Pragmatism, Dogmatism, and Theory
Jul. 15th, 2007 08:39 pm![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
peterbirksWhen it comes to poker, I admit that I am a bit of a Theorist. This isn't the best way to be. A Pragmatist is the best way to be. However, down the route of pragmatism lies danger. If you are not too careful, it becomes dogmatism, and that can cost you a lot more money than sticking to the theory.
Chen and Ankenman are the greatest proponents of "theory"; so much so that they list their justification on page one, "Common Misconceptions", Number One.
Let's jump to another area of theory -- the hands with which you should raise all-in when your stack is, say, an M of 7 and you are on the button. Andy Ward has produced a nice little table of this, which is mathematically sound. In other words, if you follow this table, it doesn't matter what your opponent does.
However, in the three or so years since I first saw that table (since mathematically expounded in the Chubukov table) Andy has change the hands with which he plays that way. Indeed, even when he first revealed them, players such as Gryko observed that they would choose different hands, although they accepted the principle.
Why? Because the players defending hadn't read the tables. On top of that, even if they had, they didn't know if Andy had. Andy discovered that there was no need to play at the point where "it doesn't matter what your opponent does". You could play even looser than that, because defending players did not play perfectly. And, to a man (or woman) the way that they did not play perfectly was that they folded too often. Pragmatism led to a better result than theory.
Now, Chen and Ankenman would observe that this advantage could only be temporary. To which the obvious reply is "Yes, we know. The trick is to adapt just before your opponent adapts."
Now, pragmatism is all very well, but the important thing about it is that you have to understand the theory first. If you don't, then you will not realize that you are playing in an exploitable way, which only happens to work because your opponents haven't yet woken up on how to exploit it. This weakness is, along with a bankroll that cannot withstand variance, one of the major reasons for people who move up in stakes failing. They play a certain way. They win. They move up in stakes. They lose. They cannot see how this can be anything other than bad luck. They are wrong.
Back in the mid 1990s, dogmatism was rife. The reasons for this aren't hard to see. For a start, theory was (as we can see now) in its infancy. What mathematical stuff there was being written (e.g. in the game theory field) was not being written by gamblers. So the concept of game theory was that both opponents played perfectly (and the concept of multi-player 'unsolvable' games wasn't even touched). In a land without theory, you just went with what worked. You didn't really care if this was 'perfect' play or 'exploitable' play, because, in practice, this didn't matter at the time.
This led to certain areas of dogmatism. For example, always check it down when another player is all-in, because eliminating that player "moves you up the ladder". The extent to which this is complete bollocks (often you want the number of players to remain constant while you build up your stack, for example) is now obvious (except, perhaps, to a few of the real old die-hards) but at the time, any bet into a dry side-pot was seen as heresy. Various other "standard" plays of the time have since been shown to be mathematically dodgy. But the established players swore blind that they were right.
The early "how-to" poker books often fell into the same trap. They told you how to beat the games as they were at the time -- which was fine -- but they also implied that this was the right way to play, which was where pragmatism fell into dogmatism. It was only when the new Internet "geeks" came along, playing sometimes in a way which had the old-timer non-Internet casino poker players just staring in disbelief, that the dogma was revealed for what it was.
But theory can have its downfalls too. Aaron Brown, in that excellent book "The Poker Face of Wall Street", is virtually the proclaimer of negative EV plays that work because your opponents do not play perfectly. In this sense it is a dangerous book for novices (and is one reason why I recommend it wholeheartedly) but his attack on the "math-geeks" is most appositely expressed when he talks about the Prisoner's Dilemma.
Let's take not the standard two people, but four people. We have two game theorists, experts in the mathematics of poker and all game-related issues, whom we shall call Mr A and Mr C, and two professional criminal partners, who we shall call Mr Big and Mr Little.
We combine these into their various pairings, AB, AC, AL, CB, CL, BL.
I'll assume that you all know the prisoner's dilemma situation. If you give up the other guy and he says nothing, then you get one year and he gets 10 years. If you both say nothing, you both go free. If you give each other up, you both do nine years.
Now, Mr A and Mr C say to themselves, "Fuck, the other guy is bound to give me up, so if I give him up, I get 9 years instead of 10, so I have to give him up".
When partnered with mathematicians Mr A or Mr C, Mr Big (who knows nothing of game theory but much of loyalty) says "Who are these guys? I don't know them! Hell, maybe they'll stay quiet and I can get off! I'll give them up. Mr Little says the same.
