Question for Mason

[TheShootah]

Mason, just got Poker Essays 1. Good book sir! Anyway, you mention 2 responses to your essays on Limit vs. No Limit Hold Em I. What kind of responses did you get to the second. BTW, I agree wholeheartedly with you, although I think they would tend to believe you more than I. [img]/images/graemlins/grin.gif[/img]

[Student]

What struck you as being the most powerful idea(s) in "Poker Essays I?" Is it suitable for beginning poker players?

[Jordan Olsommer]

[ QUOTE ]
Mason, just got Poker Essays 1. Good book sir! Anyway, you mention 2 responses to your essays on Limit vs. No Limit Hold Em I. What kind of responses did you get to the second. BTW, I agree wholeheartedly with you, although I think they would tend to believe you more than I. [img]/images/graemlins/grin.gif[/img]

[/ QUOTE ]

If I may be so impolite as to tack on a subquestion of my own to this, if limit is actually more difficult than no-limit, then why are computers (specifically the U of Alberta's AI Poki) now virtually unbeatable at limit yet relatively inept at no-limit? It was usually thought that the longer it took for computers to "crack" it, the tougher the game is - checkers, which was bludgeoned unrecognizeable by the Chinook program, isn't as difficult as chess, which Deep Blue staked out in 1997, and neither of them are generally thought of as being as difficult as Go, which computers are pretty much hopeless at at the current time.

It would seem to me that limit is more punishing for a beginner because the nature of the game means they have to make more decisions, and being beginners they are likely to make poor decisions, and more poor decisions = more -EV in the long run. But that still doesn't explain to me the discrepancy between the skill of limit holdem computer programs and their relatively donkified no-limit counterparts. Any thoughts on this?

[disjunction]

I'm more rusty on this than I should be, but here it goes:

Think of each decision in limit hold-em as a 3-way fork in the road: call, raise, or fold. In general, computers do a good job at games where there is a low "branching" factor, in other words, where it can number-crunch its way to exploring most of the possible paths. This is why computers are superb at checkers (very few possible moves each turn), very good at chess (about 20 possible moves per turn), and bad at Go (very large number of possible moves).

As you can easily see, there are more possibilities on each betting round in no-limit hold'em than limit, therefore, almost by definition, the no-limit problem is harder than limit. If you came up with a good algorithm for use in no-limit, that uses the pot-odds in some way, you could almost definitely apply it for use in limit. This is because limit-hold'em is a special case of no-limit.

EDIT: Changed "bet, raise, or fold" above to "call, raise, or fold".

[Jordan Olsommer]

[ QUOTE ]
I'm more rusty on this than I should be, but here it goes:

Think of each decision in limit hold-em as a 3-way fork in the road: call, raise, or fold. In general, computers do a good job at games where there is a low "branching" factor, in other words, where it can number-crunch its way to exploring most of the possible paths. This is why computers are superb at checkers (very few possible moves each turn), very good at chess (about 20 possible moves per turn), and bad at Go (very large number of possible moves).

As you can easily see, there are more possibilities on each betting round in no-limit hold'em than limit, therefore, almost by definition, the no-limit problem is harder than limit. If you came up with a good algorithm for use in no-limit, that uses the pot-odds in some way, you could almost definitely apply it for use in limit. This is because limit-hold'em is a special case of no-limit.

EDIT: Changed "bet, raise, or fold" above to "call, raise, or fold".

[/ QUOTE ]

Thank you very much for the well-thought-out response. But what I meant was given that that is so,how do you reconcile this with the viewpoint that limit is more difficult than no-limit?

[disjunction]

[ QUOTE ]
Thank you very much for the well-thought-out response. But what I meant was given that that is so,how do you reconcile this with the viewpoint that limit is more difficult than no-limit?

[/ QUOTE ]

I disagree with the given, and I'm not sure Mason said this, unless you can be more specific.

I don't have the essays in front of me, but I remember 2 relevant essays from my reading several months ago, to which you could be referring. IIRC -- One stated that no-limit had less variance than limit, which doesn't really make it "easier". The other was more of a devil's advocate argument, refuting no-limit players who think limit is somehow "easier". I don't remember that essay stating limit is harder, I only remember it saying that there are some reasons that limit could be seen as harder.

[Jordan Olsommer]

[ QUOTE ]

I disagree with the given, and I'm not sure Mason said this, unless you can be more specific.

[/ QUOTE ]

"I think the example just given [of how easy it is protect your hand in no-limit vs. limit] demonstrates why [no-limit] games have died out. No-limit was too easy to play well (at least many situations seem to me to be veyr straightforward, and if you didn't play well, you were quickly cleaned out...The edge that the expert no-limit player had on weak opponents is just too great, because often little doubt exists as to what the correct decision is, and when a weak opponent does not make the correct decision, he has only a slight chance and is usually severely punished for his error." - Poker Essays, p. 86 (emphasis mine)

"When [the essay 'Limit Versus No-Limit Hold'em'] was first published, it stirred up quite a controversy. My claim that limit hold'em was more complicated than no limit hold'em was more than some people could take." - _Poker Essays_, p. 87 (emphasis mine again)

"Another reason why limit hold'em is more complex is that you often go all-in on an early round when playing no-limit. Any situation where you just look at the cards on fourth and fifth streets has to be much less difficult than when you are either faced with a bet or have the option to put your money in." - _Poker Essays_, p. 88

[disjunction]

I should note that if you assume your opponents play optimally, no game is "easier" than another, since everyone is 0 EV, so we're really stuck with a definition problem here, unless somebody can clarify.

[k_squared]

that account fails to take into consideration the complexities involved in both games. The problem with limit is that playing well (at leat protecting hands in big pots) often requires counter-intuitive strategies that are not even close to straight forward. The decision is not clearly A, B, or C... because a number of complicated criteria exist in having to decide when to check, raise/bet or fold.

What makes no-limit 'easier' to play is that it is much easier to manipulate the pot-odds to make it a very clear whether or not your opponents play is correct. You run into the sort of case where multiple people are making the right play much less often. Any game that gives a significant advantage to a skilled plaer over an unskilled one is 'easier' if unskilled players are involved.

It is an entirely different matter when you make the game full of skilled players...

-k_squared

[disjunction]

[ QUOTE ]
that account fails to take into consideration the complexities involved in both games. The problem with limit is that playing well (at leat protecting hands in big pots) often requires counter-intuitive strategies that are not even close to straight forward. The decision is not clearly A, B, or C... because a number of complicated criteria exist in having to decide when to check, raise/bet or fold.


[/ QUOTE ]

I'm not sure how you mean "counter-intuitive strategies" would affect a computer. Typically computers assume that their opponents play optimally. It follows all the possible paths the action could take, and computes an expected value based on it. This is not intuitive/counterintuitive, merely a computation of expected value.

No-limit can not follow these branches, because there are too many... I agree that games where your opponents make more mistakes tend to be "easier" than games where opponents make fewer mistakes, but could this really be what Mason meant?