At first glance, the subtitle almost looks like a misnomer, while author Oliver Roeder spends a chunk of the book working through seven games—checkers, backgammon, chess, go, poker, Scrabble, and bridge1—and how the best players in each play, a large portion of the book is actually about AI and computers. From AlphaGo to Deep Blue, and the many programmers behind each computer, Roeder is fascinated by how each of these computers tried and—for the most part—succeeded in besting the best players at their own game. What does that mean for the game? And what does that mean for humanity?
To paraphrase some of these top players, the outlook is quite bleak. Seeing a top player defeated at the hands of a computer seems like an existential loss for humanity. Some argue that the art of the game, the beauty of it is gone, replaced by algorithms and cold calculations. The programmers might disagree. Is a programmer that creates an AI to play a game, not also playing that same game? They see their work as extensions of the same pursuit of greatness that some of these top players embarked on. The same top players that twisted Scrabble until they were no longer creating comprehensible words, or the players that religiously pour over every existing opening literature to master a game.
For the programmers, and perhaps to Roeder, the obsession is quite similar. When emerging Go players looked to introduce some standardization to the competition, the old guard decried it as taking the art out of Go. Then, decades later, when AlphaGo triumphed over those same now established Go players, the players decried this moment as the death of Go as an art form.
Something about a game brings it out of people. This is what drew programmers to create supercomputers with the sole purpose of being good at checkers—not something that benefits society, but a machine that can excel at a game. Sure, the grant proposals would list out goals about proving prowess, and there are some algorithmic systems that do translate to say—protein-folding, but for the most part, even the architects of these machines admit that these pursuits don’t necessarily teach us all that much about artificial intelligence or have applications outside of creating a machine that is really good at a very specific thing.
But it might teach us quite a bit about humanity. Since ancient times, games have been a way for people to create structure as a way to practice agency. So charting games and what we play not only provide cultural context, but also get at the more nuanced why’s and how’s in what it means to be human. Whenever a Go player says the art is loss, what exactly is lost? What is lost when a computer beats a human player in a game?2
Just as the games we construct for ourselves reveal quite a bit about who we are and what we value, the computers we build, the problems we seek to solve are also a mirror of sorts. It’s a very human endeavor to aspire to be the best backgammon player in the world, just as it’s a very human endeavor to want to create a backgammon computer to beat the best player in the world. It’s all a game at the end of the day.
Footnotes
-
Bridge might be a game in which computers will never best humans, a combination of the nuances of the game, and waning interest in Contact Bridge. ↩
-
What itch is being scratched by the challenge and friction of learning and getting better at a board game? Why is luck thrilling? And why would we set limitations (via rules) as a form of play (which intuitively feels would want to be unstructured and open)? ↩