Will Computers Kill Chess?
How close is chess to a “Solved Game?”
“No computer will ever beat me in a match.” -Garry Kasparov, Oct. 1994
In 1997, “Deep Blue” — an expert system with both hardware and software created specifically to play chess — defeated the World Champion in regulation match play. This computer is now on display in the National Museum of American History.
Let me say it again: computers have been defeating the very best humans at chess for so long, they are in museums.
The true beauty of chess is this:
Two people sit down across a board of sixty-four squares. For the duration of the game, this becomes the world. There is no chance, no random caprice of dice or cards. Each player can see exactly what the other sees. The contest involves human creativity, imagination, problem solving, deep thinking, stamina, and will power. In a good game, neither player makes a stupid mistake. Instead, the unfolding of the game, move by move, reveals that one player’s plan was superior.
One of the beauties of exploring the games of the very best players is that they can be wild, surprising, incredibly daring! Really strong players often play to avoid strategic errors, positional deficiencies, and prevent tactical scenarios they can’t see the bottom of. But the great players navigate into those terrifying waters.
These grandmasters, brilliant and daring, capable of seeing the tactics of risk and sacrifice through to victory simply cannot see as far, as extensively, or as deeply as a computer that can calculate tens of thousands of possibilities for each one of our own, with no lapse in the accuracy of their calculation.
Deep Blue consisted of two towering racks of custom hardware. It was essentially a network of 32 computers, each ready to solve separate lines of possibility in parallel. This machine represented the investment of one of the top technology companies of it’s era.
Since then advances in both the science of software design and the capabilities of hardware mean that the phone in your back pocket with an open-source software app freely available to the world could crush this machine… and any top grandmaster.
Chess is not, technically, a “solved” game. You probably remember the point in your life when you realized that tic-tac-toe is a solved game. (Chickens can be taught to win.) So you get it.
Chess doesn’t have that kind of “solution” — but no amount of human creativity will ever again show the mindless machine who’s boss. Some games with more complex computational characteristics took longer to reach this state: for a long time the elegant game of “go” was resistant to strong computer play. But it is all falling by the wayside now.
Where does that leave humans and chess? Where does that leave our relationship to computers?
“So what if they can play better chess than humans? Cars go faster than humans, too, but we still have the Olympics.” — Eric Martin, rec.games.chess.misc, 1996
Chess is an exquisite competition of human versus human. The elements referenced above remain a powerful drug: creativity, imagination, willpower, intellect. We love to strive against each other, at every level of skill.
And this was always true. For any person who does not happen to be world champion, there was always that knowledge out there that our moves are never the very best possible ones, there is always someone who can crush us like a bug. We used to go to those people to teach us to get better. (We still do.)
But now we can go to the computer for those lessons as well: databases with every recorded game in the history of competition, inventories of chess puzzles, play against a machine at tunable levels of strength. We use the machine not to measure our pathetic human failings, but to develop our faculties, expand our knowledge, train our pattern recognition…
… and to cheat.
Oh, yeah, we do that.
“On Friday, Emil Sutovsky, director general of the World Chess Federation (FIDE), wrote in a Facebook post that Rausis was “caught red-handed” during a recent tournament in Strasbourg, France, confirming long-held suspicions about the Latvian-Czech player.” — Washington Post, 2019.
I am writing this during the covid pandemic, most or all tournaments are purely online now. What is to keep a person from having chessbase in one window, stockfish running analysis while you navigate the optimal opening tree? How do you know you are playing a human, and not a human who is just the go-between for a computer?
Well, that would take all the fun and all the glory out of winning wouldn’t it?
Remember: some people do play for money, and when money’s on the line, humans have a history of cheating. And other people… well… I don’t know what to say about the pretend-glory of winning a game on behalf of your iphone. Perhaps we should ask Igors.
Computers to the rescue!
Many tournaments require a video conference setup where you and your screen can be seen at all times. A camera over the player’s shoulder making sure that no untoward windows are being referenced, or secondary devices utilized.
And the games can instantly analyzed by computer as well to determine the accuracy of play. A class B or C player such as myself (this is a seemingly infinite distance from grandmaster play, by the way) who suddenly makes surprisingly optimal moves from a computer’s perspective is almost certainly using a computer.
Even as cheaters find ways to be more sophisticated, so do detection methods.
Remember Kasparov losing to Deep Blue in 1997?
His first outraged rant after the humiliating defeat was: they cheated!
Kasparov’s accusation of cheating was because the computer didn’t play like other computers of its time, automatically capturing pieces as the only measure of success. (Want a deeper dive: here is analysis of the game and position, with reference to subsequent software approaches to the same position.)
The historical conclusion is: Deep Blue not only didn’t cheat, it even made a few mistakes in that game. Kasparov should have been able to play to a draw.
Chess may or may not be a “solved” game in our lifetime or any. The full decision tree of all possibilities exceeds the thermodynamic barrier of practical solution. (But the full decision tree may not be required to mathematically solve the game, so the jury remains out on this one.)
What is abundantly clear is that many people who love chess and all the wonderfully human experiences of playing the game and improving in skill at the game have grown up in a world where neither they nor any living world champion could ever reliably beat a computer.
The necessary conclusion is, just as Eric Martin insightfully predicted in a casual Usenet post in 1996, it’s not even close. As fast as cars (or, soon enough, robots) can go, we still love our olympics. And as good as computers become, perhaps even to the full solution, we still love our chess.
Hell, some people still play tic-tac-toe, chicken or no chicken.
How about a chess puzzle, today from the site “shredderchess”…
