[xxhong]

No AI can beat good human players in a deep stack no limit holdem game. All the draw pokers, limit holdem and omaha are much simpler in that sense. No limit poker is about observation, psychological and strategical pattern recognition and assigning the right constraints to all the possibilities. All those things are very very difficult to program.

hrothgar, on 2011-May-26, 11:13, said:

The strategy space for Hold'Em seems remarkably limited compared to any stud variant or even draw

There aren't nearly as many stages to the game.
You don't have nearly as much information that you need to process on the fly.

I don't know how you interprete the word "complicated", but from a game theoretic perspective you're dead wrong.

FWIW, I did my master's thesis on Poker years and years ago.
I was identifying equilibria for 5 card draw.

Typical types of results:

If you get dealt trips, its better to draw one card than two.
Its more valuable to be able to conceal whether your drawing with

1. Two pair
2. Trips
3. Trying for a four flush

than the additional chance of improving your hand...

(Describing this sort of thing is pretty easy. Proving it... shudder)

[jjbrr]

billw55, on 2011-May-26, 09:03, said:

2. People get to pretend they are playing for more money than they really are. For a $50 buy in, or even less, you might get 5000 "dollars" in tournament chips. Now people can say "I bet three thousand" and feel like a cowboy or a high roller. When really the most they can lose is their buy in. They're paying just a little for this entertainment, this rush of fake high stakes. And they have a non-trivial chance to win something on top of it, per above.

loldonkaments

[Phil]

People actually spellcheck forum posts?

[jjbrr]

xxhong, on 2011-May-26, 11:41, said:

No AI can beat good human players in a deep stack no limit holdem game. All the draw pokers, limit holdem and omaha are much simpler in that sense. No limit poker is about observation, psychological and strategical pattern recognition and assigning the right constraints to all the possibilities. All those things are very very difficult to program.

NLHE has a GTO strategy.

edit: your point is valid; it's hard to program a winning NLHE bot, but you make it sound like it's impossible, which it is not.

we just don't have an AI *YET* that can consistently beat humans. we will eventually.

[hrothgar]

xxhong, on 2011-May-26, 11:41, said:

No AI can beat good human players in a deep stack no limit holdem game. All the draw pokers, limit holdem and omaha are much simpler in that sense. No limit poker is about observation, psychological and strategical pattern recognition and assigning the right constraints to all the possibilities. All those things are very very difficult to program.

My understanding was that the Poker Group up at the University of Alberta had a poker bot that could beat (essentially) anyone in the world at two player limit. The group was moving on to two player no limit and expected that they'd be in a similar position within a couple years.

I think that you are significantly misrepresenting the difficulty of this problem.

[jjbrr]

hrothgar, on 2011-May-26, 11:57, said:

My understanding was that the Poker Group up at the University of Alberta had a poker bot that could beat (essentially) anyone in the world at two player limit. The group was moving on to two player no limit and expected that they'd be in a similar position within a couple years.

I think that you are significantly misrepresenting the difficulty of this problem.

This bot is called Polaris fwiw. There are membership sites available where you can play against it, if you so choose.

[Aberlour10]

There were 2 groups of polish bridge players in the second half of the 20th century...

1) Hundreds of thousand (mostly well-educated )people played rubber bridge as one of the main elements of their social life, such a private bridge parties were a mass apparances in times (communism) with not a wide range of possibilities to spend a free time..

2) 5-8 thousends members of the polish NBO who were interesting in playing competetive bridge.

And nowdays? The first group smelts like the arctic ice. The life in the last 20 years is much faster, the competition in job much much harder, nobody has a lot of free time and there are 1001 more attractive possibilities to spend it than in 70s or 80's.

The second group remains relatively small but is much more active now, especially the youth. In the sector college and university teams is Poland leading in Europe, I suppose.

