The Two Groups of Bridge

[nige1]

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]

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 ....

1. 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).
2. 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]

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]

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]

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.

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.

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.

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.

[jjbrr]

No, you have some misconceptions.

 

A game optimal strategy is not necessarily the best strategy or the most lucrative strategy; in fact in this case it's not even close. It's simply the strategy that cannot be beaten by any other strategy. It's well documented that many heads up limit bots win against bad players at a much lower rate than humans against bad players, because it doesn't actively exploit weaknesses or leaks. It just plays mathematically perfect poker (or close to it. we probably haven't discovered the equilibrium strategy yet). Against an opponent playing the same optimal strategy, you would expect this bot to break even in the long run, and against those good players who were beating the bad players for a lot more money, it would be the favorite.

 

Unpredictability is wrong. The optimal strategy is certainly predictable, but it is a mixed strategy, ie for any given situation, the bot will fold x%, bet (raise) y%, and check (call) z%.

 

You are correct that the goal should be to lump some of the hands together.

[jjbrr]

also, if we're assuming infinite stack sizes, i'll bet my entire net worth, and lets extend it to 5 years.

[NickRW]

...we probably haven't discovered the equilibrium strategy yet

 

Its a while since I read all the papers that Darse Billings + chums have produced, but iirc they had a pretty darned close go at it for heads up limit - that's why their latest bots are so difficult beat over a long match.

[jjbrr]

yeah i think they're close if they haven't gotten it yet. but anyway in practice nearly optimal is good enough.

[nige1]

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,

No, you have some misconceptions. A game optimal strategy is not necessarily the best strategy or the most lucrative strategy; in fact in this case it's not even close. It's simply the strategy that cannot be beaten by any other strategy. It's well documented that many heads up limit bots win against bad players at a much lower rate than humans against bad players, because it doesn't actively exploit weaknesses or leaks. It just plays mathematically perfect poker (or close to it. we probably haven't discovered the equilibrium strategy yet). Against an opponent playing the same optimal strategy, you would expect this bot to break even in the long run, and against those good players who were beating the bad players for a lot more money, it would be the favorite.

Assume the program is playing in game with a good human player and a poor human player. The good player adopts a technically unsound but effective and fast-winning strategy when up against the poor player. Thus he takes more money from the poor player than the program does, because it is handicapped by playing what jjbrr calls "optimal strategy".

 

With the poor player eliminated, the good player switches to optimal strategy, The human is now favourite, because he has most of the money. IMO a better strategy for the program would be to take into account all available information about its opponents.

[NickRW]

Assume the program is playing in game with a good human player and some of poor players. The good player adopts a technically unsound but effective and fast-winning strategy when up against against a poor player so he takes more money from the poor players than the program does, because it is handicapped by playing what you call "optimal strategy".

 

With the poor players eliminated, the good player switches to optimal strategy, The human is now is strong favourite, because he has most of the money. IMO a better strategy for the program would be to take into account all available information about its opponents.

 

Yeah, but how does a machine (and a computer is a machine) figure all that out. It is a difficult problem as soon as you even introduce a third player to the table. Is player X, say, playing some sort of generally mixed strategy or is this player playing tight against player Y, but loose against player Z. Obviously, with enough observation, a computer can figure it out, but by that time X may have cleaned up most of the chips from the fish.

[JLOGIC]

People are really failing to grasp how much harder no limit is than limit for a computer. You are changing the possible variations by an enormous amount just because of bet sizing, not to mention that you are opening up the possibility of your bot making huge mistakes. It is simple enough in limit to figure out how many value hands vs bluffs you have on the river, and bluff the appropriate amount of time accordingly. It is also simple enough to do the reverse with calling down. Try doing that in no limit, and account for small bets, large bets, pot sized bets, etc. It is not possible that in 2 years time there will be a HUNL bot that is beating humans just based on where people are at right now in trying to solve that problem (something that would make them a LOT of money if they did). It really is a very long way off despite how far along the limit games are (not just LHE, but bots are very good at limit draw games etc).

[nige1]

I don't play Holdem, so please correct me if I'm wrong.

• Presumably, a good program will first try to make Bayesean inferences. What kinds of hand can an opponent have to explain his betting pattern (allowing for concealment and bluffing possibilities), in the light of what it can glean about his previous habits and betting patterns.
• Nobody (except jjbrr) has an "infinite stack", What is called "no-limit" is really severely limited. Your maximum commitment is the chips in front of you -- or the chips in front of your opponent if they are fewer. This makes betting strategy more complex than ordinary limit-poker but I'm sure it wouldn't take a competent games-theorist long to crack it.
• I suppose your objective also makes a difference. Your strategy may be different if you aim to wipe out your opponent rather than simply end in profit.
• The task seems tedious but not overwhelmingly hard.

