Komodo vs. Movsesian

[lkaufman]

Human are smarter than computer, so 660 elo can be more vs human. I already said that a 2700 human can easy draw and probably win vs the f7 pawn handicap.

This time the human is high rated, and has time to get prepare. So I except him to win at least one game at this handicap, and draw every others games.

My final result was 10 wins, 7 draws, and one loss for White, which is plus 191 elo, which would make the handicap about 640 elo. But when I looked at the games, the situation is even worse. In three of the games White gambited back the e4 pawn on move three to gain develoment, but only scored one out of 3 in those games. No human would gambit a pawn like this against a much stronger player, but the engine didn't know it was playing a stronger player. So excluding those games the score was 12.5 to 2.5, which is about a 280 elo gap, which would make the handicap 730 elo, without even considering the factor you mention. I'm running the exchange handicap the same way tonite. It will probably turn out to be a bit less, probably not much less though. So at this point I'll be happy with a drawn match.

[Jesse Gersenson]

http://www.chess.com/article/view/the-romance-of-chess

A volume, entitled, "A Selection of Fifty Games, from those played by the Automaton Chess Player during its Exhibition in London in 1820. Taken down by permission of M. Maelzel, at the time they were played," was published about this time, from which we learn, that of nearly three hundred games played by the Automaton Chess Player (giving the Pawn and Move), it lost only six.

Well, you can bet Komodo won't lose that many.

There's some irony in that I'll be sitting at the board and the automaton will be feeding me moves.

[lkaufman]

My simulation results at Exchange odds are not quite as bad as at f7 handicap. So far, again with the winning side (in this case Black) giving 30' + 30" to 1' + 1" time odds on quad, Black leads by three wins to one, with eight draws, which is 58 elo. So, in round numbers, it looks like Exchange odds is around 500 elo and f7 odds around 650 elo (or more if you discount the strange games where White returned the pawn on move three). Since there are twice as many f7 games, the weighted average of the handicap values for the match is at least 600, so if we call Movsesian 2700 we are in effect playing an opponent of at least 3300 strength.

[Jesse Gersenson]

My simulation results at Exchange odds are not quite as bad as at f7 handicap. So far, again with the winning side (in this case Black) giving 30' + 30" to 1' + 1" time odds on quad, Black leads by three wins to one, with eight draws, which is 58 elo. So, in round numbers, it looks like Exchange odds is around 500 elo and f7 odds around 650 elo (or more if you discount the strange games where White returned the pawn on move three). Since there are twice as many f7 games, the weighted average of the handicap values for the match is at least 600, so if we call Movsesian 2700 we are in effect playing an opponent of at least 3300 strength.

The assessment of f7 has gone from 450 to 650 in the past two weeks. Whatever the projections say, history tells us that giving pawn and move to a player as strong as Movsesian is absurd. This classical handicap is reserved for the strongest players in the world playing patzers. But there are clear examples of what happens when White is not a patzer.

Pillsbury-Steinitz, 1892 is a good example. Pillsbury, who'd started playing chess just 4 years earlier, beat Steinitz 2-1 at f7 odds. These odds are not for the faint of heart.

Want another example? Milov beat Rybka 3 by a score of 1,5 to 0,5.

Komodo recently beat a GM at these odds. But, for the record, that was the GM's first time in his life playing f7 odds.

[Ozymandias]

Pillsbury-Steinitz, 1892 is a good example. Pillsbury, who'd started playing chess just 4 years earlier, beat Steinitz 2-1 at f7 odds. These odds are not for the faint of heart.

In defense of Steinitz, it should be noted that Pillsbury is part, with Fischer and Morphy, of that strange american chess dynasty.

[lkaufman]

My simulation results at Exchange odds are not quite as bad as at f7 handicap. So far, again with the winning side (in this case Black) giving 30' + 30" to 1' + 1" time odds on quad, Black leads by three wins to one, with eight draws, which is 58 elo. So, in round numbers, it looks like Exchange odds is around 500 elo and f7 odds around 650 elo (or more if you discount the strange games where White returned the pawn on move three). Since there are twice as many f7 games, the weighted average of the handicap values for the match is at least 600, so if we call Movsesian 2700 we are in effect playing an opponent of at least 3300 strength.

The assessment of f7 has gone from 450 to 650 in the past two weeks. Whatever the projections say, history tells us that giving pawn and move to a player as strong as Movsesian is absurd. This classical handicap is reserved for the strongest players in the world playing patzers. But there are clear examples of what happens when White is not a patzer.

Pillsbury-Steinitz, 1892 is a good example. Pillsbury, who'd started playing chess just 4 years earlier, beat Steinitz 2-1 at f7 odds. These odds are not for the faint of heart.

Want another example? Milov beat Rybka 3 by a score of 1,5 to 0,5.

Komodo recently beat a GM at these odds. But, for the record, that was the GM's first time in his life playing f7 odds.

The most important historical match at f7 handicap, in my opinion, was De Vere vs. Steinitz, 1865. De Vere playing White beat Steinitz by 8 to 4 (seven wins, three losses, two draws). The importance of this match was that two years later De Vere had risen to World number 2 according to the Jeff Sonas historical rating list. However at the time of the f7 match De Vere had no known tournament history, and was just 19 or 20 years old, which in those days meant he had very little experience with strong players. So it doesn't really tell us much about the elo value of the handicap. I looked over the games for ideas for our match, but many of Steinitz's opening choices are easily refuted with the aid of computer analysis. Black has a very limited number of ways to defend that don't clearly lose too easily.
All in all I expect we can break even at Exchange odds, but breaking even at f7 will be a challenge.

[mjlef]

The match is live now here on our site:

http://u.komodochess.com/games/Live.html

and live on chess.com:

http://www.chess.com/tv

[JJJ]

First game easy draw. That's not surprising. The next one is gonna be a lot tougher in my opinion for the GM.

[lkaufman]

First game easy draw. That's not surprising. The next one is gonna be a lot tougher in my opinion for the GM.

Second game also drawn (exchange odds). Due to delays caused by technical difficulties game three postponed until tomorrow at 9 am eastern time, with all four games to be played that day.

[lkaufman]

We discovered that the problem was that cleaning people pulled out the internet cable when I was in another room, and it wasn't obvious. Won't be a problem tomorrow.