Poker theorists know very, very little about poker?

[Lottery Larry]

"Jackpot Jay: Is everything important already "known" about poker? If not, what is there still to learn?

Mantros: Not even close. We, as poker theorists, actually know very, very little about poker. As far as I know, we don't even know what would constitute optimal play in a heads-up match, let alone in a multihanded game of no limit hold 'em. It's hard to say what there is still to learn because we really don't know anything right now, relatively speaking.

Basically, all the poker strategy you read about anywhere is an educated guess at what the "correct" strategy is. But we don't actually know what the "correct" strategy is."


Any protests? Or do we have 1000 more 2+2 books to plan on?

[gumpzilla]

I don't think this is particularly controversial. There are clearly many individual decisions in specific contexts that we can assess the profitability of, but taking all of the isolated decisions and unifying them into one clear, coherent strategy is a much different beast.

[GeeeJay]

I don't believe that this is correct. On the other hand, poker theorists DO know EXACTLY what they're talking about. It is my opinion that you are confusing your opponents play (and your response) of the hand with the actual theoretical applications. If you could SEE what the other had, the play would be rather standardized, based on pure probability, wouldn't it ? It is because of the VARIABLITY of the way the hands are played based on purposeful deception that confounds the issue.

[TStoneMBD]

originally, when i first read the topic, what immediately jumped into my head was that most poker theorists have alot to learn about heads up play. there are no heads up books, but there will/should be. the internet is changing things, and when enough hands are compiled by intelligent people, you will start seeing some staggering results.

[Lottery Larry]

Maybe this wasn't as stupid a statement as I originally read it to be?

[splashpot]

It's so incredibly hard to gauge what is correct play in every situation. The reason for this is that the most important factor in most poker situations is the way your opponent will play. Since this is so relatively unpredictable, it is difficult to determine correct play.

[Cyrus]

[ QUOTE ]
"We, as poker theorists, actually know very, very little about poker. As far as I know, we don't even know what would constitute optimal play in a heads-up match, let alone in a multihanded game of no limit hold 'em. It's hard to say what there is still to learn because we really don't know anything right now, relatively speaking."

[/ QUOTE ]

If this was truly said by a poker expert, then he is not an expert at much, let alone poker.

Among other things, he succumbs to the usual confusion between strategy (overall, long term, general planning) and tactics (situation-specific planning). Poker experts, such as the resident stable of racehorses, certainly already know most that needs to be known about poker strategy in today's poker.

The application of that knowledge is what befuddles the masses, who mistake the lack of specific instructions for ignorance. It's typical of most people to resent sound but general advice, such as "Casino roulette has a negative expected value for the player", and prefer instead astrology, which comes on as specific, e.g. "The number 14 is in your future".

In poker, as in chess, we certainly know what is optimal, under the assumption of specific ircumstances. The underlying principles behind a player's strategy in a game, or his tactics in a specific hand, should be robustly correct, i.e. logically mathematically sound. A chess master who plays a game of chess sees a position on the board for the first time in his life very, very often -- despite the fact that chess masters spend most of their time with a chess book or in front of a chess board analyzing new positions. The chess master, then, applies his knowledge (of correct play, of information about his opponent, etc) and uses his mind (to analyze the position itself) and comes up with what he believes to be "the objectively optimal play". The path the chess master takes might subsequently proven not to be optimal, but the underlying principles should be the optimal ones.

We cannot assess each and every possible situation (the number of possible situations is too big) but we can certainly identify robustly correct play and analyze situations and formulate optimal strategy. (We can also run things by the computer, now.)

Poker play is analogous.

--Cyrus

[ACW]

I think the situation with poker at the moment is similar in many ways to the explosion of bridge in the 1930's.

Ely Culbertson published his Blue Book which taught people the basics of the game, helped them to correct their previous moronic strategies and tactics, and was a great publicist for the game.

There were still huge developments to come. Some of his ideas turned out to be vastly inferior to alternatives (e.g. for the bridge literate amongst you compare Roman Key Card Blackwood to the Culbertson 4-5NT convention). Key areas of the game were, as far as I know, completely untouched by his writings (Principle of Restricted Choice, Law of Total Tricks).

I think the current state of understanding and literature on Poker is more advanced than Culbertson was, but probably not by an order of magnitude. I think there's a long way to go, with significant discoveries still to be made.

Sklansky wouldn't be spending nearly so much time thinking about poker theory if he thought there was nothing left to learn - do you see why?

[EjnarPik]

Do you consider RKCB to be vastly superior to 4nt-5nt?

Ejnar Pik, Southern-Docks.

[ACW]

Perhaps I'm being a little unfair, given that I've never actually used the Culbertson convention in anger. The fact that 90%+ of top players use RKCB and <1% use Culbertson 4N-5N seems compelling evidence. RKCB is widely abused, and suffers from the idiotic problem of having two conflicting but equally valid variants, but that doesn't mean it's not a great tool if correctly used.

[gumpzilla]

[ QUOTE ]

We cannot assess each and every possible situation (the number of possible situations is too big) but we can certainly identify robustly correct play and analyze situations and formulate optimal strategy.

[/ QUOTE ]

I think you have it ass backwards. What we can do is precisely the tactical stuff - analyze very specific situations. Optimal strategy is precisely what we don't have, though. Particularly in something like heads up, it should be possible to formulate an unexploitable mixed strategy; however, such a strategy would be incredibly complicated. I know some people in the STT forum have done some looking at this, but even there, their findings are specific to very particular stack sizes relative to the blinds and they assume even stack sizes when you start, which are some fairly strong restrictions.

Perhaps you mean something different; would you mind elaborating?

[EjnarPik]

It's only superiority over 4nt-5nt, is that it is easier to learn.

Anyway this is a sidetrack, your argument is valid. There was a lot of things Culbetson said, that were way better than things were before him, and a lot worse than things after him.

Ejnar Pik, Southern-Docks.