Less work on the software, more on the actuators
"Our bot has been known to rip people's arms out of their sockets" would be a good opening gambit.
The biggest deal in the poker calendar kicks off tomorrow, but only humans will be allowed to sit in on the final table of the World Series of Poker's main event. This Las Vegas knock-out tournament whittled 7,319 entrants down to nine back in July, and after a four-month wait to build media interest in the mainly no-name …
poker isn't just about playing each hand optimally, there's state in the form of the cash total in each player's pot. If any one player has significantly more money than the others, they can use that advantage to bully the others. If one player starts pulling ahead, there may then be an advantage in bidding aggressively on bad hands if your fellow losers do to cause the leader to back off and lose some pots. You can also gang up to kill the people who are furthest behind to simplify the game.
The other thing that can happen with multiple players is that you are more likely to have erratic behavior since the poor players haven't been winnowed out yet. I'm sure that causes hell for any assumptions about optimal play and rational behavior.
Yes you can count the cards , but when all is said and done, poker is about, seeing who has the biggest swinging pair between their legs! When all is said and done , a computer cant look into a persons eyes and tell if they really have a flush or not, and long may it continue :D
and i'll play the solid maths and statistics, see who comes out ahead!
Misrepresenting your hand is a part of the game, but only a part, being able to read your opponents like a book won't help you if you can't do the, fast, maths to back it up. But, if you can do the maths, you can get away with not being able to read your opponents physically. in-game behaviour is all you have to go on online anyway, and a computer can analyse this.
Also, there is a world of difference between playing online and playing around a real table, there is a lot more pressure on a real live player. You occasionally get a player that makes the transition successfully, but they never become the big names, as they don't have the character built around them.
Being a poker nut I have spent/wasted many hours trying to learn this game.
Good poker players have recently been found in research to be excellent strategists.
Computers and AI strategising neural nets seem to be missing at present. There would certainly need to be more effort on this part before a better poker bot can be built.
I strongly suspect that's why they peg their poker AI as only being at the "amateur" level. Amateurs may possess the basic know-how to play the game and maybe a few of the basic tactics needed to try to control the table flow, but there is still a lot of human factor in poker: especially multiplayer poker. There is the reading of tells, the concealment of your own tells, the idea of the occasional "stunt" play to affect psyches, and so on.
Poker is supposed to be about playing the person not the cards.
Sure, there are things to quantify what to do when a potential hand is showing and chip stack size, etc. but it comes down to tells.
Perhaps they can make a computer to pick up in tells, twitches, thermal imaging, the works.
"Poker is supposed to be about playing the person not the cards.
Sure, there are things to quantify what to do when a potential hand is showing and chip stack size, etc. but it comes down to tells."
Doesn't really account for why the best real-life players generally clean up in online poker too (if they enjoy playing it). I don't think it's a good idea to try and simplify poker down to 'it's only the psych' or 'it's only the math'. It's both, plus a ton of other things too. That's why it's so interesting, and why a lot of high-level players of other games move into poker. (well, that and the huge stacks of prize money; hard to win $9m playing Magic The Gathering...)
mind you, the BEST thing about the WSOP is the awesome commentary. I believe they are the Ragin' Cajuns...
“We're at mid-range amateur level. Nothing particularity spectacular,” says Bowling from the university of Alberta's gold plated underground monorail equipped poker department.
"Nothing to see here ... please don't trip over the piles of gold bars on your way out - we aren't very good at online poker honest"
Bet at those who fold a lot, don't bet so much at those who don't fold alot. After that the rest boils down to observational analysis and good strategy.
I think the scientists making these bots are asking the wrong questions: i.e. does this player fold often (more than 50% of the time)? do they normally bet if they have something/anything? simple questions can be very revealing about a players strategy. At the same time players have to vary their playing style per player so as to become unpredictable to others they play against. playing correctly against one player may not be correct against another. e.g. if a player does fold often and I have nothing, but theres another player behind me that almost never folds, betting would *usually* only get rid of 1 player (the tight one, rather than the loose one). Being out of position is a disadvantage, therefore "check" rather than "bet" would be best in most instances.
People change through time though. If they lose a big pot they often go on tilt meaning they stop folding and start pushing their chips into pots with hands they ordinarily wouldn't consider. Of course, you cannot win every hand in poker which is where strategy comes into play - overplaying too many hands ineveitably leads to loss.
I actually don't see what value having a poker bot would add to the world of poker though, other than it may provide casinos with a new better way to tap into player's bank balance. I'd be interested to hear of the advantages of a poker bot if anyone has any. The only time I've seen them put to use is for malign purposes.
Biting the hand that feeds IT © 1998–2020