[xxhong]

Well, it is an over optimistic expectation IMO. In 2008, it played 6 games, won 3, tied 1 and lost 2 against good human players head to head in limit holdem. However, it has a lot of problems facing multiple opps and facing opps who often change strategies. In No limit game, this problem will become even more severe, because very strong players frequently change strategies in their game. Also, the challenge of computing resource in no limit game is much larger than in limit games.

hrothgar, on 2011-May-26, 11:57, said:

My understanding was that the Poker Group up at the University of Alberta had a poker bot that could beat (essentially) anyone in the world at two player limit. The group was moving on to two player no limit and expected that they'd be in a similar position within a couple years.

I think that you are significantly misrepresenting the difficulty of this problem.

[Foxx]

Vampyr, on 2011-May-26, 10:00, said:

Remember that celebrity thing they did on Sky? The bidding was really basic, the card-play was appalling, but the players were clearly having fun and there was expert commentary so that the viewer could learn a little. I think that this format could be successful, but it needs to be on a less obscure channel than Sky Sports 2.

How about the Bathurst team broadcast one of their Bermuda Bowl practice matches on Versus.

[JLOGIC]

hrothgar, on 2011-May-26, 11:57, said:

My understanding was that the Poker Group up at the University of Alberta had a poker bot that could beat (essentially) anyone in the world at two player limit. The group was moving on to two player no limit and expected that they'd be in a similar position within a couple years.

I think that you are significantly misrepresenting the difficulty of this problem.

Not a chance in hell for no limit. Limit is far easier for an AI for obvious reasons. There is no way that Darse Billings and the Alberta guys claimed they would be in a similar position in a couple of years unless they were talking about an abnormally small stack size.

xxhongs posts have been right on imo.

(FWIW I was involved extensively in a HU limit program that is better than Polaris that is now in casinos as a rake free skill based slot machine. You can imagine that once we got that done, all the casinos and manufacturers wanted were a no limit program that could do the same thing. The same programmer who is an AI/neural net genius could not even come remotely close to something that could beat me, while attempting it for years, even for just a 50 BB stack).

[JLOGIC]

Foxx, on 2011-May-26, 16:43, said:

How about the Bathurst team broadcast one of their Bermuda Bowl practice matches on Versus.

Practice matches?

[hrothgar]

JLOGIC, on 2011-May-27, 05:06, said:

Not a chance in hell for no limit. Limit is far easier for an AI for obvious reasons. There is no way that Darse Billings and the Alberta guys claimed they would be in a similar position in a couple of years unless they were talking about an abnormally small stack size.

I tracked down the quote in question:

Billings claimed the following

“Our no-limit program is probably at a very strong amateur level, not at a professional level,” he says. He believes that in a couple of years, in no-limit games, his team will be able to approach its success of 2008 when Polaris, the collective name for the group's bots, came out ahead competing against six of the world's best heads-up (two-player) limit players. The score was close: three wins to Polaris, two to the pros, and one draw.

http://www.theregist...h_up_for_grabs/

Earlier in the article, Billings stated that Alberta had programs that could beat the world's best.
I had confused these programs with Polaris which plays much worse and was being used as the basis for the comparison.

[JLOGIC]

Sorry for triple post, but one other thing about computer programs in poker; money talks.

Lets say you can develop a program that can beat humans and cannot make money commercializing on it, you can make a TON of money by putting it on a poker site to play for real money. This is against the terms of service, but of course people would do it for the amount of money at stake.

There have been many occasions of bots being found and banned online. At first, they were all in heads up limit hold em. They were beating stakes as high as 50/100 for tens or possibly hundreds of thousands of dollars.

There were also many occasions that this happened in hyper turbo and turbo sit n gos (1 table tournaments where you start with a small amount of big blinds, like 10, and the blinds go up very fast). Obviously this is a perfect format for bots because it is push/fold and all simple math based which computers should crush. Again, these programs made a lot of money.

There have been no occasions that I know of of bots being banned from the heads up no limit games. If this happened for any significant stakes, it would be well known, because the sites refund players who were cheated when this is found.

This, imo, is obviously because there are no bots that can beat significant stakes of HUNL at this time. There are other possible reasons like the sites covering it up or something, but stuff like this gets out, and combined with all the other evidence (no commercial product that is at all good at HUNL, like the UofA team for LHE, etc), is enough for me to believe that this does not exist, and I'm sure it will not exist for a long time. It really is too hard for our current methods/processing power.

