I saw and read this article when it first appeared in the online NYTimes. As a live poker player, although not a great one and not that strong at heads up poker or limit poker, I still found the article fascinating, since poker is a fairly complex game incorporating incomplete information and variations in strategy.
By the former, I contrast poker with complex games with complete information, such as chess - or for that matter, such as video poker -- where you may not know what's coming next, but you know all the possibilities (in live poker, "all the possibilities" are too numerous to be useful, and players will therefore try to narrow the number of possible hands their opponent might have, as much as possible, based on their play of the particular hand so far, and to the extent that they have additional information, based on their play of similar hands in the past). By the latter, I mean that a skillful player therefore will not play the same way in the same situation every time, so that their actions hopefully do not provide any additional information to their opponent, since the more a player knows about their opponent's likely hand, the easier it is to play correctly (i.e., in strict accordance with the mathematics of the hand).
Therefore, the application of game theory is important in "live" poker, as it would presumably continue to be when playing a machine. More specifically, a given hand in a given situation may reasonably lend itself to straightforward play or deceptive play, and therefore a choice between, for example, calling a bet or raising, or between, for example, checking, checking and calling, or checking and then raising any bet, as well as the introduction of bluffing in appropriate circumstances and with appropriate frequency, may be randomly decided based on a pre-determined frequency (e.g., bet 60% of the time, raise 40% of the time). Such variations among the player's options make it more difficult for opponents to utilize betting patterns to determine a player's likely hand.
As I understood the article, the poker-slot had to be "dumbed down" not so much to make it interesting, as to comply with gaming regulations that prohibit a machine from changing its operation in response to a particular player's play; the article said:
"Casino commissions, however, mandate that a gaming machine cannot change its playing style in response to particular opponents. A poker game must play a World Series of Poker champion the same way it does a neophyte, so Dahl's machine would not be able to learn from its experience in a casino."
If this information is correct, it would prevent the machine from using all aspects of the artificial intelligence algorithms that were available to the developers. Of course, the quote says the machine can't change its playing style in response to particular opponents, which does NOT lead to the following remark, that it could not learn from its experience -- so the actual prohibitions on how the machine can operate are not readily discernible from this quote. But there are most certainly regulations that sometimes apply -- we are aware that Vegas-style video poker machines must deal the cards randomly from a 52-card deck, while other styles of machines ("bingo" style - and please excuse my likely incorrect use of the terminology here) are not "fair" in the same sense, in that they will provide winners and losers based on factors that do not arise from a "fair" deck of cards dealt randomly.
If the poker machine were able to identify who it is playing against and utilize data on that player's style of play, it would have a tremendous advantage over the player -- most (all?) poker players have at least some, if not many, aspects of their game that are consistent and can be exploited, because they are unaware of that consistency - even if they try hard to eliminate that element from their game.
If a machine could not identify who it was playing, it could not utilize player-specific data to improve its play against the human opponent.
Those few who are winning against the machine have apparently found flaws in some aspect of the computer strategy, and the machine is not allowed to "improve" its play in recognition of those flaws as a cause of losing to the player. However, it is likely that two things will happen: future iterations of the machine, if it catches on, will have improved strategy -- and as we know, skillful players may be prohibited from playing the game, just as they are often barred from blackjack and advantage-play video poker
The comment about the "denomination" of the machine is interesting, but I would speculate that, like some video poker machines, this one is multi-denominational too.
Of course, the response of players is always educational :) I sent the article to a bunch of my fellow live poker players, with the caveat that only a handful of the best could beat the machine and they should avoid it if they see it. Responses ranged from "why would anyone want to play against a computer?" to "I can't wait to play one of these". The most intelligent response noted that the application of game theory might thwart the machine -- e.g., rotating more than one skillful player against the machine on a random basis as a team, with each player using his/her own best strategy / approach --- so that the machine would not be able to make any "intelligent" adjustments to the player's playing style --- however, it may be the case that the machine is not allowed to do this anyway; the rules of the "casino commissions" are not that clear in the article, despite the quotation above (and of course, the use of the term "casino commissions" probably reflects the gambling sophistication of the writer, unless the term was "dumbed down" from "gaming commissions" for public consumption).
Like video poker, slot machines with an element of chance and an element of skill are particularly appealing to some gamblers (some of whom would not even classify themselves as "gamblers" when playing games with an element of skill, even though there is most definitely a significant element of chance involved).
--BG
=======================
2a. Re: Fascinating new poker machine
Date: Wed Sep 11, 2013 1:13 pm ((PDT))
check this out:
http://forumserver.twoplustwo.com/showpost.php?p=39069549&postcou\nt=215
--- In
vpFREE@yahoogroups.com, <
vpfree@yahoogroups.com> wrote:
Article says: "So far, over the course of 500 hours, he says, he
averages about $135 per hour in profits, playing at the $20- or
$40-bet level."
Translation?: $135/hr at 20-40 limit (=$10 denomination machine,
=6.75BB/hr)
Why?
The article tells you they had to dumb the bot down in order to
get any interest. But maybe that's just an excuse, maybe their
technique for making a bot is not so hot. Sure it bluffs and can
fold at the river at a reasonable frequency now, but is that all
there is to winning poker?
--- In
vpFREE@yahoogroups.com, <
vpfree@yahoogroups.com> wrote:
Last paragraph of the article sums it up when writer is in Vegas
and catches up with Chan:
"...I asked him if he thought anyone could win big over the long
haul, beyond what Mike Reed has managed to grind out.
“Nothing is impossible,� Chan said. Then he added
that some people see Texas Hold ‘Em Heads Up Poker as a
potential gold mine. “You probably have five groups out
there right now, testing the machine, writing down all the
results, getting every play blow by blow and figuring out what to
do on every hand. One hundred percent, I’m sure there are
teams working on it...�
HAL2000 is on the way.
--- In
vpFREE@yahoogroups.com, <
vpfree@yahoogroups.com> wrote:
http://www.nytimes.com/2013/09/08/magazine/poker-computer.html?_r\=0
New format of posting is extremely difficult. Did this using
reply to another message.
