Mikhalevski vs Komodo match

Discussion of computer chess matches and engine tournaments.

Moderators: bob, hgm, Harvey Williamson

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
lkaufman
Posts: 3733
Joined: Sun Jan 10, 2010 5:15 am
Location: Maryland USA
Contact:

Mikhalevski vs Komodo match

Post by lkaufman » Fri Jan 01, 2016 8:39 pm

Our handicap match with GM Victor Mikhalevski (FIDE 2528, FIDE Rapid 2590, age 43) starts tomorrow, Jan. 2. It is unfortunate that this is so close in time to our match with Nakamura starting Jan.6, but the Mikhalevski match was scheduled before the Nakamura match was rescheduled. It will be a six game match at the usual time limit, 45' + 15" increment. Handicaps same as Hess match, four games at pawn (f7) and two moves, two games at Exchange (a1 rook for b8 knight, place a8 rook on b8) handicap. As usual, Komodo will run on my 24 core Intel machine, and he will play live on a regular chessboard at my home.
broadcast:
http://u.komodochess.com/games/Live.html
live pgn:
http://u.komodochess.com/games/live483.pgn

Tentative schedule: Jan 2,3,and 4 at noon and 5 pm Eastern Standard time. If any details change I'll post here.
Komodo rules!

whereagles
Posts: 561
Joined: Thu Nov 13, 2014 11:03 am

Re: Mikhalevski vs Komodo match

Post by whereagles » Sat Jan 02, 2016 12:12 am

Great news.. looking forward to follow it live :)

APassionForCriminalJustic
Posts: 415
Joined: Sat May 24, 2014 7:16 am

Re: Mikhalevski vs Komodo match

Post by APassionForCriminalJustic » Sat Jan 02, 2016 4:23 am

lkaufman wrote:Our handicap match with GM Victor Mikhalevski (FIDE 2528, FIDE Rapid 2590, age 43) starts tomorrow, Jan. 2. It is unfortunate that this is so close in time to our match with Nakamura starting Jan.6, but the Mikhalevski match was scheduled before the Nakamura match was rescheduled. It will be a six game match at the usual time limit, 45' + 15" increment. Handicaps same as Hess match, four games at pawn (f7) and two moves, two games at Exchange (a1 rook for b8 knight, place a8 rook on b8) handicap. As usual, Komodo will run on my 24 core Intel machine, and he will play live on a regular chessboard at my home.
broadcast:
http://u.komodochess.com/games/Live.html
live pgn:
http://u.komodochess.com/games/live483.pgn

Tentative schedule: Jan 2,3,and 4 at noon and 5 pm Eastern Standard time. If any details change I'll post here.
No disrespect to you Larry. Clearly you are a legend. I thought that I would just post this from an Internet friend of mine who is a strong Fide Master with one IM Norm:

"I heard about these Kaufman brainchild odds matches on chess.com, and even watched GM Williams play against Komodo live once. I agree it is pointless, it is well known that any top engine on a Pentium 4 (not just K93 on a v3 Dual Xeon), and even many second-tier ones could beat the best players consistently and rarely lose a single game to a human (if not for some engine glitches.) Plus, handicap chess isn't real chess it's more like handing you a ready made winning position and just asking you to finish it off. To me, it would be more interesting from a chess perspective to play these engines straight up, knowing that the result would be heavily in the computer's favor".

Do you have any real plans for a real chess series match? Of course even the Super GMs would get destroyed. But how badly?! For instance, could Anish Giri or Carlsen obtain a draw in a ten-game match?

lkaufman
Posts: 3733
Joined: Sun Jan 10, 2010 5:15 am
Location: Maryland USA
Contact:

Re: Mikhalevski vs Komodo match

Post by lkaufman » Sat Jan 02, 2016 6:01 am

APassionForCriminalJustic wrote:
lkaufman wrote:Our handicap match with GM Victor Mikhalevski (FIDE 2528, FIDE Rapid 2590, age 43) starts tomorrow, Jan. 2. It is unfortunate that this is so close in time to our match with Nakamura starting Jan.6, but the Mikhalevski match was scheduled before the Nakamura match was rescheduled. It will be a six game match at the usual time limit, 45' + 15" increment. Handicaps same as Hess match, four games at pawn (f7) and two moves, two games at Exchange (a1 rook for b8 knight, place a8 rook on b8) handicap. As usual, Komodo will run on my 24 core Intel machine, and he will play live on a regular chessboard at my home.
broadcast:
http://u.komodochess.com/games/Live.html
live pgn:
http://u.komodochess.com/games/live483.pgn

