4 ms·
The Real Kings of Chess Are Computers
- asdfologist 12y agoA major point missed in the article is that computers aren't just stronger than humans but also play an extremely monumental role in human chess itself. The entire opening book played by top grandmasters is the result of thousands (millions?) of hours of computer analysis, and many top-level games these days, especially the infamous ~20-move "grandmaster draws" are little more than one grandmaster's home-prepped computer engine analysis being pitted against the other. In other words, many "human" games are effectively a game between two computers :)
- pk2200 12y agoChess opening theory is the result of ~150 years of mostly human effort. Grandmasters do use computers to assist with opening preparation, but it's certainly not the case that computers have completely rewritten the opening books. In fact, I can't think of a single example of a computer significantly altering the evaluation of a major opening system. They do make small improvements, but almost all of the major openings that were popular 30 years ago are still popular today.
- blazespin 12y agoGreat article, http://www.newyorker.com/business/currency/are-computers-making-society-more-unequal http://www.newyorker.com/business/currency/are-computers-mak... which talks about human/computer chess competitions. So rather than computer versus computer or human versus human, it's a hybrid. What I'd be curious about though, is there ever a case where the strongest computer program + a human could beat the strongest computer program?
- asdfologist 12y agoYou're referring to Advanced Chess [0]. Yes, humans do add value, or otherwise this form of chess wouldn't exist at all, i.e. the human's best strategy would be to always take the engine's top recommended move. [0] http://en.wikipedia.org/wiki/Advanced_Chess http://en.wikipedia.org/wiki/Advanced_Chess
- gwern 12y agoBut at some point Advanced Chess will be 'dead' in the sense that the best human players no longer help the best software win: the human will pick better moves less than poorer moves, and blow games. It's hard to tell when, but given how large the margin is, and how finely balanced the best chess engines are now (Lipton has some interesting posts on computer chess, most recently https://rjlipton.wordpress.com/2014/12/28/the-new-chess-world-champion/ https://rjlipton.wordpress.com/2014/12/28/the-new-chess-worl... ), I wonder if that moment has already passed?
- Houshalter 12y agoAnother observation is that humans only add value for their pattern matching ability. Computers have worse heuristics, but make up for it by being able to explore many moves into the future. But now deep neural networks have been catching on. They've shown good results on Go, being able to predict the move an expert would make 44% of the time.
- stolio 12y agoNakamura's been the main GM willing to at least publicly enter that arena. See 2008's Rybka vs. Nakamura [0] for some background and the more recent Rybka+Nakamura vs. Stockfish[1]. In his computer enhanced form he's affectionately referred to as "Rybkamura" [0] - http://www.chess.com/forum/view/general/nakamura-defeats-rybka-in-a-loooong-game http://www.chess.com/forum/view/general/nakamura-defeats-ryb... [1] - http://www.chess.com/news/stockfish-outlasts-nakamura-3634 http://www.chess.com/news/stockfish-outlasts-nakamura-3634
- raymondh 12y agoThe article makes a surprising and unattributed assertion, "Mr. Kasparov’s obsession with Deep Blue’s surprising strategic insight was partly responsible for his loss in their 1997 match. Today, smartphones play better than IBM ’s supercomputer did then." Does anyone have a link or attribution for this? I would be gobsmacked to learn that 2015 smartphones play better chess than Deep Blue in 1997. Could this possibly be true?
- jrpt 12y agoI was curious too so I Googled it. Looks like it's true: http://www.phonearena.com/news/A-modern-smartphone-or-a-vintage-supercomputer-which-is-more-powerful_id57149 http://www.phonearena.com/news/A-modern-smartphone-or-a-vint... You'd still need the right chess AI running on it but smartphones today are more powerful than Deep Blue, apparently.
- jules 12y agoSince 1997 we've had 12 Moore's law doublings in transistor count (4096x), and chess algorithms have improved quite a bit. http://www.quora.com/How-does-Deep-Blues-playing-strength-and-algorithms-compare-to-modern-chess-engines-eg-Houdini http://www.quora.com/How-does-Deep-Blues-playing-strength-an... http://en.chessbase.com/post/komodo-8-the-smartphone-vs-desktop-challenge http://en.chessbase.com/post/komodo-8-the-smartphone-vs-desk...
- ChrisLomont 12y agoThis link [1] claims Komodo 8 on an android phone from 2013 should be around 3000 strength, far stronger than Deep Blue. [1] http://en.chessbase.com/post/komodo-8-the-smartphone-vs-desktop-challenge http://en.chessbase.com/post/komodo-8-the-smartphone-vs-desk...
- Athas 12y agoYou have to remember that Deep Blue was not a top-of-the-line supercomputer in 1997. It was a large machine for sure, but the software made it a chess champion, not just the raw hardware. Many 1997-era computers were faster.
- 12y ago
- kingkawn 12y agoA computer can win chess, not invent it.
- rpearl 12y agohttp://boardgamegeek.com/boardgame/33767/yavalath http://boardgamegeek.com/boardgame/33767/yavalath is a board game that is commercially available and was generated by a computer. It's got a decent rating, too. Higher than chess, in fact :)
- kingkawn 12y agohahaha i was wondering how long it would take for my statement to be wrong
- toolslive 12y agoDoes Usain Bolt feel bad about himself, because a Ferrari is faster on the 100m flat? Should he stop competing because of it? Does a weightlifter feel bad because a fork lift exists? Does this kill weightlifting as a competition? So why would this be the case for Chess ? Anyway, is it just me or does the article feel like having been written by someone who doesn't know about computer chess?
- Kenji 12y agoNot until the line between human and machine blurs (e.g. cyborg). At that point it stops making sense.
- tbrake 12y agoViswanathan Anand himself had a quote which reflects this mentality and is a good way of looking at it : "[...] but since most of chess is tactically based they do many things better than humans. And this imbalance remains. I no longer have any issues. It’s bit like asking an astronomer, does he mind that a telescope does all the work. He is used to it. It is just an incredible tool that you can use."
- asdfologist 12y agoYou're fighting a straw man. Nowhere does the article even remotely imply that human chess is under threat because computers are stronger.
- stolio 12y agoCan't find the link but I once saw Carlsen (or maybe Nakamura?) say a computer miscalculated the value of a pair of pawns in a position. Essentially the value of the pawns wouldn't become clear until much later in the game and the heuristic the computer attached to space could overvalue it in certain circumstances. More common than that is a top player might make a move that the computer disagrees with until it gets to a depth of 20+ moves when it suddenly thinks it's THE winning move. For a really interesting computer vs. computer game see Kingscrusher's commentary on Houdini vs. Rybka: https://www.youtube.com/watch?v=xWdMqvGMxF4 https://www.youtube.com/watch?v=xWdMqvGMxF4