AI Ruined Chess. Now, It's Making the Game Beautiful Again (wired.com) 38
Chess has a reputation for cold logic, but Vladimir Kramnik loves the game for its beauty. "It's a kind of creation," he says. His passion for the artistry of minds clashing over the board, trading complex but elegant provocations and counters, helped him dethrone Garry Kasparov in 2000 and spend several years as world champion. Yet Kramnik, who retired from competitive chess last year, also believes his beloved game has grown less creative. From a report: He partly blames computers, whose soulless calculations have produced a vast library of openings and defenses that top-flight players know by rote. âoeFor quite a number of games on the highest level, half of the game -- sometimes a full game -- is played out of memory," Kramnik says. "You don't even play your own preparation; you play your computer's preparation." Wednesday, Kramnik presented some ideas for how to restore some of the human art to chess, with help from a counterintuitive source -- the world's most powerful chess computer. He teamed up with Alphabet artificial intelligence lab DeepMind, whose researchers challenged their superhuman game-playing software AlphaZero to learn nine variants of chess chosen to jolt players into creative new patterns.
In 2017, AlphaZero showed it could teach itself to roundly beat the best computer players at either chess, Go, or the Japanese game Shogi. Kramnik says its latest results reveal beguiling new vistas of chess to be explored, if people are willing to adopt some small changes to the established rules. The project also showcased a more collaborative mode for the relationship between chess players and machines. "Chess engines were initially built to play against humans with the goal of defeating them," says Nenad Tomasev, a DeepMind researcher who worked on the project. "Now we see a system like AlphaZero used for creative exploration in tandem with humans rather than opposed to them." People have played chess for around 1,500 years, and tweaks to the rules aren't new. Nor are grumbles that computers have made the game boring.
In 2017, AlphaZero showed it could teach itself to roundly beat the best computer players at either chess, Go, or the Japanese game Shogi. Kramnik says its latest results reveal beguiling new vistas of chess to be explored, if people are willing to adopt some small changes to the established rules. The project also showcased a more collaborative mode for the relationship between chess players and machines. "Chess engines were initially built to play against humans with the goal of defeating them," says Nenad Tomasev, a DeepMind researcher who worked on the project. "Now we see a system like AlphaZero used for creative exploration in tandem with humans rather than opposed to them." People have played chess for around 1,500 years, and tweaks to the rules aren't new. Nor are grumbles that computers have made the game boring.
Mind Boggling (Score:3)
Re: Mind Boggling (Score:3)
Re: (Score:2)
It could be the strategies from then are fine but he just has dementia after 600 years.
Re: (Score:2)
Uuuuh, half the 2200 rateds at a tournament remember every game they ever played. They don't have a memorization step.
I'm only 2100 and I don't remember my games like that, but I certainly remember lots of variations in the openings I play.
Re: (Score:2)
I'm a 1400-1500
If you are Elo 1450, you're definitely not playing in the same league, by far. It's like comparing algebraic geometry and 2 x 7 calculus. Totally incomparable.
Re: (Score:2)
I think the game is still useful for certain people to help them develop their minds potential. I think competition is still useful in that people always like to believe that they are better at something than anyone else. But that skill at c
Re: (Score:1)
house rules (Score:1)
Re: (Score:3)
Yep.
Kramnik says its latest results reveal beguiling new vistas of chess to be explored, if people are willing to adopt some small changes to the established rules.
I really don't think that you need AI to make house rules to change the game so people have fun playing again. The entire point of chess is that the rules are black and white and everyone knows them.
You just made a new game. Congratulations! You didn't make chess beautiful again, because it's not chess. And if you sat down with anyone around the world who knows how to play chess, they wouldn't know how to play your game.
And how long until they throw AI at this new ruleset, and produce the same problem o
Re: (Score:2)
"I really don't think that you need AI to make house rules to change the game so people have fun playing again. The entire point of chess is that the rules are black and white and everyone knows them.
You just made a new game. "
So you're still old style, no "en passant" and no "two-square first move for pawns" since that was the last rule change, I guess you didn't like that one either, those middle-ages morons ruined the game.
"Fairy Chess" (Score:2)
You just made a new game. Congratulations! You didn't make chess beautiful again, because it's not chess.
I once heard the term "Fairy Chess", defined as: The set of all games played on a chessboard with chess pieces which are not chess.
Re: (Score:2)
Chess is full of rules tweaks. Travel back a few hundred years, and you'll get very angry reactions if you try to castle, move a pawn two squares on it's first move, capture en passant, move your queen more than one square at a time....
What can't they do? (Score:3)
What if computers could be used to translate character sets so we could avoid this:
?
Too bad that is impossible.
Re: (Score:2)
Bigger Chess/Go board (Score:2)
I suspect a lower size board favours the computer, while a higher size board favours the human.
Maybe if we moved from 8x8 to 16x16 in chess, humans would fare more favourably against computers again. Likewise with Go, moving from 19x19 to say, 29x29 board size, humans could start winning again even over the best Go computers?
