So, it’s official. Computers are better than humans at Go - AlphaGo has beaten the top pro decisively and the pros generally agree that AlphaGo is significantly better than the top pros.
Chess has been in that unenviable position since Deep Blue, but Chess as a game seems to be going just fine, with human teachers and clubs and so on. One comparision I’ve heard is that humans don’t race against cars.
How did you take it, though? It feels kind of futile that I’m playing a game that is for all intents and purposes solved. But I suppose in the end, all games will are solved in that sense… help me out here.
I think Dutch chess player and writer Tim Krabbe put it quite well when he said (something like) “Big deal. Man has once again built a machine that can lift rocks heavier than he can lift by himself”. I haven’t played chess regularly for years, but when I do, I’m just interested in beating the computer (on whatever low difficulty level I choose) or person in front of me. I don’t care that a computer exists who can play the game much better than me. I mean, taken to its logical conclusion, the position that we can’t enjoy chess because computers have (virtually) solved it is equivalent to saying that participating in any activity is pointless unless you are the best in the world at it.
Here’s another analogy that may help - you and I probably wouldn’t have much fun playing noughts and crosses (tic-tac-toe) against one another, as I (and I presume you) would never make a losing move. But for (say) the average six year-old, the game could be a fascinating battle of wits.
There are other aspects too - often in chess you can create something almost beautiful, or at least unusual, or interesting. Then there’s the fact that chess, like many activities, lends itself to study and self-improvement.
I could probably go on but I hope some of this is of interest.
There are plenty of people in this world who are better at playing chess than I’ll ever be. So what? I can still enjoy playing chess even if I’m never going to be the best player in the world.
The growth of computer superiority brings the number of classes of entities which can defeat me at chess to two: most human beings, and some computers.
There have been vehicles that can traverse 100m in a straight line a lot faster than even Usain Bolt for a long time. And yet we still have the 100m dash.
For me, the issue isn’t that computers play the game better than humans do. As pointed out, lots of people play chess better than I do, and that didn’t stop me from beating my head against them in tournaments.
I stopped playing when it was clear to me that the “solution” was inevitable. And despite any number of impassioned declarations that that’s not true, I maintain nonetheless that computers will “solve” chess, in essence first, then in fact. At that point, I realized that all I was doing was being bad at solving a puzzle (which fact was emphasized somewhat by the increasing depth to which people had to memorize openings thanks to the computerization of opening knowledge).
I’ve played and taught chess for about 50 years now (my highest ELO was 2390.)
I realised early on that there would always be better human players than me (even if I had turned professional), but I have enjoyed doing my best.
Chess is a pure battle of analysis and experience.
It’s immensely satisfying when you successfully work out the likely continuation and predict the next few moves of the game. (It can also keep you awake at night about some blunder you made!)
I’ve travelled all over the World to play in tournaments and met some amazing people.
There is no limit to the possibility of computing power in this regard. I should think everyone would realize this by now. It’s not a matter of “if”, but of “when”.
There are few things a 10-year old kid can’t do better than I can, and the number shrinks every day…I’m usually near the bottom of any on-line gaming leaderboard or scoring. I still have a long disappointing life left ahead of me if I let it get me down.
Chess will never be solved via brute force, but that’s not the same thing as saying it’ll never be solved. By analogy, one can construct a Nim game with an even larger solution space than chess, and yet Nim, of any size, is still solved. Now, obviously the solution to chess will be more complicated than the solution to Nim, but I have no difficulty envisioning it being solved using restricted brute force, similarly to (say) how the four-color map theorem was finally proven: That is, some human or computer or combination of the two breaks chess up into some large but not combinatorically-large number of subproblems, and then computers solve each of those subproblems via brute force.
I’ve long thought the way to accomplish it is to remove what we already “know” to be trivial moves at the beginning, and use brute force on the remainder. That doesn’t truly “solve” it, but it does effectively solve it, as we can be pretty certain that 1. a4 is not a winning opening move.
Tic-tac-toe is solved. We know how to either win or draw any possible game of tic-tac-toe: Make the first move in the center position. It’s impossible to beat you if you play intelligently from there. After you’ve made that move, the intelligence required to not lose is minimal.
Chess, on the other hand, is not solved. Computers can beat humans because computers can evaluate moves better, because they can look more moves ahead than humans can. That’s it. That’s what the best chess playing algorithms do. We don’t know any specific rules for how to always win or draw, even if you stipulate that you’re always going to be white. All we know is that looking ahead more moves will result in victory. Which isn’t a very exciting conclusion, frankly.
You might be able to shrink the solution space by a factor of several hundred that way, or even a thousand. But a thousandth the size of the solution space is still far too big for brute force solution. And you can’t even shrink the solution space by as much as you think, because the truly idiotic moves lead to shorter games, more quickly won by the one not making idiotic moves, and hence don’t take up as much solution space as the good moves do.
Also, if you shrink the solution space in the early game, then whatever conclusion you reach at the end will be tentative (unless you only shrink white’s first move, and then find a forced win for white).
If you didn’t bother to evaluate dumb moves then the game is not solved, since some apparently dumb moves can lead to an advantage.
