master level bot created?
 

The University of Alberta’s Computer Poker Research Group has developed an artificially intelligent automaton known as “Vex Bot,” capable of playing poker at the master level, though as yet it can only apply its gambling genius to two-player games. Vex Bot has been used by researchers to test the frontiers of artificial intelligence – and as the basis for a commercial poker tutorial program, Poki’s Poker Academy -- but some fear it may become a blueprint for programmers with more sinister motives.

Re: master level bot created?
 

What does "master level" mean, and how do they know that this thing plays at it?

Re: master level bot created?
 

unlike other games where computers have proven to be equal or supieror(sp?) to humans, i.e. chess and backgammon, poker has a large factor of "human" element. computers will never be able to match the human element. they cant think, only process.

Re: master level bot created?
 

Head. Sand.

Re: master level bot created?
 

[ QUOTE ]
Head. Sand.

[/ QUOTE ]

oh, so cryptic, so vague. you make reference to a big bird who hides his head in the sand when scared. i fail to see any connection to the discussion.

Re: master level bot created?
 

[ QUOTE ]
unlike other games where computers have proven to be equal or supieror(sp?) to humans, i.e. chess and backgammon, poker has a large factor of "human" element. computers will never be able to match the human element. they cant think, only process.

[/ QUOTE ]

Yes, I was actually thinking about chess, where it makes more sense to talk about playing at master level, since it is actually well defined. There are master titles, and if you can play even with titled players, you could say that you play at master level. In poker, the meaning is unclear.

However, I don't agree that it wouldn't be possible to ever create a bot that plays (online) poker as well as any human. This is exactly what was said about chess computers until the late 80s. The computers lacked the necessary "intuition" and "ability to think" to play as well as a human. Later development has shown that there are ways to compensate for this.

Re: master level bot created?
 

Let me make this comparison between poker and chess and the use of computers.

if you take the worlds greatest chess player and have him play against an average joe chess player, he will win 10 times out of 10.

However, the greatest poker player can easily lose to an average poker player (especially online). Look at the world series of poker as an example. The best players in the world havent won in 4 years. Its random new guy after random new guy. And at the WSOP the great players have the added benefit of reading players. If you took the best 5,000 chess players and had them compete in a big tourney the best players would win each and every time.

there is clearly no comparison between chess "bots" and poker "bots."

Re: master level bot created?
 

i have no doubt in my mind that, at some point, there will be a computer vs human heads up match just like the chess one. there is also a strong posibility that the computer could win. the computer will "remember" all the plays the human made. the longer they play the better the computer can map out the humans style and then choose an action. the humans best shot is to end it fast before the sample size becomes large enough.

Re: master level bot created?
 

YSSCKY

Re: master level bot created?
 

[ QUOTE ]
[ QUOTE ]
Head. Sand.

[/ QUOTE ]

oh, so cryptic, so vague. you make reference to a big bird who hides his head in the sand when scared. i fail to see any connection to the discussion.

[/ QUOTE ]

Would a computer be able to assimilate those two words and understand what he meant? I suspect not. And that's the difference between poker and chess. In chess every possible outcome can be analysed if you have enough computing power. In poker there are a lot of unknowns and context is important.

Having said that I don't doubt that competent bots are possible or already exist and if undetectable would impact severely on internet poker. I don't think this would be because most players would be losing players (they are already), but because playing against a machine is just less satisfying.

Re: master level bot created?
 

[ QUOTE ]
Head. Sand.

[/ QUOTE ]

Useless. Response. Got any reason for saying that? Yes? care to enlighten us?

Re: master level bot created?
 

[ QUOTE ]
Let me make this comparison between poker and chess and the use of computers.

if you take the worlds greatest chess player and have him play against an average joe chess player, he will win 10 times out of 10.

However, the greatest poker player can easily lose to an average poker player (especially online). Look at the world series of poker as an example. The best players in the world havent won in 4 years. Its random new guy after random new guy. And at the WSOP the great players have the added benefit of reading players. If you took the best 5,000 chess players and had them compete in a big tourney the best players would win each and every time.

there is clearly no comparison between chess "bots" and poker "bots."

[/ QUOTE ]

I don't think this argument is convincing. First of all, chess isn't as exact a science as you imply. Good players get defeated by worse players all the time. But I agree that the variance is much smaller than in poker.

This however, does not mean that you cannot program a bot to play as well as a human. Of course the bot could still get its virtual ass kicked in any one session, but it would win in the long run against worse opponents. By your argument, you couldn't even talk about good poker players, since they don't win every time. What I mean is that I am not convinced that you cannot program a bot to play poker as well as the best human players.

