Slashdot is powered by your submissions, so send in your scoop


Forgot your password?
Stats AI Supercomputing

Humans Dominating Poker Super Computer 93

New submitter IoTdude writes: The Claudico super computer uses an algorithm to account for gargantuan amounts of complexity by representing the number of possible Heads-Up No-limit Texas Hold'em decisions. Claudico also updates its strategy as it goes along, but its basic approach to the game involves getting into every hand by calling bets. And it's not working out so far. Halfway through the competition, the four human pros had a cumulative lead of 626,892 chips. Though much could change in the week remaining, a lead of around 600,000 chips is considered statistically significant.
This discussion has been archived. No new comments can be posted.

Humans Dominating Poker Super Computer

Comments Filter:
  • by ranton ( 36917 ) on Friday May 01, 2015 @10:26PM (#49598459)

    How are we supposed to have any idea what a cumulative lead of 626,892 chips means without knowing how many total chips there are? If there are 650,000 chips then the game is almost over, but if there are 1,000,000,000 chips then there hasn't been any movement at all.

    This is some pretty poor journalism.

    • They're playing a set number of hands, or for a set amount of time. So there's no chip limit.
      • by Anonymous Coward

        They're playing a set number of hands, or for a set amount of time. So there's no chip limit.

        Do you know how poker works? If there is no chip limit, then there is no gamble, no risk of loss, no opportunity for gain. There is no poker without a chip limit.

        • by Mal-2 ( 675116 )

          You can play Limit without worrying about chip stacks, but yeah, No Limit is heavily dependent on having to decide when to go all in and when to live to fight another hand. I have to assume there are re-buys.

          • They are playing "Doyle's Game" or variant, I forget the exact name. Every hand resets the amount of chips to 1,000 for both of two, heads-up players. This eliminates the advantage of stack size. Also, with previous no-limit hold'em bot vs. bot games, they would switch the ownership of hands after a hand, wipe mem, and replay. Can't do that exactly with humans, but I am waiting for the whole thing to finish before I read it in greater detail.

    • Re: (Score:2, Informative)

      by Anonymous Coward

      80000 hands * 20000 stack each hand = 1600000000 chips if humans win all hands.

    • They played 20,000 hands each.

      It was in mirrored play. So player A and B got one set of hands that they played against the computer; and C and D then got the computers hands and the computer got the same cards as A and B.

      This greatly reduced the variance in outcome.

      They were playing with 200 deep stacks (20,000$, with blinds of 50/100). So the humans were up by 31 buyins; or an average of about 8 buyins. Certainly not a huge margin of victory for that many hands.

  • by PAjamian ( 679137 ) on Friday May 01, 2015 @10:29PM (#49598469)

    So let me get this straight. At the start of the game the computer calls all bets and then it lets the other players train it and change their strategy to take advantage of that training? And they thought this would beat a seasoned professional poker player?

    This is basically a beginning poker player (fresh blood) but who is more consistent. A pro will absolutely clobber it.

    • by Idarubicin ( 579475 ) on Friday May 01, 2015 @10:43PM (#49598509) Journal

      This is basically a beginning poker player (fresh blood) but who is more consistent. A pro will absolutely clobber it.

      In other words, either the researchers involved are complete idiots, or a Slashdot poster jumped to a useless conclusion based on a strawman argument spun from the summary. Hm.

      • by glitch! ( 57276 )

        Yes, your boolean logic is correct. And a very simple thought experiment will provide a solution. If you have a 2 and a 3, will you ever call? No. Never. Unless you plan to bluff. But there are so many middling hands to bluff from that might have a chance of actually winning anyway. So there is NO REASON to call with such a crappy hand, unless you really want your opponents to think you are psychotic. Can that be used to an advantage?!?

        My very limited experience with "Dogs playing poker" tells me that cards

        • by belthize ( 990217 ) on Saturday May 02, 2015 @12:18AM (#49598767)

          That's good entry level tight strategy but will get you cleaned out playing no limit against a seasoned pro. You really have to learn how to win with any two random cards against any flop. You don't make a run at the hand every time but you must occasionally. If you know your opponent only plays face cards and likes to slow play strong hands preflop I'd rather call a minimum bet preflop with 2 3 than with KQ. I can get out cheap or absolutely clobber him with a good flop and I avoid the risk that he has some Ax, KK, QQ, JJ or some other face cards that can put my KQ in an uncomfortable spot.

          Lastly, if in doubt never call. Either raise or fold. Calling should be a very deliberate play. Most folks call because they have no clue if they're winning or losing a hand. Hint, you're losing.

          • by Anonymous Coward

            as the saying goes. If you look around the table and can't work out the weakest players then you are the weakest player.