But when Mr Big and Mr Little are both put in the box, they say "Ahh, he's a pro. He won't grass." And they both stay silent, and both go free.
As Aaron Brown points out, he can think of better things in life than to be stuck in a cell with a guy for nine years while that guy congratulates himself on what a good game theorist he is.
You often come across this situation at poker tables in multi-wayers. A player might make the correct play assuming that both of his opponents will play perfectly (to which the reply, as Brown might put it, is "what the hell are you doing in a game where you assume that your opponents will play perfectly?") but it's the wrong play given one of the player's known foibles/weaknesses (perhaps he folds too much, perhaps he calls too much, perhaps he's too passive, perhaps he's too aggressive). When you are the third player in this kind of scenario, it's incredibly annoying.
So, I try to start with the theory and adapt it as and when I see fit. In poker, this changes dramatically from site to site, from level to level, and on the time of day. In particular, late at night, players are far more likely to call, whereas in the afternoon, they are more likely to fold. These are equally exploitable. To call one game "good" and the other game "bad" on the gounds of looseness is wrong. The only game that is "bad" is where your opponents make the correct decisions. "Loose" or "tight" has nothing to do with it. But it is here that dogmatism comes into play. "Loose, action" games are seen as "good" and tight, no-action games are seen as bad. And the reason this is stated is because it is, in the main, true. You tend to make more money in the looser action games. But beware of dogmatism. The reason for this is nothing to do with the looseness, it's to do with the negative EV decisions that the opponents are making. If players fold inappropriately, those are negative EV decisions too, but the games show up as "% Seeing flop 14%, Avge pot 4x the big blind" and the pros avoid those games like the plague. In fact, you can move in, steal blinds for a couple of circuits (and perhaps actually make a hand) and move on. There's lots of poker sites you can move around doing that. It's not like it's a single casino.
Chen and Ankenman are the greatest proponents of "theory"; so much so that they list their justification on page one, "Common Misconceptions", Number One.
We can gain an edge over our opponents by knowing their strategies and exploiting them. But this edge can only be temporary".
Let's jump to another area of theory -- the hands with which you should raise all-in when your stack is, say, an M of 7 and you are on the button. Andy Ward has produced a nice little table of this, which is mathematically sound. In other words, if you follow this table, it doesn't matter what your opponent does.
However, in the three or so years since I first saw that table (since mathematically expounded in the Chubukov table) Andy has change the hands with which he plays that way. Indeed, even when he first revealed them, players such as Gryko observed that they would choose different hands, although they accepted the principle.
Why? Because the players defending hadn't read the tables. On top of that, even if they had, they didn't know if Andy had. Andy discovered that there was no need to play at the point where "it doesn't matter what your opponent does". You could play even looser than that, because defending players did not play perfectly. And, to a man (or woman) the way that they did not play perfectly was that they folded too often. Pragmatism led to a better result than theory.
Now, Chen and Ankenman would observe that this advantage could only be temporary. To which the obvious reply is "Yes, we know. The trick is to adapt just before your opponent adapts."
Now, pragmatism is all very well, but the important thing about it is that you have to understand the theory first. If you don't, then you will not realize that you are playing in an exploitable way, which only happens to work because your opponents haven't yet woken up on how to exploit it. This weakness is, along with a bankroll that cannot withstand variance, one of the major reasons for people who move up in stakes failing. They play a certain way. They win. They move up in stakes. They lose. They cannot see how this can be anything other than bad luck. They are wrong.
Back in the mid 1990s, dogmatism was rife. The reasons for this aren't hard to see. For a start, theory was (as we can see now) in its infancy. What mathematical stuff there was being written (e.g. in the game theory field) was not being written by gamblers. So the concept of game theory was that both opponents played perfectly (and the concept of multi-player 'unsolvable' games wasn't even touched). In a land without theory, you just went with what worked. You didn't really care if this was 'perfect' play or 'exploitable' play, because, in practice, this didn't matter at the time.
This led to certain areas of dogmatism. For example, always check it down when another player is all-in, because eliminating that player "moves you up the ladder". The extent to which this is complete bollocks (often you want the number of players to remain constant while you build up your stack, for example) is now obvious (except, perhaps, to a few of the real old die-hards) but at the time, any bet into a dry side-pot was seen as heresy. Various other "standard" plays of the time have since been shown to be mathematically dodgy. But the established players swore blind that they were right.
