 2) have there still be "a good number" of players on each side near/at the end of the game (as suggested above by e.g. Kirby in terms of having the end of the game happen at roughly the same time as the last person being voted off)

If either of those is the case, then it's a matter of choosing your timing: when do you start making moves that benefit your side more than the other? This will either just have to be the case at some point late enough in the game, or I can imagine there being a move big enough that you should sacrifice/reveal yourself to play it (though with the voting method instead of the taking turns method, maybe several people on your team would be revealed at once).

If the rules do allow a point with only players on one team, maybe when that is the case one should require that none of their moves be truly awful (e.g. filling in an eye) as determined by the moderator.

[Edited for clarity shortly after posting.]

[Joaz Banbeck]

...I can imagine there being a move big enough that you should sacrifice/reveal yourself to play it (though with the voting method instead of the taking turns method, maybe several people on your team would be revealed at once)...

Hmmm...I hadn't thought about sacrifices. It can get really unstable then.
It happens like this: one guy observes that he can sacrifice himself by filling an eye and thereby killing a big group. He propoposes filling the eye. He thereby has revealed himself, and everybody knows that he will be voted off by the opponents in the next move.
The only way for his teammates to make his sacrifice worthwhile is to join hin. They have the choice between losing a teammate for nothing, or losing a teammate and killing a big group that will probably win the game. So they clearly opt for the latter. They are thereby revealing themselves.
The other team has to vote for something else. Indeed, they have to all vote for the same something else, for if they split their votes, they have revealed themselves but still not prevented the sacrifice. So they must somehow organize around an opposing candidate move.

Thus, as soon as there is a big group with minimal eyespace, one guy will sacrifice, all players will be revealed, and the game suddenly gets boring.

[MountainGo]

...I can imagine there being a move big enough that you should sacrifice/reveal yourself to play it (though with the voting method instead of the taking turns method, maybe several people on your team would be revealed at once)...

Hmmm...I hadn't thought about sacrifices. It can get really unstable then.
It happens like this: one guy observes that he can sacrifice himself by filling an eye and thereby killing a big group. He propoposes filling the eye. He thereby has revealed himself, and everybody knows that he will be voted off by the opponents in the next move.
The only way for his teammates to make his sacrifice worthwhile is to join hin. They have the choice between losing a teammate for nothing, or losing a teammate and killing a big group that will probably win the game. So they clearly opt for the latter. They are thereby revealing themselves.
The other team has to vote for something else. Indeed, they have to all vote for the same something else, for if they split their votes, they have revealed themselves but still not prevented the sacrifice. So they must somehow organize around an opposing candidate move.

Thus, as soon as there is a big group with minimal eyespace, one guy will sacrifice, all players will be revealed, and the game suddenly gets boring.

This is a good point. I guess we could just add the rule that those sorts of suicide moves are just not allowed, as determined by the moderator.

On another point, I noticed in the other thread that the proposed ruleset includes only public voting. I wonder if it would be better to have anonymous voting when you choose whom to "kill". That way we avoid the situation where a player clearly favoring one side in the move-making survives because people are afraid to vote to "kill" him since it would reveal them as being from the other team.

[daniel_the_smith]

I think nominations should be public and votes private.

[Phelan]

The reason I put down everything as public, is that people will have to analyse the nominations and votes. There is no way for someone to nominate a bad move, and the rest of his team to hide themselves.

I might have to add something to the rules about suicides, though. :/

I want to try it like this for now, and see what people come up with. The next game might have better rules, depending on what we observe happening.

[Joaz Banbeck]

...I might have to add something to the rules about suicides...

That would be terribly difficult to enforce. There are tremendous grey areas. Sure, you can enforce the obvious things that suicide a group, like filling in an eye. But what do you do when the sacrifice involves a snapback five move deep, and he swears that he didn't see it? Do you tell him "You're a 1D, therfore you should have seen it, therefore you are cheating!"?

[Phelan]

Yeah, I know, hence the :/ smiley. I don't really want to have a rule about that.

[Zwergesel]

My first post in this forum and it's in a Mafia Go thread
Anyway, here are my thoughts on Mafia Go:

I wouldn't play it with the same number of players on both teams — it's too easy for one team to just reveal themselves once they think they outnumber the others and then just vote the other team off. This is also possible on the very first move, since even if they are revealed, they would always get a draw in any vote. That's enough at least to bring the game to a halt or, if in case of a draw the vote is chosen randomly, gives them a 50% chance of winning.

Why not play it like this instead:
* Give the white team more players, let's say 6 white players and 3 black players
* The black players know each other and may communicate via private messages at all times
* The white moves are chosen by a vote
* The black moves are chosen by the black team and sent as PM to the gamemaster
* Every X moves one player gets voted off — the players may also choose to vote no one
* One white player gets secretly and randomly chosen as the "sensei" (or call it whatever you want ) at the start of the game
* In case of a voting tie that does not get resolved by one player changing his vote, the "sensei" may choose one of the tied options via PM to the gamemaster
* If the "sensei" gets voted off, he secretly passes on his title to any other player in the game
* The game ends if one side wins on the board. It also ends if the black team outnumbers the white or there are no black players left

X should be chosen so that the game can ends roughly at the point where only 2 players are left, e.g. with 9 players total on 9x9 with about 60 moves for a full game I would choose X as 60/7 =~ 9

Imho this captures the spirit of mafia games better, while still making it an interesting Go game.

[Phelan]

I thought about making it like that, but it would make choosing a komi and handicap that doesn't reveal the black players (the hidden team) harder. I like your rules, though.

I'd play it if there are more willing players, and you volunteer to moderate.