Upgrade to Chess.com Premium!

GM Nakamura Tries to Prove He is No Stockfish

  • FM MikeKlein
  • on 8/19/14, 2:35 PM.

"Playing like a machine" has become a compliment lately in the chess world (if not an accusation!). This weekend, a top grandmaster will make it literal.

On August 23, world number five GM Hikaru Nakamura will play four games against Stockfish, the current top chess engine. The caveat? He gets the assistance of a weaker program and will use an older computer.

The match will be in Burlingame, California in the Bay Area. You can see the details of the match on the Facebook events page.

The event will also be covered live on Chess.com/TV with host IM Danny Rensch -- coverage starts at 1 p.m. Eastern (GMT -5), 10 a.m. Pacific (exact game times to be determined). Nakamura has agreed to be interviewed following the match.

UPDATE: IM Rensch will host coverage of the first two rounds of the match, before piece odds for the last two games. GM Nakamura will join IM Rensch & company after the first two games before his break.

Nakamura will "team up" with an older version of Rybka (rated 3050) running on a 2008 MacBook Pro to take on the latest version of Stockfish (rated 3290), which in May won the TCEC (chess programming) title. 

In the first iteration of the concept earlier this summer, GM Daniel Naroditsky didn't fare too well. The match rules have changed slightly since that match. Games one and two will be man and machine versus machine, while games three and four will be without computer assistance. In those latter two contests, Nakamura will take the white pieces and pawn odds.

GM Hikaru Nakamura, getting in the "cyborg" mood.

The games will be 45 minutes per side with a 30-second increment per move, and Stockfish will not have access to an opening book.

"It started as pure entertainment and curiosity," said Tyson Mao, the co-organizer of the contests along with Jesse Levinson. "Though I feel like as I have these conversations with people, I've learned more, and this event I think can serve a larger purpose."

He said that one main contribution is to show how far computers have come in the last six years.

"This knowledge will only help humans improve their chess game," said Mao. "Daniel Naroditsky wasn't aware of Stockfish when he showed up on the day of the match. After the first game, his words were, 'when I get home, Komodo out, Stockfish in.'" 

Naroditsky was about to shell out some money for the program that conquered him when Mao informed him that it is free. Stockfish is also open source, meaning everyone has the chance to improve its code.

Here's Stockfish outlasting a modern version of Rybka as this year's TCEC. The timing of the resignation might look mysterious, but humans are allowed to intervene if the evaluation tips to a certain level.

The traditional man-versus-machine match is all but extinct (GMs Vladimir Kramnik and Michael Adams were both beaten in the mid-2000s), but recently matches have tried innovative strategies. This round-up from 2008 shows that humans, even strong GMs, have needed many handicaps to have a chance of beating their silicon counterparts (and even then they usually lose).

If Mao is right and computers have advanced significantly, games three and four will prove quite challenging for Nakamura. Other GMs like Jaan Ehlvest couldn't win a brief match with a pawn and Black every game, while GM Joel Benjamin narrowly lost with pawn and alternating colors. Both of those matches were in 2007 -- a lifetime ago in computer terms.

Nakamura's history with computers is nothing if not humorous. This famous internet blitz game from 2008 versus Rybka seemed to exploit some issues in the programming code.

That's not the only reason is it remarkable. Here are the game's other oddities:

  • Rybka "handicapped" itself by playing the Grob.
  • Nakamura sacrificed two exchanges.
  • Nakamura ended up with six bishops at one point.
  • And he played 271 moves in 3 minutes!

Whatever the circumstances surrounding the game, Nakamura certainly understands that keeping the position closed is the way to go! The contest surely could fit in with GM Gregory Serper's recent column on "The Art of Doing Nothing."

Mao cited this game and said it made him "feel optimistic that he's the right person to figure out how to break the machine." But after the Naroditsky match (which ended 3.5-0.5 for Stockfish), Mao's prediction has become more cautious.

"I'm not optimistic about GM Nakamura's chances," Mao said. "When I tell a lot of people about this, they don't really believe me. Everyone wants to believe humanity still has a chance."