[JLOGIC]

Nige1 I'm sorry but you really don't know what you're talking about.

 

Yes, stack size is important, as I've already said if it was a 10 big blind stack size then it is easily solvable. A standard buy in in an online site for a no limit game is 100 big blinds, live it is more, but let's go with 100 big blinds. Of course, you might lose your 100 big blinds and then rebuy for 100 big blinds and get deeper than that.

 

The difference in 100 big blind no limit and limit is this: In limit, you can only bet or raise 1 big blind preflop and on the flop and 2 big blinds on the turn or river (and you can never bet any other amount, your options are to check or bet that amount, and then fold, call, or raise that amount, etc). In No limit at any time you would be able to bet 1 big blind, 2 big blinds, 3 big blinds, 3.5 big blinds, 4 big blinds, 5 big blinds, etc, up to all in for whatever your stack is (lets say 100 big blinds, minus whatever you've already put in the pot). Hopefully you can see that this increases the complexity of the game exponentially.

 

In a typical limit game, if your opponent bets the river, there might be 8 bets (16 big blinds) in the pot. If you are faced with a bet, it will be for 1 bet. This leads to all computers pretty much showing down any pair or better, and on most boards ace high. Why? Because pairs are hard to make, and you are getting 8:1 on your money so you only have to be right 11 % of the time, and people sometimes bluff (remember your opponent, despite knowing he is getting called very often, is getting a great price on his bluff). I am not trying to trivialize heads up limit, but that is what decisions come down to. Computers are about as good as top limit players now (a little worse than the very best but they are still very good), because they are able to have good frequencies for bluffing and good frequencies for calling down since it is largely a math question and how you construct your ranges.

 

In a typical no limit game, you will be facing a river bet of 3/4ths pot or so (lets say there is 40 big blinds in the pot, a typical bet will be 30 big blinds). As such it is much less of a showdown game, and much more about interpreting betting patterns and putting them on a specific range of hands. But it's more difficult than that, because sometimes you might get overbet and see a 60 big blind bet. What does that mean? Or you might see a small bet of 15 big blinds. What does that mean? If the draws miss, and your opponent knows that you will call down a bet very light, and he still bets, what does that mean? Etc. It is just a much more complex game for computers.

 

This makes betting strategy a bit more complex than ordinary limit-poker but I'm sure it wouldn't take a competent games-theorist long to crack it.

 

I don't understand how you could say this if you don't even play poker, or understand how different NL is from limit! Seriously, how long do you think people have been working on this? Do you think that all of these "competent game-theorists" who cracked heads up limit just randomly decided to try heads up limit rather than NL, despite NL being far more popular and thus more profitable if you can create a bot for it? Heads up limit is basically a fringe game, even before the poker boom when limit was popular, nobody played it heads up (except Andy Beal!). Even before the bot problem, it did not enjoy 1/100th of the popularity of heads up NL online. The reason these bots are widespread in HULHE and very good, but no good ones exist for HUNLHE, is because it was possible to do for limit, and it is not possible right now for NL. I guarantee you more effort has been put into making HUNL bots than limit by all of these competent game-theorists for the past 10 years, for the obvious reasons that these people would love to make a lot of money.

[Bbradley62]

I nominate this for "Most Successfully Hijacked Thread".

[NickRW]

I don't play Holdem, so please correct me if I'm wrong.

 

Nigel, this is a quote from a fairly recent paper on the subject of poker bots:

 

"Running CFR on abstract Poker games for sufficiently long often takes on the order of days to weeks."

 

CFR = Counter Factual Regret Minimilization = (roughly) a technique for approximating an optimum strategy

 

abstract = they are not talking about 'real' Poker, they are talking about cut down versions of the game with such things as one or more of a) limit instead of no-limit, b) heads up instead of multi player, c) reduced deck size, d) reduced number of betting rounds or e) having a cap on the number of raises allowed per round.

 

When they talk of "days to weeks" they are not talking about doing this 9 to 5 on a single computer - they are talking about possibly having several computers working on the problem in parallel 24/7.

 

When you said "The task seems tedious but not overwhelmingly hard", tedious it may be, "not overwhelmingly hard" is a wee bit of an overbid on your part.

 

Nick

[tolvyrj]

This sort of "scientific" approaches and topics r one of the reasons why bridge players r considered least say peculiar human beings.

Is it just me or does anyone else think this conversation is waist of time?

[wickedbid1]

Dear Phil, et al.

 

This was an interesting thread until it got hijacked by the poker botphiles.

