Computer Chess / Handicap Chess
The Return of
Handicap Chess Everything old is new again
By GM LARRY KAUFMAN
between them with any chance of a human victory. Just as around 1860 Paul Morphy proved so dominant over his rivals that he was able to give “pawn and move” (the f7-pawn) handicap to some of the top European players and even knight odds (b1) to the third-ranked U.S. player and come out ahead, top engines can now give f7 odds to very strong grandmasters successfully, despite the vastly higher standard of play now. As part of the Rybka team in 2007 and 2008 when it was the top engine, I got the ball rolling with a series of handicap matches against grandmasters, international masters, and FIDE masters at fairly serious time limits. Rybka proved able to spot a pawn successfully to respected grandmasters, but came out behind giving the traditional f7 “pawn and move” handicap or the Exchange (rook for knight) to 2700- rated GM Vadim Milov. Rybka was totally unable to give knight odds to a FIDE master, losing all four games badly to John Meyer. After that I switched to working on the engine Komodo, which is
H
now generally considered to be the world’s strongest, having won the prestigious Thoresen Chess Engines Competition recently beating nearest rival Stockfish by 9 to 2 in the final (plus 89 draws!). In 2014 Stockfish scored one win and one draw against our top player Hikaru Nakamura giving him pawn odds, but these were the b7- and h7-pawns, much smaller handicaps than the f7-pawn. In 2015 we (Komodo) decided to start holding serious matches (at 45’ plus 15” increment for each side, with money prizes for wins and draws) at various handicaps against FIDE titled players, mostly grandmasters, culminating in a handicap match with U.S. Champion and World Title Candidate Hikaru Nakamura. These matches have provided interesting and entertaining games to the spectators, have helped to quantify just how much stronger Komodo is than top human players, and have helped us to improve Komodo. Some of these matches were held in Prague with our webmaster Jesse Gersenson operating, some were held at my home office, and some were live on
chess.com. Almost all were played with Komodo running on my 24-core computer, a $6,400 machine. Some of the games have been quite spectacular. As of this writing we have not lost a single game out of 36 official games in which we gave grandmasters handicaps such as pawn (f7) and move, pawn (f7) and two moves, rook for knight (both colors), and even two pawns other than f2, Komodo winning 17 games and drawing the other 19.We did have some losses when we got too generous, giving f7 and threemoves or two pawns including the “f” pawn. Let’s now talk about the various handicaps one by one. Traditionally,
andicap chess has returned with a vengeance lately, for the simple reason that computer programs are so much stronger than any human player that only handicap games can be played
the smallest handicap (other than just conceding the white pieces) was draw odds. Rybka did play one eight game match conceding draw odds and white to GM Joel Benjamin, beating him six to two. I think even World Champion Magnus Carlsen would need more than this handicap against Komodo. Pawn handicap traditionally meant the “f” pawn, because it is the only
pawn whose removal offers no compensation in the form of opened file or diagonal for rook, queen, or bishop. Instead, it weakens the king, so it is a larger handicap than just a pawn. Normally in handicap chess the odds-giver plays white and moves first, but in all recorded historical games involving the “f” pawn the odds giver played black, removing f7. Of course removing f2 as white is a valid handicap, but it was probably not used because it is not much larger than draw plus white odds, which was preferred as more like normal chess. With the f7-pawn missing Black has to grovel in the opening after 1. e4. The reply 1. ... e5?? loses immediately to 2. Qh5+, similarly the Sicilian is a dubious gambit, 1. ... g6 is weak due to 2. h4, and the Caro-Kann 1. ... c6 doesn’t work after 2. Nc3 since 2. ... d5? loses a pawn after 3. exd5 cxd5 4. Qh5+. So Black has to play the French 1. ... e6, the Pirc 1. ... d6, the Nimzowitsch 1. ... Nc6, or the bizarre 1. ... Nh6!? with the idea of going to f7. Analysis shows that White always should get a substantial positional advantage on top of his extra pawn, so I consider the f7 handicap to be a one and a half pawn handicap equivalent. Despite this, so far Komodo has played 10 games giving this handicap to grandmasters, including four with Sergei Movsesian (2658 FIDE and in the world top 10 in 2009), four with World Open Champion Alex Lenderman (FIDE 2628), and one with U.S. Champion Hikaru Nakamura (FIDE 2787) without defeat, scoring five wins and five draws. This was the most popular handicap in the recorded games of Paul Morphy, and also a Bobby Fischer favorite. World Champion Emanuel Lasker once defined a chess master as a player to whom no one could successfully give pawn and move odds, but by this definition even the above strong grandmasters would not qualify (unless you argue that Komodo is not “someone”). Probably Lasker would not qualify either! I think that this shows that chess is a much deeper game than generally thought a century ago. Nakamura managed a draw in a difficult position by a nice piece sacrifice, and also drew with black at f2 odds. We also played five games giving grandmasters “pawn and two moves”,
another popular handicap from the 1800s in which the f7-pawn is removed, and White gets to make two moves (should be 1. e4 and 2. d4) before Black gets to move. Komodo beat Simon Williams (FIDE 2439), drew and beat Victor Mikhalevski (FIDE 2528), and drew twice
www.uschess.org 31
PHOTO: DAVID LLADA
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46 |
Page 47 |
Page 48 |
Page 49 |
Page 50 |
Page 51 |
Page 52 |
Page 53 |
Page 54 |
Page 55 |
Page 56 |
Page 57 |
Page 58 |
Page 59 |
Page 60 |
Page 61 |
Page 62 |
Page 63 |
Page 64 |
Page 65 |
Page 66 |
Page 67 |
Page 68 |
Page 69 |
Page 70 |
Page 71 |
Page 72 |
Page 73 |
Page 74 |
Page 75 |
Page 76