          • I'm not sure how many of those are actual words and how many of them you're making up.

        • The two lowest card you can not use in a straight, 2 and 7, are probably worse than a 2 and 3. I'm too lazy to do the math, but you can check for me :) I see a lot of straights when watching poker on tv.
    • by Anonymous Coward

      CMU has been at the forefront of poker AI for nearly a decade now... I'm pretty sure it's a lot more sophisticated than you suppose.

    • Yeah starting out by calling all the time is STUPID. I think this game has four players, so it'll have losing cards 75%. In the first-order analysis, you should therefore fold 75% of the time. Sometimes you should raise, so you should only call about 10%-15% of the time. With further analysis it gets (much) more complicated, but those percentages are in the ballpark.

      Their LEARNING algorithm might be good, it might be very good, but they're starting out with a strategy that sucks. Really sucks. The

      • if I understand, it is playing 4 heads up matches simultaneously. It is losing in 3 of the matches, and winning in 1. To even be winning in 1 is pretty damn amazing though, as I'm sure they don't lose to anyone who isn't damn good. It also looks like a second match could swing back its way. Nothing certain as they are only half way done, but this computer is doing damn well. I'm sure if I played 20 top level players in heads up, no limit, I would lose against all 20 by a pretty damn large margin.

        • > I'm sure if I played 20 top level players in heads up, no limit, I would lose against all 20 by a pretty damn large margin.

          Perhaps. Actual tournaments aren't generally heads up, they are 8 to a table, with as many tables as needed. In these actual tournaments, you can easily "beat" over half of the players by simply folding anything but the strongest hands. Source - I've actually done this in WSOP events. You can get "in the money" (win prize money) by simply playing super, super tight, folding

          • In chess, you CAN beat half the grandmasters while playing heads up against each of them. Simply copy the your first opponent's move against your second and vice versa. They end up playing each other, with you as the messenger.

            This would assume simultaneous play and an equal distribution of white vs black where you could choose what order the games would be addressed in. What if you were playing 3 grandmasters and you got black against all of them? What if you were white? You would have to get the white move from another game before you could make your first move.

      • A naive strategy that would beat most non pros would be as follows:
        At each round of betting, evaluate how many stronger hands there are than your cards and how many weaker hands there are. If there are N other players at the table, you should bet if the number of stronger hands divided by the number of possible hands is less than 1/N because you should assume that whoever has the strongest cards is still in the game, so the strongest opponent is the strongest hand of N random hands.

        • Yeah that's a restatement of my first-order analysis. At a table of four, you'll have the best hole cards 25% of the time, so you should fold if your cards aren't in the top 25% of possible pairs. There are patterns you can learn to know approximately how many hands beat yours.

          Further analysis brings out the fact that sometimes you can call cheap, get a good flop, then have someone bet large into your strong hand. So if you're playing against a table who bets small preflop and large postflop, you might

          • You are just going to get stacked when you have 95 percentile hand and someone has the nuts. Sometimes you just have to assume they have the nuts when you have the second nuts. Anyway, this is about heads-up poker. So you are thinking 50 percentile hands are playable, what do you do just jam any hand significantly better than that? It is really hard to exactly quantify, hence Claudico getting owned.

            • I was speaking preflop. Post flop is a whole other ball of wax. Post, you have more information not just about the cards, but from betting. So you can then analyze what the nut is, and who may have it (from their betting), and what it might cost you to showdown.

          • They are not playing four handed.

            It is four simultaneous games of headsup poker and it is mirrored hands (so we see where a human loses more of less than the bot with the same cards, and 'coolers' mostly get canceled out)

  • by Idarubicin ( 579475 ) on Friday May 01, 2015 @10:48PM (#49598519) Journal

    I don't know about the technology or the algorithm(s), but the linked article is certainly nonsense.

    “You could use the same basic framework to do robust decision making like trying to come up with insulin and glucose monitoring plans [for diabetes patients],” says Neil Burch, a computer scientist at the University of Alberta who helped design a poker-playing AI earlier this year. “You get regular snapshots of glucose levels, and you have to decide how much insulin you should take, and how often.”

    Look, I get it. Nobody wants to admit that they're spending their grant money this way because it's fun to get a computer to play Hold 'em. But that's got to be the dumbest justification I've ever read. Human metabolism is complex, but the pancreas doesn't bluff.

    • by khallow ( 566160 )

      But that's got to be the dumbest justification I've ever read. Human metabolism is complex, but the pancreas doesn't bluff.

      It means they're solving a harder problem.