Re: master level bot created?
 

[ QUOTE ]


Would a computer be able to assimilate those two words and understand what he meant? I suspect not. And that's the difference between poker and chess. In chess every possible outcome can be analysed if you have enough computing power. In poker there are a lot of unknowns and context is important.



[/ QUOTE ]

Yes, in chess, every possible variation could be analyzed, if you have enough computing power. But this is irrelevant to the discussion, since noone has that kind of computing power, and noone will in a forseeable future. This is not how chess engines defeat grandmasters. They use computer power, yes, but they also have to use heuristics to value positions. Engines do not play perfect chess, just very very good chess.

Re: master level bot created?
 

I think there are two different points here:

1. Can you develop a poker program that can beat humans?

2. If so would you get huge numbers of these bots playing online poker and ruining the game for us?

I don't think we need to worry about the first question too much, because poker sites do not and will not tolerate bots. If Party can already scan your hard drive when it wants, and check your running processes, you'd have to be pretty dedicated and tech savvy to stand a chance at avoiding detection. Some bots may get through, but not in large enough numbers to have any serious effect on the game.

Re: master level bot created?
 

[ QUOTE ]

I don't think this argument is convincing. First of all, chess isn't as exact a science as you imply. Good players get defeated by worse players all the time. But I agree that the variance is much smaller than in poker.

This however, does not mean that you cannot program a bot to play as well as a human. Of course the bot could still get its virtual ass kicked in any one session, but it would win in the long run against worse opponents. By your argument, you couldn't even talk about good poker players, since they don't win every time. What I mean is that I am not convinced that you cannot program a bot to play poker as well as the best human players.

[/ QUOTE ]

Maybe good players get beat by worse players, but great players would not lose to you or me. It just wouldnt happen. Could you imagine bobby fisher losing a chess match to me? It simply wouldnt happen in 100 games.

but if you or me played against the best poker players in the world, we would win a few times.

The chess board is very complex, but it there are limited number of moves and calculations. If you move to a certain spot, the computer can calculate your next most probable move. In poker, the game can change when the flop or turn or river hit. The computer won't know whats coming and would have a much harder time adjusting to the ever changing situations. not to mention a full table of players and having to adjust to each and every players differences. Chess is heads up, so its easier for a computer to adjust.

With poker you have players getting up and new players coming in. So not only does poker have more variables heads up, it has a greater level of difficulty when playing against a full ring.

There is no doubt that programs can be made to win at poker. I just don't think it will ever be good enough to dominate the game like deep blue was able to dominate chess.

Re: master level bot created?
 

[ QUOTE ]
unlike other games where computers have proven to be equal or supieror(sp?) to humans, i.e. chess and backgammon, poker has a large factor of "human" element. computers will never be able to match the human element. they cant think, only process.

[/ QUOTE ]
That's ironic because good poker players don't think very much. They make almost all decisions very quickly using a trained ability to recognize patterns. Neural net programs often do well at this type of application. I wonder if it would be possible to design one that read large numbers of hand histories played by an expert and learned to imitate his play.

From a game theory perspective there is such a thing as mathematically correct poker play that cannot be beaten in the long run. The mathematics is too complicated to work out exactly for complicated games like hold'em, but there is no reason that programs could not eventually be developed that would defeat any human opponent in the long run using game theory.

Of course beating expert players is not solely what it's all about. You would also want your computer poker player to analyze opponents past play, identify leaks, and modify its own play to exploit them. The computer's potential to remember and analyze thousands of hands will be a big plus here.

In summary, expert computer poker is hard and it isn't solved yet, but there is no special reason why it can't be done.

Re: master level bot created?
 

[ QUOTE ]
[ QUOTE ]
Head. Sand.

[/ QUOTE ]

Useless. Response. Got any reason for saying that? Yes? care to enlighten us?

[/ QUOTE ]

For Sciolist and mattw;

I think the poster was inferring that he preferred to be oblivious to the possible threat of poker bots... out of sight, out of mind.

I don't think there was any sinister or malicious intent in his post.

Ease up.

Re: master level bot created?
 

[ QUOTE ]
Head. Sand.

[/ QUOTE ]
See:

http://www.pbs.org/wnet/nature/critters/ostrich.html

Right at the bottom.

Re: master level bot created?
 

I wasn't trying to be an ass. I just get a kick out of comments like this:

[ QUOTE ]
computers will never be able to match the human element. they cant think, only process.