Tentative schedule: Jan 2,3,and 4 at noon and 5 pm Eastern Standard time. If any details change I'll post here.
No disrespect to you Larry. Clearly you are a legend. I thought that I would just post this from an Internet friend of mine who is a strong Fide Master with one IM Norm:

"I heard about these Kaufman brainchild odds matches on chess.com, and even watched GM Williams play against Komodo live once. I agree it is pointless, it is well known that any top engine on a Pentium 4 (not just K93 on a v3 Dual Xeon), and even many second-tier ones could beat the best players consistently and rarely lose a single game to a human (if not for some engine glitches.) Plus, handicap chess isn't real chess it's more like handing you a ready made winning position and just asking you to finish it off. To me, it would be more interesting from a chess perspective to play these engines straight up, knowing that the result would be heavily in the computer's favor".

Do you have any real plans for a real chess series match? Of course even the Super GMs would get destroyed. But how badly?! For instance, could Anish Giri or Carlsen obtain a draw in a ten-game match?
Well, on the first point, the very fact that it is accepted by most that any top engine even on an old machine can beat the World Champ in a match is the reason for playing the handicap matches; it is a way of measuring how wide the gap is. But yes, I would very much like to try a match with a top-five player in which he gets White and draw odds in every game, perhaps with no opening book for Komodo and big time odds as well, just to see how many draws he might be able to obtain. We would be willing to contribute to the financing of such a match, but we would need some other major sponsor to justify the expense. The practical problem is that it is not very interesting for chess.com or ICC or another such company to show a match in which the goal is to make draws. They prefer a format that makes for exciting games, which in general the handicap matches have done.
Komodo rules!

lkaufman
Posts: 3733
Joined: Sun Jan 10, 2010 5:15 am
Location: Maryland USA
Contact:

Re: Mikhalevski vs Komodo match

Post by lkaufman » Sat Jan 02, 2016 6:06 am

Two changes: First, the games will be at noon and 4:30 Eastern time each day, not 5:00 as previously stated. Second, two of the four games at f7 + two moves will be replaced by two games at two pawn odds (b2 + c2, c2 + d2) in the interest of maximum variety. Previous matches have shown that two pawns including a2 is too difficult for an ordinary GM, while two pawns including f2 is too easy, so these two pawn handicaps should be about right.
Komodo rules!

fenchel
Posts: 36
Joined: Thu Dec 04, 2014 5:01 am

Re: Mikhalevski vs Komodo match

Post by fenchel » Sat Jan 02, 2016 7:34 am

APassionForCriminalJustic wrote:
lkaufman wrote:Our handicap match with GM Victor Mikhalevski (FIDE 2528, FIDE Rapid 2590, age 43) starts tomorrow, Jan. 2. It is unfortunate that this is so close in time to our match with Nakamura starting Jan.6, but the Mikhalevski match was scheduled before the Nakamura match was rescheduled. It will be a six game match at the usual time limit, 45' + 15" increment. Handicaps same as Hess match, four games at pawn (f7) and two moves, two games at Exchange (a1 rook for b8 knight, place a8 rook on b8) handicap. As usual, Komodo will run on my 24 core Intel machine, and he will play live on a regular chessboard at my home.
broadcast:
http://u.komodochess.com/games/Live.html
live pgn:
http://u.komodochess.com/games/live483.pgn

Tentative schedule: Jan 2,3,and 4 at noon and 5 pm Eastern Standard time. If any details change I'll post here.
No disrespect to you Larry. Clearly you are a legend. I thought that I would just post this from an Internet friend of mine who is a strong Fide Master with one IM Norm:

"I heard about these Kaufman brainchild odds matches on chess.com, and even watched GM Williams play against Komodo live once. I agree it is pointless, it is well known that any top engine on a Pentium 4 (not just K93 on a v3 Dual Xeon), and even many second-tier ones could beat the best players consistently and rarely lose a single game to a human (if not for some engine glitches.) Plus, handicap chess isn't real chess it's more like handing you a ready made winning position and just asking you to finish it off. To me, it would be more interesting from a chess perspective to play these engines straight up, knowing that the result would be heavily in the computer's favor".

Do you have any real plans for a real chess series match? Of course even the Super GMs would get destroyed. But how badly?! For instance, could Anish Giri or Carlsen obtain a draw in a ten-game match?
Personally, I'm very happy these matches take place. What follows are my reasons. I am not negating your friend's comments; indeed, the ELO gap between humans and engines is irrelevant to me, my reasons are mostly orthogonal to that:

* translating ELO gaps into material.

* Making chess engines more useful to human play. As chess engines improve, their lines might become less useful since they are designed to refute play which a human will never follow. Taking this to the extreme, if chess were a draw and chess engines could identify this, it'd most likely be completely useless to human play.