      • Re: (Score:2, Interesting)

        by Anonymous Coward

        but poker is about much more than just the odds.. it's about emotions, reading players expressions and tells, intimidation, and taking advantage of all that. playing 'perfectly' based on percentages only gives you a chance at staying in the game, it's mastering the 'human factor' that gets you all the chips.

        i don't see the point to training a computer solely based on odds and what happens 'on the table', and i don't see how you could possibly program a machine (with today's technology, anyway) to exhibit th

        • "but poker is about much more than just the odds.. it's about emotions, reading players expressions and tells, intimidation, and taking advantage of all that."

          That's your saying.

          "i don't see the point to training a computer solely based on odds and what happens 'on the table'"

          Therefore the research.

        • or did anybody else picture Data playing poker with the command crew of the Enterprise? [] Thanks for another smile Mr. Roddenberry [] :)
        • What about online play? IMO, reads are only reliable when one player is significantly outclassed or comes to the table dragging a huge amount of cognitive-dissonance between their actual and self-assessed level of emotional intelligence. Good players can wear sunglasses or not, more just an idiosyncrasy of the individual. They don't care if you can see the white's of their eyes or not, because they can be a brick wall if they want to or they can act happy or sad. And more importantly, they know that YOU can

    • The pancreas don't bluff, but sensors do. In the real world, you can't measure everything with perfect certainty, so you have to make guesses. Thing also start to get interesting when you have to make decisions off of these guess, should you gather more information to improve your guess or should you go ahead and make a decision?

      One of the people involved also claimed to have solved texas hold em' poker, if this is proven to be true then they've made a major advance in game theory. It means that systems tha

    • If I had diabeties I don't think I would want a poker playing AI controlling the insulin. What if it decides to go "all in"?

  • Not surprising (Score:4, Interesting)

    by GammaKitsune ( 826576 ) on Friday May 01, 2015 @11:06PM (#49598563)

    Lt. Commander Data struggled with the intricacies of poker as well.

  • by arbiter1 ( 1204146 ) on Friday May 01, 2015 @11:19PM (#49598609)
    Playing poker is a lot more about reading your opponent. figuring out if they are buffing or if they really have something. No computer can do that. Not at least at this point in time.
    • Ever heard of card counting? Computers are impossible to beat at card counting.

      So any game where card counting is part of a useful strategy the computer has a huge advantage. That's why they were able to solve two-player limited Texas Hold 'em, card counting well gives you a slight advantage so the computer can win 52-53% of the time even tho it can't read it's human opponent very well.

      The version they're playing now has some problems, but apparently the problems are caused as much by the programmer's poor

      • by AuMatar ( 183847 ) on Saturday May 02, 2015 @03:58AM (#49599157)

        There's no card counting in Texas Holdem. The deck is reshuffled after each hand dealt. Only 7 cards are shown to a given player, and all of them can be read at any time. There's no advantage to card counting, because you don't need to count. They may have some other card game they beat, but it isn't holdem.

        • So I know 7/52 of the cards, two of which are secret to the other player, and I use that info to predict the cards the other player is statistically likely to have, thus informing my strategy on raises/calls/etc., but that's not card counting?

          Why not?

          • Card counting [] is a betting strategy that's mainly about when to play or not. Computing odds is not card counting, even though that sometimes involves counting the number of cards that satisfy some condition.

          • Because everyone can count to seven and unless they are blind and deaf they don't even have to, the dealer will gladly tell you what cards you can see, and therefore count, at any time. What advantage do you hope to gain by counting (remembering) the cards that are currently visible to everyone?

          • by AuMatar ( 183847 )

            Card counting is keeping track of cards between hands in an effort to figure out altered odds on the current hand. For example, if you're playing single shoe blackjack and have seen 10 high cards out of 11 cards, you know low cards have a higher probability than normal.

            That doesn't exist in Holdem, because there's no carry over between hands. Each is an individual event, with no altered probability from previous hands. You can calculate odds, but that's easy even for a human at holdem- if there's X card

    • Luckily for the computer, no one can tell if a computer is bluffing either.
    • by AuMatar ( 183847 )

      No, it isn't. Or at least, it isn't by looking for tells. You win money by analyzing their betting pattern on this hand, comparing it to what makes sense, and putting them on a range of possible hands. One of those possible hands is always a bluff. Then you see what you beat of those hands, what beats you, and what your drawing odds are to improve and make a choice off that information. That is definitely something a computer can do. But the question is never "is he bluffing" its "is my hand strong en

      • by Kjella ( 173770 )

        And the corollary to this is that if you play like range => bet, a good opponent will easily spot bet => range. Unlike chess where it always make sense to make the same play in the same position, it's crucial in poker to play the same hand in different ways. Of course you need to make the "correct" play most of the time, but it's when you set up a play where you have an unlikely hand like raising with garbage but hitting the flop or triple-barreling a monster your opponent thinks is a bluff that huge

  • by dlenmn ( 145080 ) on Friday May 01, 2015 @11:29PM (#49598647)

    I almost did a double-take with this story; a few months ago I read about computers having solved heads-up _limit_ Texas hold’em: []

    Well, it looks like the computer can win when there is a limit, but humans can still win when there is no limit.

