Unless anyone can beat Deep Blue, I’d say DB is the world champion, now and…at least until something better comes along, and it’s not likely to be human.
While that’s true in some technical sense, it’s so misleading that it might as well be labeled “false”. First of all, solving the storage problem in this case means something considerably greater than storing a bit in every subatomic particle in the Universe. Second, even if you somehow magically had that much storage space, and even if your database were organized perfectly logically and systematically, finding anything in a database that size in less than the lifetime of the Universe would require nearly as much processing power as just playing the game perfectly to begin with. Third of all, while using a tablebase doesn’t require processing, creating it is. So it’s not really “no processing”, it’s just “someone else already did the processing”.
Because Chess is a competition between 2 people, under specified rules.
Math is not like that.
It’s difficult to say, since there is rating inflation plus modern players have all the benefit of seeing more historical games / ideas.
My personal top 3 would be
Emmanuel Lasker
Paul Morphy
Garry Kasparov
Morphy was so dominant in his time he basically broke the game. Lasker was champion for 27 years (!) and when he came out of retirement in his 60s to play in a chess tournament, he came second.
Garry Kasparov is known as “the great one”. Enough said.
Though Carlsen could find a place in this top three, but he’d have to achieve a hell of a lot first.
Deep Blue was disassembled immediately after beating Kasparov, and the game is controversial in terms of whether it showed at the time that computers could out perform humans. For example, IBM had studied Kasparov’s past games in detail, but Kasparov had not seen any of Deep Blue’s games (not the 1997 version of the program anyway). Plus Deep Blue was human-assisted.
However, in the years since, processing power and AI algorithms have improved a great deal, and no-one disputes the strongest “players” are computers.
Chess is absolutely reducable to pure mathematics and can be studied that way. Now, as Chronos points out, it’s not feasible to solve it since the solution space is so large, but that doesn’t mean it’s not math.
It’s an interesting thought. I would say this is very unlikely. Right now anyway.
For one thing, chess computers will use sound openings and (as far as they can analyze) sound strategies.
The first time you sit in front of a human they will throw traps at you which may not be sound, but which are often effective against unprepared players. You’d get torn to shreds at the amateur level.
Of course if your experience of playing computers means you can analyze millions of moves a second, then you’d probably be fine, but that’s not how humans play chess: we pattern match.
(The reason I said “right now anyway” is because eventually human-like AI will likely advance to the point where you could play human-like AI opponents. But right now it’s remains surprisingly difficult to make an AI do all the dubious things humans might do.
And it’s not really a goal anyone’s shooting towards (because why)).
With some interesting results. I personally am the sorriest chess player imaginable and you could probably give me a Queen every time, but I enjoy reading about it and I stumbled across this page on seven-man tablebases. Amazingly, there are endgames where a pawn only halfway up the board beats four minor pieces… and some positions where a pawn beats four Queens (sic) even with the stronger side to move.
Not a sure thing, just a claim by Kasparov. Methinks it was sour grapes.
If by “human assisted,” you mean the program modifications between games, I don’t see any difference between that and human players analyzing their own moves between games.
- He misinterpreted a computer bug as deep intelligence, too, so I don’t have much confidence in this statement.
Everything is reducable to mathematics. Baseball is reducable to mathematics.
True, but going from theory to application is a lot harder in baseball.
Now here is an scenario that I can run with. I am a baseball instructor, trained not only through a lifetime of playing and coaching, but also by major league ballplayers. I know of many cases where an extremely talented player did not make the major leagues due to circumstance. I have yet to meet the major league player who does not have one or more stories about a player who was much better than them in high school, college or minor leagues but never made it to the majors. Baseball takes at least as much luck and being in the right place at the right time as it does talent.
To assume that “we” have identified the best at anything, from potentially billions of choices, to me seems ludicrous.
And you would be wrong. The most famous example I can think of is golf’s Bobby Jones. He could have made a ton of money and set many more records had he turned pro, but he had no desire to do so. Does that make him a less skilled player than Tiger Woods? Nope. Bobby played the tournaments he wanted to play, regardless of the competition, solely because he loved the game.
I have personally bowled in a league with a guy who could have taken the PBA by storm. He scored 300 with regularity. Why didn’y he go pro? He couldn’t afford the entry fees and no one would sponsor him because he was a loose cannon with a drinking problem. How many more are out there like him?
I played a little in college and coached 2 years of high school baseball, myself. I’m very familiar with the small town kid who doesn’t get the kind of recognition that leads to big time scholarship offers and MLB draft noise. But, we’re not talking about “an extremely talented player.” We’re talking about the best on the planet. I simply can’t believe that the best baseball player in the world is washed up in some slow pitch softball league right now because he never got his shot. That kind of talent can’t help but get discovered. Not MLB-level talent, but all-time-great talent.
You went from good golfers with no desire to tour to Bobby Jones. Bobby Jones wasn’t hacking away in an after work beer league. He played in PGA tournaments against the best. He played in 20 major tournaments.
Yes, completely solved. Tablebases can tell you not only if a position is a win or a draw, but also:
- if it’s a win, precisely how many moves until checkmate
- if it’s a draw, which moves draw and which lose.
I’ve used Tablebases to analyse endings and it’s been helpful.
For example:
White: King g5, Qb7, ph5
Black: King g8, rook f6, pg7
is a draw, whereas
White: King g5, Qb7, pf5
Black: King g8, rook f6, pg7
is a win.
I agree that 32 man tablebases are impossible because of the current problems you point out.
And yes it’s true that they ‘have already done the processing’.
But I wanted to show that tablebases do exist for endings and that using them involves no analysis.
As a chess teacher, I tell my students not to set traps.
Play good moves and your reward will come.
Playing for traps is a bad habit!
Of course.
I’m just saying that a hypothetical player who honed her skills only playing chess computers would get tripped up when they played real games and suddenly face many traps. These traps wouldn’t have to be sound to trip her up; her unfamiliarity would be a huge disadvantage.
The “traps” might even be accidental, from a fallible human who believes them to be sound moves.
Not the playing of baseball. Playing chess is.
Sure, whatever, you win. No player of any sport, no artist, no singer that has ever lived was better than what we identify as “the best”. Nevermind that we can’t even objectively identify “the best” at anything among those we do know.
Sorry, missed this post:
I think there is a critical difference between someone telling you you missed this or that variation, or have the following weakness in your play, and performing brain surgery on you to change your style of play. Especially mid-game.
The thing to appreciate is that it had been possible to make chess programs that beat grandmasters for some time, but such programs still needed ways to reduce the search space, which would manifest as “styles” of play to human opponents. The result being a program that consistently beat top players would get trounced once people noticed such patterns in its play.
So modifying deep blue in-game, I would consider unfair, even if I wouldn’t call it “cheating” (Kasparov did agree to those terms).
And of course it’s largely moot nowadays. Not one disputes the best chess computers running on dedicated PCs (and soon we won’t even need to say that) can beat any human.