The Mystery of Go, the Ancient Game That Computers Still Can’t Win

I think the main reason has to do with score estimation. With chess, you can say that a queen is worth 8 points and a rook is worth 5 and a bishop's worth 3, and get a quick idea of who's winning a game. In Go, every piece has the same value, and whether a stone is useful depends on which other stones it's connected to, and whether those stones are going to be alive at the end of the game. So just to work out whether you're winning, you have to be able to do search all the way to the end of the game and work out who wins! And computers have been getting that process wrong until recently.

The extra search space doesn't make Go harder for humans too, because humans play both Chess and Go very differently to computers. A human never tries to calculate the value of a million possible moves, so the question of how many million possible moves there are to search through just isn't relevant to a human. Humans play both Chess and Go largely using pattern recognition and insight in a way that computers are not currently able to model.