    I guess that's not too surprising: the limit really cuts down the number of choices, making a brute-force calculations more practical, and brute-force calculations are what computers do best. Without the restrictions of a limit, the AI needs to be a lot more clever. I wonder how long it'll be until computers win at this.

  • There's no way the computer is going to win at Strip Poker.
    • Well, they could win, but it'd be disappointing for everyone.
      • I don't know. Me, a girl who'd like to see me with fewer clothes (and I'd like to see in fewer clothes), and a computer that's kicking both of our asses? Sounds like an ideal strip poker opponent.

        You can even throw in a couple of people I don't want to see naked (and/or do not want to see me naked) and it's still a good time.

        • Haha. Fair enough.

          Although I'm not sure how well you'd fair with the line "hey baby. Wanna go back to my place and place strip poker with my super computer?"
    • In strip poker against a computer, *nobody wins*

  • A person can read the "tells", or body language and mannerisms, of another player. The computer cannot. Presumably the computer has no "tells" of its own, but an experienced player should usually win - because poker isn't a game of math, it's a game of psychology. Or rather, it's only PARTLY a game of mathematics and probability.

    • The computer has tells. If you figure out the algorithm you can figure out by it's pattern of calls/raises what it's hand is. In theory if you knew the code well wnough you could reverse engineer the computer's hand, and figure out it's strategy, based entirely on the cards showing and it's behavior; altho in practice virtually nobody could pull that off.

      The problem seems to be it has no way to read human tells, and it's designed to pursue a bad strategy (calling) most of the time.

    • Poker is only a game of psychology to things with the ability to exploit and be exploited by psychology. To a computer, poker can just be all math. a computer can also try to analyze behavior to look for patterns (e.g. like tells), in essence to play more like a human, but this also opens up the door to being tricked like a human. The computer may not be as good at playing like a human, but the human absolutely sucks at playing like a computer, and a computer always has to option to keep the games all ma
    • by ledow ( 319597 )

      Tells and the "psychology" of poker are about supplying misleading as to the content of your hand. That's it.

      The computer doesn't need to take any notice of you, misleading or otherwise. It knows what the chances of any particular card in your hand are. It therefore knows exactly the odds of whether its hand is likely to outmatch all the other hands on the table.

      The problem is not in playing the game, it's in betting (especially with no-limits, which gives too many avenues for recursion so it just has to

  • I'm pretty sure Deep Blue lost the first few thousand chess matches, real or simulated.
  • You brought her; you poke her.

  • by Dr. Spork ( 142693 ) on Saturday May 02, 2015 @02:06AM (#49598959)
    First of all, this is the link [] that the story should have included. It includes updates of the scoreboard, etc. On it you will see that even though the brains are collectively beating Claudico, the computer is actually over $100,000 ahead against Jason Les, a feat that almost no human could match. Yes, Claudico is down against the other three, but these are the top players in the world, and most human pros would get clobbered much worse by these guys. Are we really so hard to impress? This is the first time that something like this has been tried, and already, the computer is performing on a level that most poker pros would love to reach.
  • The Claudico super computer uses an algorithm to account for gargantuan amounts of complexity by representing the number of possible Heads-Up No-limit Texas Hold'em decisions.

    Ummm, wrong. It's the other way round. You don't put shoes on by keeping your feet warm, do you?

  • by tmosley ( 996283 ) on Saturday May 02, 2015 @07:36AM (#49599533)
    Humans were designed over a long, intense period of selection to be able to perform deep, self-referential thinking, ie to be able to know that the other guy knows that they are bluffing/lying/have some particular knowledge, and make plans accordingly. It is probably the single task that humans are best at. I'm not saying that having a machine beat humans at poker will mean the Singularity has arrived, but when they go one step farther and start beating us at politics (which requires the same skillset as poker but with more complexity, plus the ability to integrate other types of knowledge), then it's all over.

    I must say that I am looking forward to it. Humans are much better at beating others in the race to gain power than they are at actually ruling and making decisions for other people.
  • So, one Claudico and four humans. The odds are already stacked against Claudico 20%-80% that a human will come out tops.

    I wonder if the outcome will be different had there been more than one Claudico at the table.

Overflow on /dev/null, please empty the bit bucket.