 

When comparing poker to the problems of "golden age", present day, and where future bridge will go, it would do well to remember a bit of bridge history. In the "golden age" bridge was a very popular money game, and there were a lot of clubs where bridge for money was played. The centre of the game has shifted to tournie bridge for "serious" players, mainly because of the bridge scandals of the 60's & 70"s. The money aspect of the game has seriously dwindled, ecept for those who run clubs, are professional players, teachers or tournie directors, etc.

 

Poker is in a "gold rush" fad phase which has come and gone for bridge. A few scandals may turn poker into a more intellectual, elitist sort of mental sport, such as chess fans & bridge fans fancy themselves to be, but whether that happens to poker and is perceived as a good thing for it or a bad thing for it, is for poker fans to decide.

 

Our task, should we choose to accept it, is to figure out what the future of bridge ought to look like, and come up with some practical suggestions on efficent ways to get there. Or, if u prefer, to offer some interesting idea-probes and speculate upon where these suggestions might take us if we followed through on them. And then follow through on the best ones.

 

As far as bridge and television go --- I'll take vuegraph, thank you very much. Somewhere there is a survey that says a majority of Americans who have both cable and computers would give up TV before their computers. This is the future. Bridge should stick to new media.

 

Regarding bringing new people into the game, ACBL bridge teachers are a good tool, I believe, although I may be somewhat unqualified to make an objective judgment on the matter, having been an ACBL bridge teacher for a number of years. I taught dozens of people to play before becoming certified, too... it is not as difficult as some people think, although a teacher mentality is required, of course.

 

Many young people who are taught the game do leave it to focus on jobs & family, but they will return, I am sure, when other priorities diminish. On-line bridge, being so easy & cheap to join and drop, helps, I think, tremendously to keep the bridge-fires burning in the busy years of 30-50.

[Vampyr]

 

Poker is in a "gold rush" fad phase which has come and gone for bridge. A few scandals may turn poker into a more intellectual, elitist sort of mental sport, such as chess fans & bridge fans fancy themselves to be, but whether that happens to poker and is perceived as a good thing for it or a bad thing for it, is for poker fans to decide.

 

Poker is not an intellectual game. A person could sit down, be told the rules in 30 seconds (certainly for Hold'em, at least), and win a big pot on their first hand against experts. It is not possible to achieve any kind of success against bridge or chess experts without a fair amount of study or practice -- plus in bridge, building the rudiments of a partnership. This is why bridge is really up against it in our short-attention-span age.

 

Also, young people today are kind of strange. They spend hours on Facebook, and Twitter, and instant messaging... I think they may find an activity that involves seeing actual other people face-to-face rather intimidating. Of course there is online, but you need to first need to learn to play. I am certain that it is fairly straightforward to find a teacher who teaches online, but you have to be pretty motivated in the first place to find out where to look.

 

Regarding bringing new people into the game, ACBL bridge teachers are a good tool, I believe, although I may be somewhat unqualified to make an objective judgment on the matter, having been an ACBL bridge teacher for a number of years. I taught dozens of people to play before becoming certified, too... it is not as difficult as some people think, although a teacher mentality is required, of course.

 

You might find this hard to believe, but there are even teachers outside America! Anyway, as a bridge teacher, what do you do to get new people into your lessons? How do you reach out to people who haven't played before? Who have played for years around the kitchen table? It would be good if you and other teachers on the forum shared their strategies. If there are some really good ones they could be more widely disseminated.

[jogs]

I nominate this for "Most Successfully Hijacked Thread".

 

The OP did include psyching as an advance strategy. I disagree. Bluffing is the essence of poker. Psyching has no place in bridge. In poker every player on the table is an opponent. In bridge the player across is in theory your partner. Also bridge is a full disclosure game, poker is not. It is much easier to field a psyche by partner than one by the opponents. Two major reasons for this. You know your partner much better than you know the opponents. Every time the auction is 'impossible' because you hold too many cards AND both opponents are having BIT, you know your partner psyched.

[wickedbid1]

Poker is not an intellectual game. A person could sit down, be told the rules in 30 seconds (certainly for Hold'em, at least), and win a big pot on their first hand against experts. It is not possible to achieve any kind of success against bridge or chess experts without a fair amount of study or practice -- plus in bridge, building the rudiments of a partnership. This is why bridge is really up against it in our short-attention-span age.

 

Also, young people today are kind of strange. They spend hours on Facebook, and Twitter, and instant messaging... I think they may find an activity that involves seeing actual other people face-to-face rather intimidating. Of course there is online, but you need to first need to learn to play. I am certain that it is fairly straightforward to find a teacher who teaches online, but you have to be pretty motivated in the first place to find out where to look.

 

 

 

You might find this hard to believe, but there are even teachers outside America! Anyway, as a bridge teacher, what do you do to get new people into your lessons? How do you reach out to people who haven't played before? Who have played for years around the kitchen table? It would be good if you and other teachers on the forum shared their strategies. If there are some really good ones they could be more widely disseminated.