[/ QUOTE ]

For the record, I have no fear of bots at the moment. Not because I'm a good player (because I'm not) but simply because I doubt there are very many good bots out there. The reasons for this have been discussed over and over on these boards already. However, I think that believing computers will never be able to "think" is a bit naive.

CRF250X

Re: master level bot created?
 

FWIW, I'm pretty sure DN has played a significant amount of HU against this bot.

Also, I agree with posters who pointed out that winning money at low/mid level online poker doesn't take much complex thought. Beating 1/2 limit or 25NL for a reasonable amount could be achieved with a very small instruction set and no complex play at all.

Re: master level bot created?
 

[ QUOTE ]
unlike other games where computers have proven to be equal or supieror(sp?) to humans, i.e. chess and backgammon, poker has a large factor of "human" element.

[/ QUOTE ]

Actually there isn't a human element in online poker. At all. All your decisions are based on statistical analysis, pattern recognition and crude profiling.

Poki combined with a PT database of a few million hands would do very well in most online games. There just isn't such a bot around though.

There hasn't been all that much research effort made on poker bots. It's not exactly fair to compare todays poker bots to todays chess bots. I guesstimate that by 1965 there had been more research done on chess bots than there has been on poker bots in 2005.

Both Doyle and Cloutier claim that their superior memory is what sets them apart as NL players. But neither hold even a candle to the memory sun of computers.

I think that in the coming decade we'll see a fair amount of bots which can both hold their own against superstars and extract near the maximum against weak opposition.

Re: master level bot created?
 

[ QUOTE ]
you'd have to be pretty dedicated and tech savvy to stand a chance at avoiding detection

[/ QUOTE ]

Dedicated and tech savvy? As in someone who would spend countless hours developing a bot?

Shessh.

Tuco.

Re: master level bot created?
 

The issue isnt whether someone can create a bot that can win, or whether they can create a bot that can win and be undetectable. I think that will eventually happen. The problem would be keeping it undetectable. And once it is found, you are banned and so is the money in your account. Then you have to work hard at trying to become undetectable again, only to be spotted eventually when the sites catch up.

Even if someone was able to stay one step ahead of the game, that is just one person cheating. To think that it can be mass produced and threaten online poker is a joke. Once it was sold publicly, the tech people at these sites would hire a tech guy to detect it or prevent its use. Then you have people holding hundred dollar bots that don't work.

People who cheat via IM or telephone are a far greater risk to us than the eventual bot invasion.

Re: master level bot created?
 

[ QUOTE ]
[ QUOTE ]
Head. Sand.

[/ QUOTE ]

Useless. Response. Got any reason for saying that? Yes? care to enlighten us?

[/ QUOTE ]

actually its a very good response.

I've been saying this for a loooong time, but all the people who think its not possible for computers to beat humans because poker is a game of incomplete information, are completely clueless.

Whats really sad is that when it does happen and online poker is ruined, I wont be able to really enjoy spamming my "I told you so"'s everywhere, because I'll be poor and looking for a job [img]/images/graemlins/frown.gif[/img]

Re: master level bot created?
 

Vex on, vex off. Soon you will reach master level, young grasshopper.

Re: master level bot created?
 

I heard they also landed a man on the moon -- wow!

(yawn -- when was this squib pulled from, 1981??)

Re: master level bot created?
 

I agree 100%. The breakthrough in poker bots will come with a breakthrough in "learning algorithms", such as neural nets as you mentioned. Until then, bots will not be able to play at the level of top human players.

[ QUOTE ]
That's ironic because good poker players don't think very much. They make almost all decisions very quickly using a trained ability to recognize patterns. Neural net programs often do well at this type of application. I wonder if it would be possible to design one that read large numbers of hand histories played by an expert and learned to imitate his play.

From a game theory perspective there is such a thing as mathematically correct poker play that cannot be beaten in the long run. The mathematics is too complicated to work out exactly for complicated games like hold'em, but there is no reason that programs could not eventually be developed that would defeat any human opponent in the long run using game theory.

Of course beating expert players is not solely what it's all about. You would also want your computer poker player to analyze opponents past play, identify leaks, and modify its own play to exploit them. The computer's potential to remember and analyze thousands of hands will be a big plus here.

In summary, expert computer poker is hard and it isn't solved yet, but there is no special reason why it can't be done.

[/ QUOTE ]

Re: master level bot created?
 

[ QUOTE ]
And once it is found, you are banned and so is the money in your account

[/ QUOTE ]

Thats a little like saying nobody will bother trying to embezzle money because if they are caught they have to give the money back. It's not much of a deterent that money is not yours gets confiscated.