* Getting a better handle on what material really means in chess. I.e., maybe there are positions where any human would currently tell you that it's hopeless without a serious opponent blunder, but these matches can say otherwise..

Jesse Gersenson
Posts: 575
Joined: Sat Aug 20, 2011 7:43 am
Contact:

Re: Mikhalevski vs Komodo match

Post by Jesse Gersenson » Sat Jan 02, 2016 8:24 am

lkaufman wrote:Our handicap match with GM Victor Mikhalevski (FIDE 2528, FIDE Rapid 2590, age 43) starts tomorrow, Jan. 2. It is unfortunate that this is so close in time to our match with Nakamura starting Jan.6, but the Mikhalevski match was scheduled before the Nakamura match was rescheduled. It will be a six game match at the usual time limit, 45' + 15" increment. Handicaps same as Hess match, four games at pawn (f7) and two moves, two games at Exchange (a1 rook for b8 knight, place a8 rook on b8) handicap. As usual, Komodo will run on my 24 core Intel machine, and he will play live on a regular chessboard at my home.
broadcast:
http://u.komodochess.com/games/Live.html
live pgn:
http://u.komodochess.com/games/live483.pgn

Tentative schedule: Jan 2,3,and 4 at noon and 5 pm Eastern Standard time. If any details change I'll post here.
In what order will the games be played?

Chess.com usually needs the fen's before hand so they can set up - list the fen's here.
  • Game 1
    Game 2
    Game 3
    Game 4
    Game 5
    Game 6
Has Komodo learned from the exchange odds games?

whereagles
Posts: 561
Joined: Thu Nov 13, 2014 11:03 am

Re: Mikhalevski vs Komodo match

Post by whereagles » Sat Jan 02, 2016 10:43 am

I would like to see some regular chess matches with time odds. Something like 45/15, as in 45 mins vs 15 secs :D

Still, I think these handicap matches can be quite useful to calibrate several eval function parameters.

lkaufman
Posts: 3733
Joined: Sun Jan 10, 2010 5:15 am
Location: Maryland USA
Contact:

Re: Mikhalevski vs Komodo match

Post by lkaufman » Sat Jan 02, 2016 3:44 pm

Jesse Gersenson wrote:
lkaufman wrote:Our handicap match with GM Victor Mikhalevski (FIDE 2528, FIDE Rapid 2590, age 43) starts tomorrow, Jan. 2. It is unfortunate that this is so close in time to our match with Nakamura starting Jan.6, but the Mikhalevski match was scheduled before the Nakamura match was rescheduled. It will be a six game match at the usual time limit, 45' + 15" increment. Handicaps same as Hess match, four games at pawn (f7) and two moves, two games at Exchange (a1 rook for b8 knight, place a8 rook on b8) handicap. As usual, Komodo will run on my 24 core Intel machine, and he will play live on a regular chessboard at my home.
broadcast:
http://u.komodochess.com/games/Live.html
live pgn:
http://u.komodochess.com/games/live483.pgn

Tentative schedule: Jan 2,3,and 4 at noon and 5 pm Eastern Standard time. If any details change I'll post here.
In what order will the games be played?

Chess.com usually needs the fen's before hand so they can set up - list the fen's here.

Game 1 rnbqkbnr/ppppp1pp/8/8/4P3/8/PPPP1PPP/RNBQKBNR w KQkq - 0 1
Game 2 1nbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/1RBQKBNR w Kk - 0 1
Game 3 rnbqkbnr/ppppp1pp/8/8/4P3/8/PPPP1PPP/RNBQKBNR w KQkq - 0 1
Game 4 1nbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/1RBQKBNR w Kk - 0 1
Game 5 rnbqkbnr/pppppppp/8/8/8/8/P2PPPPP/RNBQKBNR w KQkq - 0 1
Game 6[/list] rnbqkbnr/pppppppp/8/8/8/8/PP2PPPP/RNBQKBNR w KQkq - 0 1

Has Komodo learned from the exchange odds games?
I don't recall any specific change made due to those games.
Komodo rules!

User avatar
Leto
Posts: 2028
Joined: Thu May 04, 2006 1:40 am
Location: Dune

Re: Mikhalevski vs Komodo match

Post by Leto » Sat Jan 02, 2016 7:30 pm

whereagles wrote:I would like to see some regular chess matches with time odds. Something like 45/15, as in 45 mins vs 15 secs :D

Still, I think these handicap matches can be quite useful to calibrate several eval function parameters.
That's a bit too extreme but I like the theme. How about human gets 1 hour, machine gets 10 minutes.

Post Reply