Can You Beat This Virtually Unbeatable Poker Algorithm? 

Can You Beat This Virtually Unbeatable Poker Algorithm? 

The challenge is on. Computer scientists say they have created an algorithm that has essentially solved a version of Texas hold ’em, and it’s guaranteed to beat every single puny human competitor in the long run. Don’t believe it? Why, you can play against the program yourself.

Cepheus, as this poker-playing program is called, plays a virtually perfect game of heads-up limit hold’em. The variant is like the popular Texas hold ’em, except there are only two players and a fixed number of bet sizes and raises. That still leaves 3.16 × 1017states in the game.

The sheer feat of a program that’s essentially solved a type of poker has computer scientists in a tizzy. Unlike other games that have been completely solved such as checkers or Connect Four, where every past action is laid clear on the game board, poker is a game with imperfect information — namely, you don’t know your opponent’s cards.

Cepheus is able to learn from its past mistakes, and it achieved its current high status by playing billion and billions of hands against itself. It can even bluff. For game theorists, Cepheus could hold the key to modelling real-life situations like negotiations or auctions where information is imperfect.

But hey, you want to know if you can beat it poker. It is of course possible to beat Cepheus in individual games, as poker involves an element of chance with each dealt hand. But in the long run, the computer will almost always come out ahead. There is a tiny margin of error, its creators say, but it is so small as to be practically negligible in our human lifetimes.

So without further ado, say hello to Cepheus.

Update: Apparently so many of you want to play against Cepheus, you’ve taken down the site. It’s been in and out — try checking back in a bit.

Picture: Kucher Serhii/Shutterstock


The Cheapest NBN 50 Plans

It’s the most popular NBN speed in Australia for a reason. Here are the cheapest plans available.

At Gizmodo, we independently select and write about stuff we love and think you'll like too. We have affiliate and advertising partnerships, which means we may collect a share of sales or other compensation from the links on this page. BTW – prices are accurate and items in stock at the time of posting.