Scientists have created a robot that will NEVER lose at a game of Limit Texas Hold’Em Poker. This robot, named Cepheus, may lose on any given hand if the cards are inferior, but over time the robot will always come out on top.
Firstly, this is proof that No Limit Texas Hold’Em is the best poker game in existence. The fact that a robot can be programmed to beat anyone at the game is proof enough that there’s no human element involved, and therefore it’s no longer a skill game. Secondly, holy shit when is this robot going to win a bracelet at the World Series of Poker.
My headline above of ‘on the verge of winning’ may be a slight stretch, but you’d be a fool to think that the day isn’t approaching where this robot takes on a field of the most elite poker players and wipes the floor with them….It’s only a matter of time.
If you’re not already familiar withe nuances between no-limit and limit hold’em, in the latter you can only bet a specified amount each time, and there is a cap to each round of betting. Why is that important here? Because it gives the robot the opportunity to minimize its losses on any given hand where the cards don’t come out as planned.
Poker being what it is, the robot, named Cepheus after a constellation in the northern hemisphere, will lose if it’s dealt an inferior hand, but it will minimize its losses as best as is mathematically possible and will slowly but surely take your money by making the “perfect” decision in any given scenario. Heads-up limit Hold’Em, it can be said, has been “solved.”
“You can play one hand of poker and there are hands that Cepheus will fold, and it loses. That’s poker. But poker is about how you do in the long run and, if you play long enough, Cepheus will never lose,” Neil Birch, cocreator of the robot, told me. “It doesn’t make mistakes.”
That’s the key, of course. Birch and his colleagues essentially “brute forced” the game of limit poker, in which there are roughly 3 x 10^14 possible decisions. That, according to some estimates, is more possible permutations than hands of poker than have ever been played in human history.
“Poker is big enough that just to specify a strategy—to say how we should play each situation—is as big, maybe bigger than the total number of card games people have ever played,” Birch said.
Cepheus runs through a massive table of all of these possible permutations of the game—the table itself is 11 terabytes of data—and decides what the best move is.
The creators of the robot (Michael Bowling, Neil Burch, Michael Johanson, Oskari Tammelin) published their incredible work in ScienceMag, which can be read HERE.
The abstract of the published study is as follows:
Imperfect-information games model settings where players have private information. Tremendous progress has been made in solving such games over the past 20 years, especially since the Annual Computer Poker Competition was established in 2006, where programs play each other. This progress can fuel the operationalization of seminal game-theoretic solution concepts into detailed game models, powering a host of applications in business (e.g., auctions and negotiations), medicine (e.g., making sophisticated sequential plans against diseases), (cyber)security, and other domains. On page 145 of this issue, Bowling et al. (1) report on having computed a strategy for two-player limit Texas Hold’em poker that is so close to optimal that, at the pace a human plays poker, it cannot be beaten with statistical significance in a lifetime. While strong strategies have been computed for larger imperfect-information games as well (2–6), this is, to my knowledge, the largest imperfect-information game essentially solved to date, and the first one competitively played by humans that has now been essentially solved.
And thus the thing that all great poker players have known for decades is proven: that limit Texas hold’em is an inferior game to No Limit. In order to create the perfect test of skill you need to introduce the maximum number of variables before a game gets out of hand. In the case of Texas Hold’Em poker, the variable of betting limits sets apart the computers from the humans.
What’s the point of playing a game if you know that you’d lose to a computer every time? Why even bother unless you’re the Sisyphus of Texas Hold’Em?
Poker is a complex game, involving uncertainty, randomness, luck and bluffing, and Texas Hold ‘em—the most popular variety—is no exception. However, a simpler version of it exists, called heads-up limit, where there are only two players, fixed bet sizes and a fixed number of raises. That’s why scientists from the University of Alberta decided to select this game for their algorithm.
For their study, the researchers improved upon a previously developed algorithm called counterfactual regret minimization (CFR). Regret minimization basically involves reviewing past moves and examining whether making a different decision, such as raising, folding or calling, could have resulted in a better outcome. The computer then calculates how much it lost because of a particular move, and stores that as a regret value. This is then applied to each opportunity that the computer has to make that same decision, so that losses can be avoided.
After using 4,000 central processing units for two months, which is the equivalent of around 1,000 years of computing time, to practice against itself in hundreds of thousands of rounds, the computer gradually improved and developed better solutions. The regrets then became so small that it couldn’t be beaten in a human lifetime.
So essentially this is the machine that every great grinder aspires to be. You might not have that one BIG NIGHT, but over the course of a lifetime you’re ensured to never go broke. I don’t know about you, but that sounds boring as fuck to me. Give me No Limit Omaha any day of the week over limit Hold’Em.