[ QUOTE ]
Once it was sold publicly, the tech people at these sites would hire a tech guy to detect it or prevent its use. Then you have people holding hundred dollar bots that don't work.


[/ QUOTE ]

Any bot worth anything wouldn't be sold. Noone would pay what it would be worth.

Tuco.

Re: master level bot created?
 

I thought this forum was chock full of Computer Science majors who actually learnt something, but this is obviously not the case. (Or rather, they have all learnt to skip ignorant ramblings like these.)

First of all, let's separate two things
1. A poker algorithm that can't be exploitable (i.e. beaten in the long run) (and remember that the long run is a *long* time in poker, and not a short tourney like the WSOP ME.)
2. A poker algorithm that beats particular opponents in a particular game at a rate reasonably close to the maximum possible level.

1 is difficult, but it's not *that* difficult, in partcular if one adds "beaten for more than the rake". It is almost definitely easier to write a poker bot which will not allow any human to beat a full ring of said bots for more than the rake (assuming fairly low limits, i.e. a significant rake) than it is to write a chess program that beats Kasparov.

Why, then, has the second happened, but not the first? For the very simple reason that hundreds (thousands) of times as much effort and money has been spent on chess compared to poker up until today.

2. is significantly harder, as it is basically 1. plus the learnings and adjustments. And it's debatale if it will ever be +EV to do this properly. It's not too valuable against world class opponents anyway and it is hopefully difficult to make anhy money on it against poor opponents.

This all hinges on one thing: that there is a game theoretical approach to poker that is not exploitable. This is not completely obvious, since the relationship between algorithm strengths could be non-transitive even for optimal algorithms.

I have seen several more or less convincing arguments for why there are optimal algorithms that are not exploitable, but nothing even remotely convincing suggesting otherwise. But I'd love to hear about any such argument!

Re: master level bot created?
 

[ QUOTE ]
Any bot worth anything wouldn't be sold. Noone would pay what it would be worth.

[/ QUOTE ]

Why? Because you say so?

It is perfectly possible that no one will pay anything for a poker bot because they think along the lines of: "If I buy this program, others will probably as well. And then there will be many users, we will all get caught and I will gain nothing from doing so."

Meanwhile, IRL, I point you to WinHoldem or whatever it was (is?) called.

Are you suggesting that if Winholdem would have been great instead of bad, no one would have bought it?

Re: master level bot sucks ass
 

I have poker academy pro. Both Sparbot and Vexbot suck. I forget which one but when the board had a broadway straight showing with no flush possible, I could get the bot to fold with a check raise on the river. I wrote the writers of the software and they said they found a glitch which would cause the bot to do that.

Stu

Re: master level bot created?
 

[ QUOTE ]
2. A poker algorithm that beats particular opponents in a particular game at a rate reasonably close to the maximum possible level.
...
2. is significantly harder, as it is basically 1. plus the learnings and adjustments. And it's debatale if it will ever be +EV to do this properly. It's not too valuable against world class opponents anyway and it is hopefully difficult to make anhy money on it against poor opponents.

This all hinges on one thing: that there is a game theoretical approach to poker that is not exploitable. This is not completely obvious, since the relationship between algorithm strengths could be non-transitive even for optimal algorithms.

[/ QUOTE ]

A bot capable of winning 2BB/100 in the Party 15/30 game doesn't have to be non-exploitable to do so! A bot that plays SSH strategy and does player profiling based on hands played, coupled with a datamining facility would still be exploitable. An aware opponent would play counterstrategies. But those counterstrategies give the aware human a miniscule advantage, while the bot still has a small to humongous advantage over non-aware humans. So if it loses pennies to our aware human it doesn't matter, because it's raking in the dough from fishes.

My game has plenty of leaks, and is so exploitable it's silly. But I still make money, and lots of it. Because of the simple fact that there are few players skilled enough to play counterstrategies to TAGs, and of those a very small subset even bother. Scrutinizing my play for leaks is far down on the EV hierarchy. (For opponents, obviously not for me!) They can and should spend that effort where it matters more, extracting more from weak players. I highly doubt anyone decides to focus on an 18/12/3 opponent to see if they could possibly extract 1/8th of a bet extra from him every 100 hands while the sea is still full of fish!

Re: master level bot created?
 

[ QUOTE ]
Head. Sand.

[/ QUOTE ]

Head. Arse.

Re: master level bot created?
 

[ QUOTE ]
Head. Arse.