The early "how-to" poker books often fell into the same trap. They told you how to beat the games as they were at the time -- which was fine -- but they also implied that this was the right way to play, which was where pragmatism fell into dogmatism. It was only when the new Internet "geeks" came along, playing sometimes in a way which had the old-timer non-Internet casino poker players just staring in disbelief, that the dogma was revealed for what it was.
But theory can have its downfalls too. Aaron Brown, in that excellent book "The Poker Face of Wall Street", is virtually the proclaimer of negative EV plays that work because your opponents do not play perfectly. In this sense it is a dangerous book for novices (and is one reason why I recommend it wholeheartedly) but his attack on the "math-geeks" is most appositely expressed when he talks about the Prisoner's Dilemma.
Let's take not the standard two people, but four people. We have two game theorists, experts in the mathematics of poker and all game-related issues, whom we shall call Mr A and Mr C, and two professional criminal partners, who we shall call Mr Big and Mr Little.
We combine these into their various pairings, AB, AC, AL, CB, CL, BL.
I'll assume that you all know the prisoner's dilemma situation. If you give up the other guy and he says nothing, then you get one year and he gets 10 years. If you both say nothing, you both go free. If you give each other up, you both do nine years.
Now, Mr A and Mr C say to themselves, "Fuck, the other guy is bound to give me up, so if I give him up, I get 9 years instead of 10, so I have to give him up".
When partnered with mathematicians Mr A or Mr C, Mr Big (who knows nothing of game theory but much of loyalty) says "Who are these guys? I don't know them! Hell, maybe they'll stay quiet and I can get off! I'll give them up. Mr Little says the same.
But when Mr Big and Mr Little are both put in the box, they say "Ahh, he's a pro. He won't grass." And they both stay silent, and both go free.
As Aaron Brown points out, he can think of better things in life than to be stuck in a cell with a guy for nine years while that guy congratulates himself on what a good game theorist he is.
You often come across this situation at poker tables in multi-wayers. A player might make the correct play assuming that both of his opponents will play perfectly (to which the reply, as Brown might put it, is "what the hell are you doing in a game where you assume that your opponents will play perfectly?") but it's the wrong play given one of the player's known foibles/weaknesses (perhaps he folds too much, perhaps he calls too much, perhaps he's too passive, perhaps he's too aggressive). When you are the third player in this kind of scenario, it's incredibly annoying.
So, I try to start with the theory and adapt it as and when I see fit. In poker, this changes dramatically from site to site, from level to level, and on the time of day. In particular, late at night, players are far more likely to call, whereas in the afternoon, they are more likely to fold. These are equally exploitable. To call one game "good" and the other game "bad" on the gounds of looseness is wrong. The only game that is "bad" is where your opponents make the correct decisions. "Loose" or "tight" has nothing to do with it. But it is here that dogmatism comes into play. "Loose, action" games are seen as "good" and tight, no-action games are seen as bad. And the reason this is stated is because it is, in the main, true. You tend to make more money in the looser action games. But beware of dogmatism. The reason for this is nothing to do with the looseness, it's to do with the negative EV decisions that the opponents are making. If players fold inappropriately, those are negative EV decisions too, but the games show up as "% Seeing flop 14%, Avge pot 4x the big blind" and the pros avoid those games like the plague. In fact, you can move in, steal blinds for a couple of circuits (and perhaps actually make a hand) and move on. There's lots of poker sites you can move around doing that. It's not like it's a single casino.
![[identity profile]](https://www.dreamwidth.org/img/silk/identity/openid.png){kind=small}
Pooh!
Date: 2007-07-16 09:43 pm (UTC)Unless poker is entirely distinct from any other odds-based human activity (and let's face it, that includes rather a lot, up to and including sex, death, and anything else that Mr Bergman would be interested in filming), dogmatism is an obvious deviation from theory -- not from pragmatism.
Dogmatism is basically the concretisation of an otherwise fluid field of enquiry. Pragmatism is basically a choice between one or more fluid fields of enquiry, all, some, or none of which might be applicable to the concrete subject at hand.
Pragmatism is why things work "hand by hand," or if you prefer, "site by site," or whatever. Pragmatism doesn't give a fuck about theory, as long as there's one that works, and maybe a better one round the corner.
Dogmatism is why we have politicians. They don't have to be as disgusting as a Stalin; they might be simply as objectionable as a Blair, or as pathetic as a Cameron. Dogmatism uses theory as a slogan.