Re: (Score:2)
It's less about the size of the board that the number of moves, although the two are loosely related. From the opening position of chess, you have a choice of twenty moves. From the opening position of go on a 19x19 board, you have a choice of 361 moves. Even if you prune out moves that are rotations of other moves (essentially the same because of the symmetry of the board) you still have 100 moves. Add into that the fact that it is harder to build a set of rigid rules to evaluate a go position than a c
Re:Bigger Chess/Go board (Score:4, Interesting)
I suspect a lower size board favours the computer, while a higher size board favours the human.
That would probably be true for brute-force engines like Stockfish but less so for neural network engines like AC0 or LC0, as those supposedly view the chess board more positionally.
Re: (Score:2)
But the size also dictates a lot of the customs or rules of thumbs. Changing the size (or anything about the game) will take decades for people to work out the new approaches.
I say any change heavily favors the machines.
Re: (Score:2)
In summary... (Score:5, Informative)
Re: (Score:3)
Saved me a click, thanks.
Re: (Score:2)
It may make things more fun for GM and above levels who may be bored because they already know all the openings and responses to them, but chess is still pretty interesting for the 99.9% of the players.
I do admit, self-sacrificing your own pieces does have an allure, though.
Golden age of Chess was early 20th Century (Score:4, Informative)
The early 20th century was the Golden Age of Chess IMHO. There were some epic chess matches. Sorted by last name:
* Alexander Alekhine (1892-1946)
* Jose Capablanca (1888-1942)
* Emanuel Lasker (1868-1941)
* Paul Morphy (1837-1884)
* Samuel Reshevsky (1911-1992)
* Wilhelm Steinitz (1836-1900)
Why? Because the outcomes of openings were still unknown. Creativity was explored. Announcers were baffled. Gambits sometimes paid off. Sometimes they cost dearly. By the time Bobby Fischer and Garry Kasparov made chess popular Chess was already boring. If you hadn't memorized hundreds of openings you were completely screwed.
Speed chess retains some of the feel of early Chess.
"House Rules" have always been around -- but I've never seen any variations have lasting power. I highly doubt "AI" is going to swing the balance of making Chess THAT exciting again.
Personally I always found Go to be more more interesting and rewarding. No "book openings" to memorize. Creativity has bigger payoffs due to the significantly larger board.
In a lot of ways codern computer RTSs (Real-Time-Strategy) games mirrored chess. New Rock/Paper/Scissors being explored and developed. A new meta as units are constantly getting re-balanced.
I see that Chess has surpassed Starcraft 2 on Twitch:
* Chess [twitch.tv] -- has 21K viewers, 924K followers.
* Starcraft 2 [twitch.tv] -- has 16.1K viewers, 2.1M followers
* Age of Empires II: Definitive Edition [twitch.tv] -- has 5.7K viewers, 269K followers. No doubt T90 Official [youtube.com] has played a part in that with 217K subscribers on his YouTube channel alone.
Yes, Chess is seeing an uptake in streaming due to Covid -- but the question is, will it remain popular after Covid is over? Will chess remain popular next year? In 5 years? I doubt it.
Re: (Score:3)
Chess is a simple game. It means it can be solved. That is, you can map out the most optimal move from every board state that will lead to the most advantageous ending. Of course, if both players do it, you're looking at a draw.
Some games are trivial - you can map tic-tac-toe in a few minutes quite easily, for example. Chess is much harder because the board state possibilities explode exponentially, but they are finite, but currently quite large that we can't map it all practically. Go has an even larger bo
I'd say "discovered", not "produced" (Score:4, Informative)
... computers, whose soulless calculations have produced a vast library of openings and defenses ...
Yes, the computers produced the vast library. But IMHO they discovered its content.
The openings and defences were "always there", in the universe of possible games defined by the ruleset. What the computers did was discover and evaluate them, filling the library with good ones.
Why not test the popular chess variants? (Score:2)
Re: (Score:2)
Posting instead of upmodding. Yes to 960!
New variants... (Score:2)
...would just favor those who memorize new computer-searched opening books.
Re: (Score:1)
Not if you combine them with Chess960. Shameless plug:
https://github.com/agt-the-wal... [github.com].
Wait, when was it not? (Score:2)
Chess has always been beautiful to those that have a passion for playing the game.
That it might have become less interesting to AI researchers does not change that.
Re: (Score:2)
I think it is hilarious that they think chess players feel their pain.
Can't speak with regards to serious players (Score:2)
But AI hasn't affected the game for casual players at all.
Chess is simple (Score:2)
When we have the family around the table we routinely play games more complicated than chess. Many of them have random elements too. Advanced Third Reich (not a family game) is well beyond the complexity of Chess, Shogi, or Go. And a computer that can play Diplomacy is in the realms of fantasy.
Chess isn't that hard if, like Google, you can afford to just throw a huge database at it. Really huge.