Tagged as a biography of seven games: chess, draughts, go, scrabble, poker, bridge, and backgammon, both in their structure and social histories. The social parts are often the most intriguing, especially how some games, at some times, become almost proxies for a person’s character features – to the extent that without proficiency, it’s hard for someone to succeed.
The least satisfying aspect, for me as a computer scientist, was the emphasis on whether, when, and how well, computers can learn to play different games. There are some great stories about the lengths to which some individual programmers have gone to create game-playing programs, that it would be good to learn more about. But it’s easy to assign too much significance to a computer’s gaming abilities. These revolve around the structure of the game and the effort put in to understanding it. But they tell us very little about how people play the game. Taking go as the example I’m most familiar with, using a “deep” network to create AlphaGo makes its playing decisions entirely inscrutable. It provides no insight into either the game itself or the way humans play it so well, which are surely two of the most interesting scientific questions to explore. There’s a tendency to regard a game as “solved” once a computer can beat any expert human – an entirely arbitrary quality threshold, surely – without trying to understand how a human, with far different (and probably less) computing power, came to offer that threshold in the first place.
3/5. Finished Monday 2 May, 2022.
(Originally published on Goodreads.)