I'd like to see this on Wall Street
Sounds like a great way to make investment decisions.
Artificially intelligent software can comfortably outmatch human poker pros and amateurs in one-on-one matches, that much is known. Now, for the first time, an AI bot has been built that can beat humans professionals on six-player no-limit Hold’em tables no less, and has been described in an academic paper published on …
Stockmarket decisions are more about knowledge than game theory, which company is hiring, which firing, which opening up new offices/depots which closing, which is about to lose their key employee and which is gaining him. So yes I am sure Facebook has a bot capable of making good investment decisions but its noting to do with poker bots.
The people who own this code aren't going to deploy it when they could earn $1000/hr (or more, since most online poker contents don't have the best players in the world at the table)
Even if they were too pure to do this, someone else will get a bot that does the same. Heck, if you had a bot that could only win $10/hr that's still $80,000 a year for the cost of electricity (or paying AWS to run it) And there are a lot of online poker sites, so you can have more than one. I'd set a dozen of them running and retire, thank you very much!
Having worked in the online gambling industry I can say the Poker companies are not doing this themselves.
Its FAR more profitable and less riskly just to provide the service and take a cut of buyin/stakes/winnings. or provide a winning pot and take the business.
Just like when people say a roulette is riged, they miss the point that all the house needs is a small odds in its favor (as is the design of the normal game) and them to play enough games.
33s to make a move is unacceptably slow for real work cases. However I can't see the researchers not having a copy of the code to "try" at home.
"Heck, if you had a bot that could only win $10/hr that's still $80,000 a year"
Generally poker sites don't allow bots or player aids. And will stop you playing after a certain amount of time.
If you're going to grind, then it's often easier to play for lower stakes and eat the fish, rather than play against people with equal skill.
Getting to a level where you can make $10-20 per hour is actually fairly easy, assuming you can do the bet sizing in your head. It's mainly down to how many games you can play simultaneously.
Playing multiple hands simultaneously sounds like WORK to me, and making only $10 to $20 an hour at it sounds like a big pay cut over what I make now.
But if I have a machine making it for me, I don't care how little it earns so long as it pays for its running cost. Then its just a passive investment like a house I'm renting or a owning a dividend stock.
Not sure how a poker site would be able to tell it is a bot, well other than the 24x7x365 thing. Have it quit for 8 hours after every 10-12 and it'll seem like a real but addicted player who is grinding out the money. Didn't think about the angle of playing multiple hands, even if it is only playing half the possible hours if it plays 4 hands at once that's $160K a year it can earn on one site. Maybe you could have it randomly skip a hand like a real person would have to do to go the bathroom, etc. to make it seem more 'real'.
Once again we are shown that the computer is only quicker, not necessarily better. I sure that, if I had the time (and the means) to play anything a trillion times, I would become the undisputed world champion of it.
No, I think you'd be locked up in an asylum after going crazy due to boredom before you could get a fraction of the way there. Being able to do the same thing over and over again without complaint is a big advantage of a lack of consciousness.
If a computer plays something a trillion times, it can remember every detail of every game and then run the analysis on best strategy.
I suspect your tongue is in your cheek, but if not, please can I borrow your memory?
Now, what did I come here for?.....
They have just told everyone it is a soluble problem, and indicated the path.
So yes, other people will develop it and won't announce...
Question - how much rent can bots extract from humans before the humans stop playing?
Biz Op - develop the software and use it to monitor play to seek out and ban the bots. Online poker rooms have a serious incentive to NOT let these bots go live, and the money to pay.
If only I wrote code...
Right now somebody probably is working on an Open Source poker bot. And "acheivable with readily-available materials" is pretty much a given, seeing as they'll be stuck with little more than readily-available materials.
Of course, if the bots get too good then nobody will want to play against them .....
it played 10,000 hands against five elite poker pro humans
I've no idea how long it takes human players, elite or otherwise, to play a hand of poker. The Cincinnati Kid seemed to take a long time, but that was fiction. Two minutes per hand, which seems pretty quick to an ignoramus, comes to about 42 eight-hour days. Don't these elite players have anything better to do?
>Don't these elite players have anything better to do?
Umm their job is to literally play poker for hours on end. Article doesn't say it but I assume some of that dosh that FB prints probably made it worth their while (else why give away your time, your knowledge and how you play for free). I agree being a pro poker player sounds like the worst kind of grind to me (ok minus manual labor to be fair) but to each their own.
Also worth noting if the pros were playing with their own money I doubt they would have stayed for 10,000 hands. Hustling isn't just about the play itself. Also I wouldn't be surprised if the number of hands these people have played in their careers is measured time wise in actual years and not days or hours.
Time taken to play a hand depends very much on the level of play and how many players are in the hand.
If you assume six players sitting, at the professional level you would probably not see any more than three players seeing the Flop (the first three common cards) the rest fold, whereas at the lower skill levels far more of the table would be in it, possibly through to the River (the last card). The hand time will also quicken as players drop out.
Better players are invariably tight/aggressive. They play less hands but when they are in it they bet more aggressively. Passive play and loose play, in the long run, gets found out.
Having said that, big professional tournaments with hundreds of 'runners' can last up to a week before you get into a money placing.
As someone else here said, it is their life....
Nobody has touched on the human and psychological angle. The top players place great store on their ability to spot "tells" and to avoid giving them. That is why they often wear hoodies,and dark glasses, and practise their poker faces. Maybe against a computer one of their important skills is worthless. Of course online poker is different.
It would not be impossible for a computer to have tells of its own. Or to pick up on other players' tells, if there is any spurious information being given away in the amounts bet or the timing (for instance, always betting an amount ending in £5 with good but out-of-suit cards; or deliberately pausing before betting with a potential winning hand to avoid scaring the other players away before the pot has built up).
Biting the hand that feeds IT © 1998–2019