100% yes. Poker will have to shift to be more social for it to survive online eventually.

Meh. For the record if you said no, then you have no clue wtf you're talkign about. If you said no for any reason other than that computers simply will never be fast enough/have enough memory, then you really really have no clue wtf you're talking about.

Poker is a game. It is a game with a solution. The solution can be found using math. Very good approximations can be found using brute force methods. The only reason this hasn't been done already is because of some combination of computers not being fast enough and the right people not being particularly interested in the problem yet.

yes and it's not close

I'm actually quite surprised this poll is a dead heat

Here's some pseudocode that finds a solution to every situation in n BB deep NLHE with integer bet sizes:

For every possible river board with every possible range for both yourself and your opponent with every possible pot:stack ratio, try every possible line. Run minimax on that crap to find the best possible line with each hand given your range getting to the river, your opponent's range getting to the river, the board, and pot:stack ratio.

Do the same crap on the turn except if the line reaches a river, just insert the equity of the perfect river line.

Do the same crap on the flop except if the line reaches a turn, just insert the equity of the perfect turn line.

Do the same crap for preflop except if the line reaches a flop, just insert the equity of the perfect flop line.

In some thread somewhere, I actually calculated how many combinations you'd have to go through to do this. It was some massive number and no computer would ever do it that way, but IIRC it was fewer than the number of possible chess board for 100 BB HUNL.

Simple anwser ^


Long anwser: we allready posses the technology to create such engines, but unlike a human a engine is pure calculations that works out of complex mathmatical algorithems,. There have allready been made a engine that beatet a long array of "pros" in limit games because its basically a math game - the Chess engines Fritz 11 or Rybka is allready so complex that it is able to beat top level chess players like magnus carlsen and anand, even old engines like deep blue beatet top level chess players.
however... the NLHE games are much more understanding your opponent, his though procces and emotions and adapting to his changing playing styles and here mathmatic algorithems fails.

So again, for Limit games - yes
for other games - no

Edit: many of the guys in here that voted proberly dont know much about how engines like this work or doesnt have a basic understanding of what makes a computer think and whats differs from humans. Again yes in the future LOOOONNNG out we may be able to, but we are talking hundreds of years. Simply getting more computaional power such as better CPU's and better math algorithems will not do the trick. Only some kind of fully functional robot with artificial intelligense, emotions, complex thinking and memory of other players as in of emotion and not just database, and yet still after all this you will proberly only get a bad NLHE player.
again... dont think so

Do you know that programmers who can barely understand rules of backgammon let alone play it could create bots which beats the best players in the world ?
The same goes for simplified poker games. The same goes for NLHE. It's just matter of time.
People who answered "no" in this poll has no clue what so over about AI techniques an computer science in general. Try to find one person who is programmer or graduated univ with computer science major who says no as answer to this question.

Please give me an example of a machine that are 'literally thinking'? That's a big statement. Every computer I've seen is a slave to it's instructions and is theoretically 100% predictable.

Big blue is not considered primitive. It uses some theoretically simple mathematics (as Noah mentioned minimiax is a core algorithm in game theory in AI) but the heuristics it uses are very smart.

Poker is a mathmatical game. Computers are flawless at maths. Noah outlined a brute force solution that theoretically would work, but you would need a lot of intelligence to apply it with heuristics to make it feasable.

A bot will come along eventually that is the best in the world. It's just a matter of time. And it's not sinister. It will just play poker very well. That's why I think poker is going to have to shift away from its current state into something more social (think VOIP/webcam) for it to survive and for people to trust it.

A more interesting question is how is the industry going to change when these bots are known to exist?

what bots? Chineese DoN rings? its only because they are exploitable by mathmatical engines running on algorithems. With current state i dont think any bot would be able to play deepstack NLHE

There have allready been experiments on creating AI poker player, and as i mentioned in my previous post they failed misserably. As it is now bots can only exploit things that lean towards being mathmatical games such as playing shallow stack games with minimum buyin, or playing DoN, or playing Limit games

They haven't failed miserably, University of Alberta are leading the field and have written some great papers on the subject.

It's complex, and they are coming. Probably sooner rather than later I would imagine, progress in these sorts of things seem to gain a lot of momentum.

For all the doubters, the wikipedia computer chess article is great:

http://en.wikipedia.org/wiki/Computer_chess

This is the core of your argument (original spelling) and it is wrong. A persons emotions, style "vocabulary", ranges etc are manifested in the context of poker in their hand histories. These are effectively, large databases. This is data. Computers do good stuff with the databases. I do have a comp sci degree.

I doubt you read my post.

I am a software engineer who is not actively working anymore (thx poker), and I am 99% sure in the 3 years it took me to improve my game to a certain level, I think i could have written a bot who probably plays HU NL 100bb deep better than I do now. I never played poker fulltime. Take it fwiw.

Same thing came to my mind,reading the op

This computer may eventually mean the end of online poker (no joke, and not quantum computing).

I believe the biggest problem in poker AI at the moment is processing power to crunch equity calcs for your entire range of possible hands vs your opponents entire range of possible hands in real time.

I dont think we will see this from our eyes... maybe in 300 years, but not before.
Theres so much more in poker than just strategy and math. It involves understanding of things such as ego and psychology, empathy, which are human attributes and influence our play.

How can people say things like this with a straight face. Do you realize how much has changed in the last 300 years?
In 300 years from now this world will probably be unrecognizable.

300 years?!?

In the future anything is possible bro.

Live poker a bot would win no doubt. They have some pretty advanced remote lie detecting/biology monitoring hardware/software. They could just write a program to observe how the player reacts when he has a good hand or making a move. Just my 2 cents

http://www.youtube.com/watch?v=mg8_cKxJZJY


They will be beating craps too

http://www.youtube.com/watch?v=zbkizy-Y3qw

I think regressionary analysis will best even the best players. Once a computer starts gathering information, it can compute not only probabilities, but change tendencies. If you try to mix up your play, the computer knows it's coming. There are thousands of variables that computers could pick up on that we wouldn't even begin to think influenced our play. The only way to really beat a computer is to randomize your play, and humans are notoriously bad at faking randomness.

This thread is seriously ******ed. A "bot" is as good as its programming. Given enough time, resources, motivation, input, etc, it will win in the long-run. And any attribute that is missing, or any leaks it has, will quickly be spotted and more code will be written and so on, thereby fixing the gap and making it even harder to beat.

People who think this won't happen, are literally analogous to fish that are not aware of the water they live their existence in.