[Zelandakh]

The point I was trying to make by mentioning poker is not that the game is similar to bridge but rather to point out what has been done with a game that was largely humdrum to turn it into the most popular card game in the world. And done in only a few years. Bridge has a larger player base than HU had at that time - if someone was able to market bridge in an equally innovative way that happened to work on TV this would represent a massive boost for the game. The alternative is probably Adam's idea of trying to reach youngsters through schools. Unfortunately there are some difficulties with that method which probably mean it is a way for the game to stay healthy in some areas but not a way to have it thrive worldwide.

[hrothgar]

Zelandakh, on 2011-May-27, 11:08, said:

The point I was trying to make by mentioning poker is not that the game is similar to bridge but rather to point out what has been done with a game that was largely humdrum to turn it into the most popular card game in the world. And done in only a few years. Bridge has a larger player base than HU had at that time - if someone was able to market bridge in an equally innovative way that happened to work on TV this would represent a massive boost for the game.

Comment 1:

I would argue that bridge's large player base is a weakness, not a strength.
Simply put, its very difficult to get young players to socialize with the geriatric set.

Comment 2

The game of "Poker" needed to make very significant changes before it took off.
It don't think that you can do the same to bridge without destroying the soul of the game...

It seems appropriate to quote the late, great George Carlin:

Quote

Baseball needs a little speeding up. You know how you speed up baseball? Everybody gets one swing, that's right. One swing, ***** you, you're out, sit down, next, let's go, come one, sit down, come on, let's go. Here's another thing that would make baseball a lot faster: If the pitcher hits the batter with the ball, the batter's out. You hit 27 guys, you got yourself a perfect game my friend. You get two really good accurate pitchers out there and you could be out of that ballpark in 15 minutes. You could be home watching football on TV and see some serious injuries. One more thing for baseball, out in the outfield I would have a series of randomly placed landmines. "There's Marshall, settling under that ball." (EXPLOSION sound effect) "Holy *****!"

[nige1]

hrothgar, on 2011-May-26, 11:57, said:

My understanding was that the Poker Group up at the University of Alberta had a poker bot that could beat (essentially) anyone in the world at two player limit. The group was moving on to two player no limit and expected that they'd be in a similar position within a couple years. I think that you are significantly misrepresenting the difficulty of this problem.

I guess that multi-player games are difficult for programs because human players can act in partnership (informally or by arrangement) and human experts probably already know who the weakest players are so can easily take their money, Hence It is reasonable for programmers to concentrate on head-to-head games. Until I read the enlightening and surprising comments of JLOGIC and xxhong, I would have thought that the problem with no-limit games would be more randomness.not more difficulty. Anyway, I agree with Hrothgar that programmers will crack heads-up no-limit holdem within a couple of years at most. (Disclaimer: I can't afford to play poker so this is idle speculation)

[nige1]

hrothgar, on 2011-May-27, 11:20, said:

The game of "Poker" needed to make very significant changes before it took off.
It don't think that you can do the same to bridge without destroying the soul of the game..

IMO radical rule-simplication would make bridge more attractive to play and to watch, while keeping its essential nature and enhancing its soul. Unfortunately ....

• As Phil implies, decisive groups like administrators, directors, professional players, and old-stagers have a considerable investment in the status quo, By and large, they resist any change however trivial. (For example, a WBFLC member hinted that some laws might be renumbered for the new 2007 edition. There were many protests by directors in on-line fora).
• Most ordinary players take no active interest in such matters. Some of the few that summon up any interest seem to take pleasure in destructive criticism (see the reaction of most posters to Carl Hudecek's suggestions. It is unlikely that any one person has a complete answer but we shouldn't reflexly reject every detail of every proposal.

[hrothgar]

nige1, on 2011-May-27, 12:36, said:

IMO radical rule-simplication would make bridge more attractive to play and to watch, while keeping its essential nature and enhancing its soul. Unfortunately ....

