Komodo vs Bolzoni, 5th game Broadcast

[Vinvin]

Did he say whether or not he practiced or analyzed knight odds this past week? Or is he just getting familiar with playing Komodo at knight odds from last weekend's games?

I didn't practice at all and didn't analyze his past games but he pointed the importance of the opening. 1.d4 seems way stronger than others !

[APassionForCriminalJustic]

One last win is needed. Probably doable. So, a Knight handicap worth 1000 elo ? Or there is a cap, any 2300 player can win versus any Elo rating with Knight handicap ?

I hope the next match with 2 pawn handicap will be interesting also.


A quick analysis from stockfish. Knight handicap is rated -3,3 pawn
Two pawn handicap ( c2 f2 ) is rated -1,85 pawn.

So, the difference will be high, and Komodo will probably have better chance.
Less than 2 pawn handicap or something around seems good for computer.

Well yes... obviously a knight handicap must be worth well over 1000+ ELO... that is what I find interesting about this series... no human is even a challenge at normal chess. Everyone knows that. So a complete amount of inferiority can be effectively reached with enough of a handicap if played by a master, and played with good technique. Think about it... as my master FM friend has told me... the game is already won at knight odds... all that's left is good technique. Nothing else.

Victor still lost one match, and drew so he hasn't been that impressive himself if one thinks about. At 3/1 it would be more so. Not 45/15.

You think Magnus would play a 2100 a full knight down? He would not even consider it. This is complete suicide.

[Ferdy]

I'm broadcasting the match here:
http://komodochess.com/games/Live.html

Thanks, Jesse. Bolzoni seems to be a stronger player than he was last weekend. My guess is that he practiced and/or analyzed knight odds in the interim. Probably he thought he would win easily before the match, but after the first half he realized it wasn't so simple.

He is easily a 1000+ ELO weaker so this should never have been thought to be easy.

I suspect that a majority of masters would say that even God (or perfect play, if you prefer) could not give them knight odds successfully in a serious game like these. Whether they would be right or not is impossible for me to say. Today's chess engines are designed to play the "best" move rather than the trickiest, which is probably the strategy needed to beat strong players at this large handicap.

The human learning is at work here. This should not be underestimated in a match play of more than 2 games. I have not considered such factor in my prediction . But I agree that something must be modified on the engine play, one is an option to prefer attack than defense. Also a specialized book to address this kind of handicap should be created.

[lkaufman]

One last win is needed. Probably doable. So, a Knight handicap worth 1000 elo ? Or there is a cap, any 2300 player can win versus any Elo rating with Knight handicap ?

I hope the next match with 2 pawn handicap will be interesting also.


A quick analysis from stockfish. Knight handicap is rated -3,3 pawn
Two pawn handicap ( c2 f2 ) is rated -1,85 pawn.

So, the difference will be high, and Komodo will probably have better chance.
Less than 2 pawn handicap or something around seems good for computer.

The elo value of any handicap rises with the level. I imagine a 1600 player can give knight odds to a 1200 player with competitive results, but it is likely that somewhere above some point in the 2100 to 2400 range the player receiving knight odds can win a match against even an all-knowing opponent.

I'm sure Komodo would win a match Mr. Bolzoni at two pawn handicap. But the GM playing at this handicap is 275 elo higher rated than Mr. Bolzoni. So it's not clear to me whether the GM has a higher or lower chance to win his match than did Mr. Bolzoni.

[Ferdy]

I'm broadcasting the match here:
http://komodochess.com/games/Live.html

Thanks, Jesse. Bolzoni seems to be a stronger player than he was last weekend. My guess is that he practiced and/or analyzed knight odds in the interim. Probably he thought he would win easily before the match, but after the first half he realized it wasn't so simple.

He is easily a 1000+ ELO weaker so this should never have been thought to be easy.

I suspect that a majority of masters would say that even God (or perfect play, if you prefer) could not give them knight odds successfully in a serious game like these. Whether they would be right or not is impossible for me to say. Today's chess engines are designed to play the "best" move rather than the trickiest, which is probably the strategy needed to beat strong players at this large handicap.

In this position Komodo played Qe2, then next played Be4 (offering an exchange to a knight). Something must have been going on in the search.
[d]r2q1rk1/1p3ppp/p1nb1n2/4p3/2Pp4/PP1B1Q1P/1B1P1PP1/R4RK1 w - - 5 13

[APassionForCriminalJustic]

I'm broadcasting the match here:
http://komodochess.com/games/Live.html

Thanks, Jesse. Bolzoni seems to be a stronger player than he was last weekend. My guess is that he practiced and/or analyzed knight odds in the interim. Probably he thought he would win easily before the match, but after the first half he realized it wasn't so simple.

He is easily a 1000+ ELO weaker so this should never have been thought to be easy.

I suspect that a majority of masters would say that even God (or perfect play, if you prefer) could not give them knight odds successfully in a serious game like these. Whether they would be right or not is impossible for me to say. Today's chess engines are designed to play the "best" move rather than the trickiest, which is probably the strategy needed to beat strong players at this large handicap.

