Saturday, May 03, 2008

Liar's poker

I say it again: it's always fun to find willful violations of Grice's maxims of cooperative communication.

Today's violated maxim: In the category of Quantity -- give complete information.

Sometimes I feel comfortable giving information at the poker table. When I'm feeling cocky.

Well I will tell you this: my first bet was a bluff. Of course that leaves it open. My later bets might be backed up by what came on the flop, turn or river.

I was chasing a hand. And maybe I caught it. I haven't given all the information.

But those aren't violations. They're cheeky. They're transparently opaque. I'm not misleading you because it should be clear that while what I've said is true you don't have enough information to know how to bet.

Poker After Dark: Match 23 -- "Of Mouth and Men". Jamie Gold and Alan 'Boston' Dvorkis face each other on a hand.

Boston is holding queen-10 and Gold is holding a pair of kings. After Gold raises Boston's bet Boston reports that he has Queen high. Gold says that he has king high.

Boston shows a queen and Gold shows a king.

Gold: I don't have king-queen so I don't have any of your cards.

After thinking and mumbling about it Boston reveals suspicion.

Boston: If you got king-queen that's really sickening
Gold: I don't have king-queen.
Boston: well you'd want to tell me that to get me call this so that i'd really make a bad call
Gold (showing his card to Paul Wasicka): Do I have king-queen?
Paul: Nope.

Boston doesn't trust Gold and suspects that he might have ace-king. Apparently thinking he wouldn't lie about not having a queen but might lie about having king high.

Gold: I said I have king high. I don't have ace-king
Boston: You have king high? You have king high?
Gold: Yes.
Boston: And you're claiming not to have king-queen?
Gold: I do not--

At this point Jamie has already said that he doesn't have king-queen. But Boston's latest question isn't about Jamie's hand. It's about Jamie's claim. Even had Jamie not yet claimed to not have king-queen, if he answers 'yes' to Boston's question there would be an entailed null speech-act of claiming to have a hand by answering a question that asks if the claim is made. Boston pushes for another claim.

Boston: So you don't have kings you just have king high?

Aha! Now he has revealed that he understands "king high" to mean only one king and another lower and not equal card. This was of course known -- but he has now made it an explicit claim.

How does Jamie respond? Does he confirm Alan's definition?

Gold: You know what -- you make your own decision. What do you want me to tell you. I gave you enough information.

So here's where Boston probably should have noticed that Gold is unwilling to answer his question. What does Gold mean by "enough information"? Gold surely knows that Boston defines king high exclusively. But he's unwilling to give the relevant information when asked. Is that cooperative? It probably would be if he made it clear that he was unwilling to answer specifically because Boston asked about a pair. But it could also be that Boston has asked for reassurance too many times. Here Gold's refusal to answer (which is a response) might be violating the maxim of manner by being ambiguous. But is it a violation in a game in which ambiguous information is an agreed manner of conversation? The problem is the changing frames of expected cooperation. In poker it is perfectly acceptable not to give a single word in response. But once you create a more specific frame of cooperative agreement the edge of the frame becomes important. Gold, by deflecting the question makes that edge fuzzy. It blurs the line between conversational implicature (I've said enoughI've already said so) and transparent opacity.

Boston calls the bet. Both players show their cards.

Gold: highest card in my hand is a king.

Boston: That's not king high.

Gold: Highest card I have is a king.

Boston: Yeah. That's kinda [bleep]. You got me.

There's an important distinction between the maxim of unambiguous information (category: Manner) and the maxim of sufficient information (category: Quantity): Violations of ambiguous information often leave the hearer feeling that information has been left out. Violations of sufficient information are likely to mislead because the hearer often believes that all information has been given.


  1. Are those really violations? I mean, this isn't a "conversation", where both speakers are presumed to be cooperating. The whole point (I gather, not being familiar with this version of poker) is to trick the other person and win.

  2. But there is an agreement that you don't mislead. And that's different from uttering a false statement.

    There's even a rule that if you tell another player your hand you can lie (because it is assumed that you would) and if you are telling the truth you have to make it clear that you are telling the truth. Either by assurance or proof. You can't say you have two aces flippantly if you do in fact have aces.

    Cooperative maxims are observed. And a flat out lie isn't considered fair play. You try to fake-out other players but only through ambiguity. Honour among thieves.

    I think we do have some violations. Because a frame was offered within the game where an agreement of cooperation was in play -- altho the maxim of unambiguous manner is the one that isn't usually expected.

    The definition of "king high" really mucks it up here because it can be either a structurally comprehensible description -- in which case 'two kings' would be a possible meaning; or it could be a lexical description -- in which case only one card is king and the other is lower.

    Gold obviously starts off playing with the difference between structural and lexical usage and when Boston indicates that he is working with the lexical use Gold shuts up about it.


Thanks for reaching out.

You can also contact me at wishydig[at]gmail[d0t]com.