Just to be clear: My original post should not be construed to support Nigel's crusade in any way, shape, or form.
“If You Quote What I Said Then You're Lying"

[jjbrr]

nige1, on 2011-May-27, 11:43, said:

I guess that multi-payer games are difficult for programs because human players can act in partnership (informally or by arrangement) and human experts probably already know who the weakest players are so can easily take their money, Hence It is reasonable for programmers to concentrate on head-to-head games. Until I read the enlightening and surprising comments of JLOGIC and xxhong, I would have thought that the problem with no-limit games would be more randomness.not more difficulty. Anyway, I agree with Hrothgar that programmers will crack no-limit holdem within a couple of years at most. (Disclaimer: I can't afford to play poker so this is idle speculation)

Human behavior and randomness have nothing to do with how the bots are programmed.

The goal of programming the bots is to have the bot play as close to GTO as possible, otherwise it would be possible to find a way to exploit the bot's strategy. It absolutely does not depend on what anyone else is doing in the game. It does not assume the other player plays the same strategy. Indeed, it doesn't depend at all on what other strategy anyone else is playing. This is why you could publish your strategy or announce it before each action and it would not have any impact on the ability of people to beat you. Randomness has nothing to do with it in the long run.


The problem with programming a bot to play GTO is a matter of computing power. I think there are something like ~3.7*10^15 possible combinations of hands+betting action in HULHE (I didn't have 20 hours of college math, so if I'm way wrong, I apologize. 4 betting rounds, 16 ways a round of betting can go, 52!/(45!*3!*2) possible board + hole card combinations). It takes a long time for a computer to figure out how to play each one of these combinations correctly.

It's not hard to see how that number jumps even higher when you add more players or allow bets in various sizes.

So while it may be possible that we have a winning NLHE bot soon enough, don't hold your breath, and certainly don't expect more than heads up. It probably won't play the equilibrium strategy either, but close enough.

[nige1]

jjbrr, on 2011-May-27, 13:55, said:

Human behavior and randomness have nothing to do with how the bots are programmed.

What is GTO? Donald Michie's match-box computer showed that optimum strategy depends on opponent's skill. In the multi-player example, against a tyro, a theoretically unsound strategy can win faster. Also, I think the computer could be good at detecting patterns in its opponent's play and exploiting play-history databases.

jjbrr, on 2011-May-27, 13:55, said:

The goal of programming the bots is to have the bot play as close to GTO as possible, otherwise it would be possible to find a way to exploit the bot's strategy. It absolutely does not depend on what anyone else is doing in the game. It does not assume the other player plays the same strategy. Indeed, it doesn't depend at all on what other strategy anyone else is playing. This is why you could publish your strategy or announce it before each action and it would not have any impact on the ability of people to beat you. Randomness has nothing to do with it in the long run.

I'm not sure about that. One goal of the programmer would be unpredictability. If for example, the computer bluffed on certain kinds of hand, in certain types of context, the programmer would not want to publish that strategy.

jjbrr, on 2011-May-27, 13:55, said:

The problem with programming a bot to play GTO is a matter of computing power. I think there are something like ~3.7*10^15 possible combinations of hands+betting action in HULHE (I didn't have 20 hours of college math, so if I'm way wrong, I apologize. 4 betting rounds, 16 ways a round of betting can go, 52!/(45!*3!*2) possible board + hole card combinations). It takes a long time for a computer to figure out how to play each one of these combinations correctly. It's not hard to see how that number jumps even higher when you add more players or allow bets in various sizes.

I've never played Holdem but I think jjbrii exagerates its complexity. I doubt that a programmer would even try to cater, individually, for each possibility and I think a computer-array would have more than enough power to handle generic cases. The difficulty is more in heuristics.

jjbrr, on 2011-May-27, 13:55, said:

So while it may be possible that we have a winning NLHE bot soon enough, don't hold your breath, and certainly don't expect more than heads up. It probably won't play the equilibrium strategy either, but close enough.

I won't hold my breath but I bet jjbrr a shilling that there will be a winning NLHU program within two years.