Komodo vs GM Smerdon knight odds match

Discussion of anything and everything relating to chess playing software and machines.

Moderators: Harvey Williamson, Dann Corbit, hgm

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
Chessqueen
Posts: 1119
Joined: Wed Sep 05, 2018 12:16 am
Full name: Nancy M Pichardo

Re: Komodo vs GM Smerdon knight odds match

Post by Chessqueen » Thu Apr 16, 2020 4:11 pm

lkaufman wrote:
Thu Apr 16, 2020 4:23 am
Chessqueen wrote:
Thu Apr 16, 2020 3:19 am
lkaufman wrote:
Wed Apr 15, 2020 11:42 pm
jp wrote:
Wed Apr 15, 2020 10:51 pm
lkaufman wrote:
Wed Apr 15, 2020 5:45 pm
Clearly Komodo needs some compensation for the knight to have a decent chance against a GM or strong IM, either a pawn or no castling for the opponent, or chess 960 to defeat routine play.
Larry, what about just giving odds of three pawns, which gives Komodo much better chances, though in material terms it's similar? I don't know which pawns; maybe both rook pawns and queen pawn. If the GM makes it to the endgame alive, he can win, but before then, he risks getting mated. I'm not sure Black actually needs to castle (even practically) when he has an extra piece.
Three pawns is about the same as knight for pawn materially, since a knight is worth about 4 pawns in the opening. But removing three pawns distorts the game much more than removing a knight and one pawn. Rook pawns are only worth about half a pawn in the opening position due partly to the immediate activity of the rook, while "f" and "g" are worth more than one pawn. Removing 3 out of the 4 b,c,d,and e pawns is certainly possible, but it feels less like a normal chess game to me than knight for b,c,d, or e pawn. As for knight for castling, of course Black does not "need" to castle with an extra piece, but being unable to do so makes winning a lot harder. Just for example, the French isn't quite so wonderful for Black, because if White exchanges on d5, the open file gives some chances, whereas normally it just leads to a likely rook exchange. Probably knight for pawn is a bit harder for the human than knight for castling (or knight odds at FRC), but I think that all of these options would have produced a close match with Smerdon. A normal IM would probably lose with these conditions. Note that the 3970x computer used for this match is probably about six times faster (effective speed, with 32 fast cores) than a typical quad laptop, although I don't know what Toth used in his demo game.

Larry I know that your peak rating was back in 2011 with a Fide rating of 2411, but how do you think your chances are now to take a Knight odds with No Castling, using your system. Probably you are closer to a Fide rating of 2200, but we understand that you are getting older and older.
Actually my peak FIDE rating was 2445 back in the '80s, and according to Chessmetrics I was World number 102 around the beginning of 1981. So although I didn't get the GM title until 2008, my best year was 1980 and I was playing well enough then to have earned the GM title if I maintained that level and played internationally. But I had another career then. When I actually got the title by winning the World Senior, I was no longer really GM level, but strong enough to beat older GMs and win the title. Now I'm pretty weak at blitz, especially on the computer (partly due to vision problems), but it seems that I can still play fairly well at 15' + 10", based on my 4.5 to 1.5 result vs. Komodo at knight odds. But if I have to forfeit castling (or remove a nonedge pawn) I think I'll lose more than I win, with some draws. I'll probably try it when I have a bit more time. By the way, if you know anyone who is is NOT getting older, please tell me their secret!
In reality the No castling Chess only delay the manual castling by 2 move, in most cases the a or h pawn move to a4 or h4 followed by the Rook to a3 or h3, but the extra Knight should balance the gout out a little, by NOT allowing an extra piece to Attack the Human King before the Manual castling take effect :
https://chessbase.in/news/No-Castling-C ... g-variant-

Chessqueen
Posts: 1119
Joined: Wed Sep 05, 2018 12:16 am
Full name: Nancy M Pichardo

Re: Komodo vs GM Smerdon knight odds match

Post by Chessqueen » Thu Apr 16, 2020 11:28 pm

Chessqueen wrote:
Thu Apr 16, 2020 4:11 pm
lkaufman wrote:
Thu Apr 16, 2020 4:23 am
Chessqueen wrote:
Thu Apr 16, 2020 3:19 am
lkaufman wrote:
Wed Apr 15, 2020 11:42 pm
jp wrote:
Wed Apr 15, 2020 10:51 pm
lkaufman wrote:
Wed Apr 15, 2020 5:45 pm
Clearly Komodo needs some compensation for the knight to have a decent chance against a GM or strong IM, either a pawn or no castling for the opponent, or chess 960 to defeat routine play.
Larry, what about just giving odds of three pawns, which gives Komodo much better chances, though in material terms it's similar? I don't know which pawns; maybe both rook pawns and queen pawn. If the GM makes it to the endgame alive, he can win, but before then, he risks getting mated. I'm not sure Black actually needs to castle (even practically) when he has an extra piece.
Three pawns is about the same as knight for pawn materially, since a knight is worth about 4 pawns in the opening. But removing three pawns distorts the game much more than removing a knight and one pawn. Rook pawns are only worth about half a pawn in the opening position due partly to the immediate activity of the rook, while "f" and "g" are worth more than one pawn. Removing 3 out of the 4 b,c,d,and e pawns is certainly possible, but it feels less like a normal chess game to me than knight for b,c,d, or e pawn. As for knight for castling, of course Black does not "need" to castle with an extra piece, but being unable to do so makes winning a lot harder. Just for example, the French isn't quite so wonderful for Black, because if White exchanges on d5, the open file gives some chances, whereas normally it just leads to a likely rook exchange. Probably knight for pawn is a bit harder for the human than knight for castling (or knight odds at FRC), but I think that all of these options would have produced a close match with Smerdon. A normal IM would probably lose with these conditions. Note that the 3970x computer used for this match is probably about six times faster (effective speed, with 32 fast cores) than a typical quad laptop, although I don't know what Toth used in his demo game.

Larry I know that your peak rating was back in 2011 with a Fide rating of 2411, but how do you think your chances are now to take a Knight odds with No Castling, using your system. Probably you are closer to a Fide rating of 2200, but we understand that you are getting older and older.
Actually my peak FIDE rating was 2445 back in the '80s, and according to Chessmetrics I was World number 102 around the beginning of 1981. So although I didn't get the GM title until 2008, my best year was 1980 and I was playing well enough then to have earned the GM title if I maintained that level and played internationally. But I had another career then. When I actually got the title by winning the World Senior, I was no longer really GM level, but strong enough to beat older GMs and win the title. Now I'm pretty weak at blitz, especially on the computer (partly due to vision problems), but it seems that I can still play fairly well at 15' + 10", based on my 4.5 to 1.5 result vs. Komodo at knight odds. But if I have to forfeit castling (or remove a nonedge pawn) I think I'll lose more than I win, with some draws. I'll probably try it when I have a bit more time. By the way, if you know anyone who is is NOT getting older, please tell me their secret!
In reality the No castling Chess only delay the manual castling by 2 move, in most cases the a or h pawn move to a4 or h4 followed by the Rook to a3 or h3, but the extra Knight should balance the game out a little, by NOT allowing an extra piece to Attack the Human King before the Manual castling take effect . Here are some sames played by strong young GMs from India: https://www.youtube.com/watch?v=MYq7lZHTMUE

Post Reply