Game theory, on the other hand, I could take or leave. Von Neumann came up with two eye-catching ideas in his life, and I prefer to make money off the other one. I no more believe in the general applicability of game theory than I do in the possibility of anyone solving the three-body problem in Newtonian gravitation.
Poker sites these days look awfully like the stock market to me: too many people with too many dickhead theories (up to and including "My granny would have done that"), and consequently too much non-linear perturbation to make any sense out of a particular, ahem, dogma. Just go with the EV, your gut feel, and your horoscope for today.
And if that doesn't work, have a nice hot cup of cocoa and move on to the next table.
Re: Pooh!
Date: 2007-07-18 12:50 pm (UTC)As you say, dogma comes from theory. What happened in "the old days" was that people learnt how to play with "what worked". This then became the theory (kind of like a general theory of relativity vs a special theory of relativity) and the old-timers wrote books stating that this method worked all the time. This was the dogma. It didn't work all the time. It only worked in special situations (loose-passive opponents, timid in tournaments, etc). So, yes, the dogma derived from theory. But the (incorrect) theory was derived from pragmatism. Where the flaw lay was that the theory derived from the pragmatism considered things to be "givens" which were not.
Pete
Re: Pooh!
Date: 2007-07-20 07:39 pm (UTC)You're pushing it a bit with the general/special theory of relativity, although I'll let that pass. The basic reason you are pushing it is that it's the wrong domain. Either one makes specific predictions, based upon (largely) static dynamics, and either one can be measured. The difference between the two -- and why bother mentioning them unless there is a difference between the two -- is that they measure different things.
Well. I meant to let that go, but I couldn't.
However, you are (in the poker domain) almost certainly right. I say almost certainly, because I've not followed online poker since 1990, any more than I've followed the stock market since 1990.
The parallels look eery to me, though. (cf any number of recent -- post dot-com crash -- books on why investors behave in unusual ways.) Things are only a "given" if they are (a) provable and (b) reproducible. This doesn't seem to work very well in a market situation, or a poker situation, so I suspect your general drift is in the right direction.
Thanks for the kind words about my book
Date: 2007-07-29 01:11 am (UTC)The difference is the dogmatic pragmatists have actually played a lot of poker, and probably made money at some time. Most will probably change their play and (more slowly) their stated opinions. Dogmatic theorists, on the other hand, are rarely players and even more rarely successful players.
I do think real game theory has a lot to say about poker, my criticism is for the textbook examples that some people insist are the whole truth. The problem isn't in the math, it's that to take a real situation and simplify it so textbook methods work ends up destroying the problem.
You have a small but crucial error (possibly a typo) in your account of prisoner's dilemma. If both people confess, both get one year. If one confesses and implicates his partner, while the other maintains his innocence, the confessor goes free and the partner gets ten years. As you say, if both confess, both get nine years.
In this situation, you always save a year by confessing (zero years instead of one if your partner stays silent, nine years instead of ten if your partner confesses). So under the assumptions of game theory, it's irrational to refuse to confess. So perfect game theory gets you nine years, whereas any six-year-old with some guts knows how to get one year.
I have never thought much about EV. I play mostly in cash games, and I know sitting down about how much I can win. Of course, I'd be happy to win more. But each dollar above my target is harder to win than dollar before it, and also causes some erosion of long-term value by chasing some players away from the game or making it unlikely I'll be invited back. If someone goes all-in so there's $20,000 in the pot, with $10,000 for me to call, I don't think about whether my chance of winning is 30% (negative EV) or 35% (positive EV). I think about how it will affect the game if I fold, versus calling and winning, versus calling and losing. Some players in some situations I don't mind calling with anything, because I know I can get the money back if I lose. Other players in other situations I don't want to call unless I have a very high chance of winning, because I think I can get their money later but if I lose now I'll never see their chips again.
I'm also thinking about the effect of this hand on the table and the game. I don't ignore the odds, but treating poker like a sequence of EV decisions misses the point (I'm speaking of no-limit cash games here, in a limit tournament EV is much more important, which is one of the reasons I don't play many limit tournaments). It's like trying to go from New York to Los Angeles by always going southwest. If your goal is a lifetime of winning poker, having $100 EV on one particular decision is at best a tiny step in that direction, and may be a step backward.
Black Diamond Engagement Rings
Date: 2011-09-23 12:56 am (UTC)