29566 reads 111 comments
20 votes

Comments


  • 2 months ago

    mcris

    Stockfish is strongest, then Komodo, afterwards Houdini.

  • 2 months ago

    D_Ostwald

    I think you played quite well too Naka!  Good show and good effort on your part.  Was fun to see you "manipulate" it as you did ... I suspect there are not many (if any) others who could do that as well as you did.  Unfortunate on Game #2 ... but gosh, it could have gone on for hours longer waiting for Stockfish to make the mistake of "pushing" a pawn.  Any how, thanks for the excitement ... perhaps a rematch?  Cheers.  :)

  • 2 months ago

    GMTkdriverx

    I did a search on a random game with Stockfish and it was able to get 22+ depth.

  • 2 months ago

    savantz

    SO TRUE... MinimusMax! that's been my experience too with rybka

  • 2 months ago

    MinimusMax

    RE: Rybka's depth, Rybka intentionally obfuscates depth and search speed, reporting them as much lower than they actually are.

    It was a silly thing that was done to mislead everyone when Rybka was released, so they would think that due its low reported depth and speed, that it was much more knowledgeable or smarter, instead of being very fast and efficient, which it actually was.

    Getting over depth 20 with Rybka takes VERY powerful hardware or very long search times.

    At TCEC, played on a monster 16 core machine with a lot of time (G/120 +30), Rybka barely averaged over depth 20.

    You just can't expect to get high reported depths from Rybka. It's not a problem with Rybka; it's actually searching to depths on par with other engines, but it's lying about it with the displayed values.

  • 2 months ago

    EvgeniyZh

  • 2 months ago

    ChessMN16

    Well played Nakamura - one of the most exciting matches I've watched in a long time. I hope chess.com will do more of these special matches.

     

    I wonder if it's even possible to beat this machine...In the pawns odd matches, a slight inaccuracy will lead to a draw; in the Rybka matches, Stockfish just defends superbly and there's no way to take it into a position it doesn't understand. 

  • 2 months ago

    savantz

    you played extremely well; and don't let the "haters" tell you otherwise!

    best wishes in the sinquefield cup 👍

  • 2 months ago

    GM Hikaru

    Ok, a few clarifications. First off, in game 2, when I plugged in some moves instead of f5 like say Rh8 Re3 Rh6, rybka said g4 was +/- 0.90. Whereas on f5, it was saying += 0.3 This is why I chose to break open the position. Of course, rybka simply did not have the horizon or depth to see deeper unlike houdini. 

    Game 4 was simply me trying to even up the match at all costs. At the end of the day, I think I played quite well. The only issue I have is how Rybka seemed to take about 1 minute to reach depth =16 and NEVER reached depth =20 no matter how long I let it sit along with a failure to hash the position from the variations. 

  • 2 months ago

    savantz

    Naka held draws in every game; every game... but decided in games 2 and 4 ~

    "what the hell, let's give 'em a show". eleven (11) straight hours he matched wits with the silicon

    nice stuff from nakamura!

  • 2 months ago

    novv

    What happened in game 2? Naka seemed to have a draw in hand when I stopped watching.

  • 2 months ago

    EvgeniyZh

    pgn someone?

  • 2 months ago

    Jukes

    Game 1: Draw

    Game 2: Stockfish wins 

    Game 3: Draw

    Game 4: Stockfish wins

    3-1 to Stockfish

  • 2 months ago

    matematikisto

    the Rybka vs Nakamura. 61...Qxe6 - 111.Bxe4 and 124...Nxf5  -  174.c4 they almost reach the 50 moves rule. Hmm so Rybka refused to draw two times!!!

  • 2 months ago

    mnhsr

    lol

  • 2 months ago

    2WatchTV

    3-1 to Stockfish

  • 2 months ago

    Gitananda

    Here is the up-to-date chess engine rating list: http://www.computerchess.org.uk/ccrl/4040/

  • 2 months ago

    awesomechess1729

    Why did chess.com/tv stop streaming the game?

  • 2 months ago

    exiledwanderer

    what were the results...i can't find them and missed everything after the first game.

  • 2 months ago

    geminicricket

    I attempted to download and install stockfish last week.  No success.

Back to Top

Post your reply: