# Chess Computers Reach the Next Level

13 years on from Deep Blue beating Gary Kasparov, the idea of a computer beating a champion human at chess doesn’t seem that outlandish. But now a computer has beaten a leading player at shogi, more colloquially known as Japanese chess.

Why’s that so impressive? Well, the general concept of the game is similar to chess, with a variety of pieces of different ranks that can move in specific ways, the aim being to capture opponents and set up a position where the king is under attack and the player can’t make any legal move to rectify it.

Things are slightly more complicated in that each player has 20 pieces rather than 16, and that several pieces can be promoted upon reaching the opposing side of the board. (Each has its own “promoted” rank rather than taking on the characteristics of another piece.)

But what really adds to the complexity is that once a player captures an opposing player’s piece, he or she can then use it themselves.

Exactly how much more complex the game is is a matter of interpretation. One interpretation is that there are 10 to the power of 224 different combinations of moves across an entire game. That compared to 10 to the power of 123 in chess.

If your mathematics is as rusty as mine can be at times, you might think that means that shogi has about twice as many possible combinations. In fact, shogi has 10 to the power of 101 times more combinations: that’s a one followed by 101 zeroes.

To put that into context, chess has more possible combinations than there are estimated to be atoms in the universe. In turn, shogi is more complicated than chess by more times than there are atoms in the universe.

While traditionally humans do better at computers at such games because they do a better job of considering multiple possibilities and consequences at once (single-core computer processors work much quicker but can only do one thing at a time), a system named Akara 2010 has now beaten leading player Ichiyo Shimizu. The system brought together four different computer programs and made whichever move was suggested by the majority on each occasion.

According to the Mainichi Daily News, human fallibility likely proved the difference on this occasion: ” Shimizu made a questionable move partway though the game, and Akara went on to win.”