Heads-Up No-Limit Hold'em Poker Is WAY Harder To 'Solve' Via Computers Than ChessChess 'A Distant Dream,' So No-Limit Hold'em Almost Impossible |
|
For years there has basically been a consensus that heads-up limit hold’em is “solved,” but recently a team from the University of Alberta unveiled the latest in unbeatable poker bot technology. In other words, a computer can tackle the entire complexity of the game and play what in human terms can be considered perfect. However, what are the chances of one-versus-one no-limit having the same fate?
It’s almost unbelievably unlikely.
The difficulty in solving a game is determined by the number of total positions possible within the game, as a report from fivethirtyeight.com underscored (graphic right). “Solving chess…remains a distant dream,” the report said about the centuries-old board game.
Despite computers being able to beat the best chess players in the world, in order to make a computer play perfect chess, our current technology would have to be pitched in favor of something better. In other words, our computers are too stupid at present.
Wondering why the best in the chess world don’t play the best computers anymore in high-profile matches like Garry Kasparov once did? It’s not because the human couldn’t score a victory, but because there’s so little money in such matches these days.
Solving limit hold’em via sophisticated algorithms isn’t actually as impressive as solving checkers was in 2007, if you go by the number of total possible positions. Seen in this light, the Cepheus Poker Project isn’t really a game-changer, but it’s still pretty cool for poker theory.
However, when talking about heads-up no-limit hold’em, the complexity is near mind-boggling, as the chart indicates. With a “state space” of 10^140, which dwarfs chess’ 10^50, figuring out to make a machine “indistinguishable from perfect” in what has been dubbed the “Cadillac of poker” might not ever happen. Let’s not even mention a nine-handed NLH poker game.