The human learning is at work here. This should not be underestimated in a match play of more than 2 games. I have not considered such factor in my prediction . But I agree that something must be modified on the engine play, one is an option to prefer attack than defense. Also a specialized book to address this kind of handicap should be created.

That's a good point. Human learning... human consciousness is such a pest.

[APassionForCriminalJustic]

One last win is needed. Probably doable. So, a Knight handicap worth 1000 elo ? Or there is a cap, any 2300 player can win versus any Elo rating with Knight handicap ?

I hope the next match with 2 pawn handicap will be interesting also.


A quick analysis from stockfish. Knight handicap is rated -3,3 pawn
Two pawn handicap ( c2 f2 ) is rated -1,85 pawn.

So, the difference will be high, and Komodo will probably have better chance.
Less than 2 pawn handicap or something around seems good for computer.

Two pawns should be perfectly doable for the computer.

[APassionForCriminalJustic]

I'm broadcasting the match here:
http://komodochess.com/games/Live.html

Thanks, Jesse. Bolzoni seems to be a stronger player than he was last weekend. My guess is that he practiced and/or analyzed knight odds in the interim. Probably he thought he would win easily before the match, but after the first half he realized it wasn't so simple.

He is easily a 1000+ ELO weaker so this should never have been thought to be easy.

I suspect that a majority of masters would say that even God (or perfect play, if you prefer) could not give them knight odds successfully in a serious game like these. Whether they would be right or not is impossible for me to say. Today's chess engines are designed to play the "best" move rather than the trickiest, which is probably the strategy needed to beat strong players at this large handicap.

That's some good points Larry. Here is actually an excerpt from my email correspondence with my FM friend who said some interesting things related to knight odds:

"Although I respect your computer's chess strength immensely, I feel that a knight would be too much material to spot me. I am well aware that as a human I am definitely capable of making mistakes, even in the simplest of positions. Part of the reason Poly is so strong and rarely loses is because it is always better on position, and doesn't dig itself into a hole it can't get out of. Starting without a knight is a deep, deep hole. I suspect I should be able to beat it rather easily with such odds, but it would be interesting to see".

He continued with:

"I'm confident that I can spot a knight to someone under 2000, perhaps even 2100 USCF in blitz (more than 400 points difference) and come out way ahead as I have played many such propositions in the past. In this case, I'm estimating that the difference in rating between me and your engine is probably around twice that (800-900.) But I believe that there are fundamentals of the game which a 2400 USCF player like myself has down which someone who is not at least a master is lacking. A knight is simply too much to expect to come out ahead at the master level, even if god were giving the odds. Basically, if the computer is able to complicate the game in a thought-provoking way (not easy to do) it has chances to not lose".

I think he backs up a lot of what you are saying Larry. The handicap itself increases as Elo increases. But I think his main point was that there are fundamentals of the game that all masters have that would make winning knight odds matches versus them very difficult. But as you said Larry, if the engine searches for tricks or complications then it has better chance for greater success...

[lkaufman]

I'm broadcasting the match here:
http://komodochess.com/games/Live.html

Thanks, Jesse. Bolzoni seems to be a stronger player than he was last weekend. My guess is that he practiced and/or analyzed knight odds in the interim. Probably he thought he would win easily before the match, but after the first half he realized it wasn't so simple.

He is easily a 1000+ ELO weaker so this should never have been thought to be easy.

I suspect that a majority of masters would say that even God (or perfect play, if you prefer) could not give them knight odds successfully in a serious game like these. Whether they would be right or not is impossible for me to say. Today's chess engines are designed to play the "best" move rather than the trickiest, which is probably the strategy needed to beat strong players at this large handicap.

In this position Komodo played Qe2, then next played Be4 (offering an exchange to a knight). Something must have been going on in the search.

Probably Komodo concluded that otherwise Black would eventually achieve the crushing breakthrough ...e4.
[d]r2q1rk1/1p3ppp/p1nb1n2/4p3/2Pp4/PP1B1Q1P/1B1P1PP1/R4RK1 w - - 5 13

[duncan]

I'm broadcasting the match here:
http://komodochess.com/games/Live.html

Thanks, Jesse. Bolzoni seems to be a stronger player than he was last weekend. My guess is that he practiced and/or analyzed knight odds in the interim. Probably he thought he would win easily before the match, but after the first half he realized it wasn't so simple.

He is easily a 1000+ ELO weaker so this should never have been thought to be easy.

I suspect that a majority of masters would say that even God (or perfect play, if you prefer) could not give them knight odds successfully in a serious game like these. Whether they would be right or not is impossible for me to say. Today's chess engines are designed to play the "best" move rather than the trickiest, which is probably the strategy needed to beat strong players at this large handicap.

In this position Komodo played Qe2, then next played Be4 (offering an exchange to a knight). Something must have been going on in the search.

Probably Komodo concluded that otherwise Black would eventually achieve the crushing breakthrough ...e4.
[d]r2q1rk1/1p3ppp/p1nb1n2/4p3/2Pp4/PP1B1Q1P/1B1P1PP1/R4RK1 w - - 5 13

thanks for arranging this fascinating tournament. komodo did brilliantly considering the handicap, but ... if there was a rematch what do you think would happen ?