 

 

Poker attracts the attention of a lot of very smart people, so I have no problem should those people decide to call it an intellectual game. Even if it can be played at a very basic level by those of us who have not delved into it deeply. The point I was trying to make is that if poker ever evolved into a game not played for money, it would be played for reasons of aesthetic/intellectual satisfaction, as bridge is for many of us.

 

Although, I do believe that the social aspect of the game is the most important part for the vast majority. To test this, one would ask the survey question: would you rather spend an evening playing bridge with very poor players whom u like a lot, or very good players u hate? How about a lifetime?

 

Since u asked a direct question, Vampyr, I will try to provide a direct answer... where are the bridge students lurking?

 

In my own case, I have taught mainly at a big university with an active bridge club. I became the "resident teacher" because i'd played & taught informally during my undergraduate days at another institution. All I did was put up signs around campus and encourage the club members to tell their friends i was starting a new block of lessons. A group of 15-25 students showed up for each lesson block at the start of the spring & fall semesters. Getting them in the door isn't much of a problem. Keeping them can be tricky. Most are showing up for a fun evening or afternoon, not more SCHOOL (ick).

 

I charged a very small fee, as the club provided the space and boards, and the ACBL provided the books for larnin' (Audrey Grant's Club & Diamond Series). I condensed & simplified even this basic material -- into eight or nine lessons, with lots of play practice. The level of play in the "main" roon varied greatly, so it was no problem to introduce the newbies fairly quickly -- when the regular club and its director r friendly, the newbies will lose their nerves before they lose their nerve.

 

There were a fair number of older people at the club too (it was not quite restricted to students, professors & alumni). Some seniors and middle aged people did show up for lessons from time to time too, yet the majority were undergraduate & graduate students from all countries, with all sorts of majors, and so the older students had to work to fit in socially with the younger, not the other way round. This, of course, is a lot easier for all concerned, as the older people have been there, done that.

[hrothgar]

Psyching has no place in bridge.

 

Strangely, the laws disagree with you here...

Perhaps you might be happier playing some other game?

 

Still, quotes like this are a sad commentary of the lasting effects of Don Oakie

[mgoetze]

Is it just me or does anyone else think this conversation is waist of time?

 

Time is obese and his waist is quite substantial.

[wickedbid1]

"Waist naught, want naught!"

[rduran1216]

I think the point is more that there are "social bridge players" versus "tournament bridge players." The former group are mostly out to have a good time; they are certainly trying to do well when they play but are not interested in devoting a huge amount of time to reading bridge literature, discussing hands or methods, etc. They tend to be uninterested in traveling to tournaments (unless it's part of a vacation or bridge cruise maybe). They are sometimes hostile to players perceived as "sharks" (they don't like to play against pros, or against unusual methods, or against people perceived as "too competitive").

 

The "tournament players" view bridge as a very competitive activity. They will often spend a lot of time and/or money in order to improve their game. They much prefer to play in tournaments than club games (the bigger the tournament the better). They often don't like playing against people who are too "chatty" or "not serious" and want the director to strictly enforce the laws; it tends to upset them when opponents can't explain their methods or agreements.

 

Tournament players tend to be better than social bridge players because they devote a lot more effort to improving, but this isn't 100% true because players have different levels of experience and talent.

 

Obviously there is space between these two extremes. But some of the problems in bridge come up when these groups rub against each other the wrong way.

 

 

Long and short, 95% of bridge pros and genuinely good players are total assholes.

 

I'm lucky enough to have been groomed a bit at the club level around people who were good enough players in their day, and play competitively, but only to a certain point. I feel safe in saying the vibe in Long Beach is much different than other clubs I've been to. The reason for this is that the core group of good "A" players who are consistent winners, all socialize together and don't mind maintaining the social aspect.

 

I personally have grown to loathe outsiders who rely on rule enforcement and try to pinch "C" players for every matchpoint they can get. To me its despicable. Arrogance is a despicable attribute of most good players. Worse is slimy traits, like watching where old ladies play their cards from, peeking into hands, and not turning deaf ears to people who talk too loudly.

 

We recently had a player come in, who many of you know, who I won't mention, playing with a client. He opened 2NT. His partner, who he'd been badgering the entire night, held xxx xx xxx QJxxx. Thanks to his club fit, and well placed cards, he made 3NT. When his partner asked if she should bid on, he passive aggresively opined, "I always respond to 2NT when I have a good 5-card suit." This poor client was terrified the entire time. It just highlights whats wrong with the undescribed group.

 

Bridge is a game. The only people who are THAT good, are rarely the ones who are nitpicky losers. This forum is a good place, with good and great and really great players who are kind enough to answers questions and be good resources for people like myself who are attempting to climb the ladder.