While the threats posed by certain types of poker bots (a/k/a automated poker decision-making software) are a real and ongoing threat to online poker, there are some more fun stories related to the topic that are still worth the read. For instance, there’s the recent announcement jointly made by Facebook’s Artificial Intelligence initiative and AI researchers at Carnegie Mellon University about an inexpensive, self-learning bot that crushed a large handful of many notable pros during extended sessions of six-handed no-limit hold’em.
The software’s success is notable in large part because it’s the first report of a poker program succeeding against top human competition in a format other than heads-up no-limit hold’em, which offers a far simpler, in comparison, strategy tree. Carnegie Mellon’s multi-player bot was built upon the success of Libratus, the poker bot that clobbered several pros in a match in a series of matches in late 2016.
According to a statement Facebook published about the Pluribus study, the program adjusted for the complexities of six-max NLHE by not seeking to look too far down the road of possible outcomes, but instead focusing on short-term outcomes and probabilities and continually adjusting its strategies on the fly. If one digs into the study, one discovers that among the many strategies options the Pluribus program selected was to stop limping, instead adhering entirely to a raise-or-fold strategy.
The strategy worked well enough to score consistent profits against a bevy of top pros, including Chris “Jesus” Ferguson, Greg Merson, Darren Elias, Jimmy Chou, Seth Davies, Michael Gagliano, Anthony Gregg, Dong Kim, Jason Les, Linus Loeliger, Daniel McAulay, Nick Petrangelo, Sean Ruane, Trevor Savage, and Jake Toole.
The Facebook presser stated, “In recent years, new AI methods have been able to beat top humans in poker if there is only one opponent. But developing an AI system capable of defeating elite players in full-scale poker with multiple opponents at the table was widely recognized as the key remaining milestone.”
The study also claimed these as its main findings:
- Pluribus is the first AI bot capable of beating human experts in six-player no-limit Hold’em, the most widely played poker format in the world. This is the first time an AI bot has beaten top human players in a complex game with more than two players or two teams.
- We tested Pluribus against professional poker players, including two winners of the World Series of Poker Main Event. Pluribus won decisively.
- Pluribus succeeds because it can very efficiently handle the challenges of a game with both hidden information and more than two players. It uses self-play to teach itself how to win, with no examples or guidance on strategy.
- Pluribus uses far fewer computing resources than the bots that have defeated humans in other games.
- The bot’s success will advance AI research, because many important AI challenges involve many players and hidden information.
Instead of joining the “death of online poker” parade of stories talking about Pluribus, let’s offer a healthy dose of reality. Multi-player botting software has been around for years, and this doesn’t change that a bit. Instead, though, as the bots improve — and the Facebook report unfortunately offers some strong pointers to botters on how to improve their bots — the reality of it all will continue to force online-poker operators to invest more resources into bot detection. Virtually any bot can be identified and blocked if sufficient energy is invested; bot play differs from live-human play in any number of ways that an online site can detect. As we’ve said all along, the sites that are serious about blocking such software will find ways to adapt and thrive, and those that don’t, won’t.