[/ QUOTE ]

Ah, I see you're in deep.

Re: master level bot created?
 

[ QUOTE ]
[ QUOTE ]
Any bot worth anything wouldn't be sold. Noone would pay what it would be worth.

[/ QUOTE ]

Why? Because you say so?

[/ QUOTE ]

A bot worth anything would probably be worth millions. Maybe two or three licenses could be sold before the games dry up. Note that with a good enough bot, you'd run it for one seat at every table at the mid to high limits. 1BB/100h at 60h/hr at even 5-10 is $1000 a week, enough to pay for the hardware to run the bot. Every subsequent week is pure profit.

[ QUOTE ]
Meanwhile, IRL, I point you to WinHoldem

[/ QUOTE ]

That bot was worth about $0, from what I heard about it.

Re: master level bot created?
 

[ QUOTE ]
A bot that plays SSH strategy and does player profiling based on hands played, coupled with a datamining facility would still be exploitable. An aware opponent would play counterstrategies.

[/ QUOTE ]

You'd have to know it was a bot, and if you suspected that, then the poker site probably would, too. At least, it'd be in your interest to advise them of the possibility.


[ QUOTE ]
there are few players skilled enough to play counterstrategies to TAGs

[/ QUOTE ]

Sure, most of us have big leaks. Some of my leaks are miscounting outs, not betting for value, etc -- decisions that a bot wouldn't make. The biggest technical hurdle for bot-makers is ajusting to opponents, especially opponents that will themselves adjust to their opponents. I'm a crap player at the micros, and even then I'll play tighter against a fellow TAG. It's trivial to code enough game theory into a bot to thwart other TAGs (at least compared to writing the rest of the bot).

Re: master level bot created?
 

First, I have to ask why is a University making a poker bot? Don't they have some real research to do.

A winning bot will be certanly made. All it has to do is have 16% VPIP bet with the nuts, call all raises and it will beat the Party .50/1 game.

There will always be competition between bot makers and huge corporations like Party, UB and others. We all know who the winner of this competition will be. Internet poker casinos have a lot of money, they make more from rake in one day than a poker bot maker could in a lifetime of succesful poker bot making. Plus, party cotrols the source code, they can change it and screw up the bot maker without investing too much money.
I'm surprised that there is still a poker bot thread every week. There is nothing to be afraid of, you think Party is going to ruin the integrity of its bussines by alowing poker bots to take its rake and more importantly scare the fish away? There is just no way that will happen. Internet casinos hold all the cards they are not stupid like the people that belive in poker bot take over, santa clause, the easter bunny, and calling an all in to catch their two outer. As long as there is fish there will be internet poker.

Re: master level bot created?
 

[ QUOTE ]
Why? Because you say so?

[/ QUOTE ]

Because economics says so.

[ QUOTE ]
Are you suggesting that if Winholdem would have been great instead of bad, no one would have bought it?

[/ QUOTE ]

Are you suggesting that Winholdem can be used as an analagy to a program that could literally win millions of dollars for someone? If you are, im done arguing with you.

Tuco.

Re: the real story
 

Many don't see a threat
But skeptics – and there are many – argue the complexities of the game and the changing strategies ensure that creation of a program that can “read” opponents’ cards using screen scanning techniques and respond in real time is years away at best. They point to the handful of commercial products that purport to give online players significant advantage, which they roundly deride as woefully inadequate, as proof today's bots are no match for humans.

Rao and his fellow believers have a ready answer: A bot capable of playing against the best humans already exists.

The University of Alberta’s Computer Poker Research Group has developed an artificially intelligent automaton known as “Vex Bot,” capable of playing poker at the master level, though as yet it can only apply its gambling genius to two-player games. Vex Bot has been used by researchers to test the frontiers of artificial intelligence – and as the basis for a commercial poker tutorial program, Poki’s Poker Academy -- but some fear it may become a blueprint for programmers with more sinister motives.

says the scanning of opponets cards is only a few years away though that is kind of....

anyway this was a http://www.msnbc.msn.com/id/6002298/ article. thought it might be kind of interesting!

Re: master level bot created?
 

[ QUOTE ]
A bot capable of winning 2BB/100 in the Party 15/30 game doesn't have to be non-exploitable to do so!

[/ QUOTE ]

You are right. My point was:

It is *easier* to write a poker program that beats world class opponents in a non-exploitable way than it is to write a bot that beats crappy opponents on a maximum or close to maximum level.

A lot of people in this thread have argued that the opposite is true, or a t least that it's exceedingly hard to write a world class level poker program.