S.Taylor wrote:Comment: It seems like it is very very hard work (for humans) with N and B vs R endings, but that the side with N and B CAN polish off pawns with very hard work, Vs Rook.
Would a human GM manage to do all this without error?
I am not at all sure that endgame was won. Note that Rybka's eval was drawish in the beginning.
It is dangerous for a human who doesn't even play chess to criticize the moves of top rated computer programs, but to me the a2-a4-a5 maneuver at moves 39 and 40 looks very dubious. The pawn on a5 serves no useful purpose for white, and is an easily attacked weakness for black. If white ever tries to defend it by advancing the b pawn, the pawn on c3 becomes weak instead.
If white had kept the pawn on a2, I don't see an easy way for black to break through.
Game 24, I honesty think Rybka could have made an overwhelming attack here by 21.f6 or first 21.Nd5 and then 22.f6.
Rybka obviously didn't see any win in thiis, in its horizon, however, i wonder if there was one just a little further after that.
If so, then it is clear that machines still don't see enough.
Obviously though, i too must admit as did Tom above, that "It's dangerous", to argue with the top computers befiore trying it out myself. Blacks night on d4 can get back to e6, as the pawn on f5 would have gone. This may make the difference (which i hadn't noticed) so that now, Black may have enough defense.
(While I'm at it, perhaps i can also ask how game 23 was such a short sudden draw?)
Again, I think that any normal GM game normally goes on a few more moves.
I could already see great ideas for White, but I didn't expect never to see one of them being tried out, and simply the next game is suddenly underway.
If a person has a computer handy, then that's nice. But I actually do not have, at the moment.
[I see a few nice features on chessbomb, but not enough of a substitute which is easy enough to follow].
I still think SF can do as good as Houdini. And we have less draws, more defined games. very nice.
I'm following games in Chessbomb, my favorite viewer from long time!! Its nice there to see game evaluations using SF 1.8 but only using around 30 secs for evaluation. Yo can see some blunders in red they not are real blunders. And probably the same goes when chessbomb evaluates a game between 2 GM.
Game 27 was weird. Does anyone understand the purpose of the Be7-d8-a5-c3 maneuver? It seems to me that black just spends three full moves transferring a tremendously useful defensive piece to a distant square where it does actually nothing, and leaves white with effectively a full piece up for the kingside attack. Am I missing something?
Tord Romstad wrote:Game 27 was weird. Does anyone understand the purpose of the Be7-d8-a5-c3 maneuver? It seems to me that black just spends three full moves transferring a tremendously useful defensive piece to a distant square where it does actually nothing, and leaves white with effectively a full piece up for the kingside attack. Am I missing something?
G27, move 35... Bd8 maneuver, Rybka doesn't like passive defence... It is seeking counterplay in the Queenside. Passive defence by Black will lose the game. So, there's nothing wrong with the maneuver because Rybka is trying to seek counterplay in a lost position. What else can Black do?
Martin Thoresen wrote:Currently the "resign" option in Fritz GUI is set to "early", which means both engines have to have an evaluation 4.8 pawns or more for 6 consecutive moves.
Two other options are "late" and "never".
"Late" is the same as early but with 6.5 pawn advantage.
"Never" is, well, played until mate. The disadvantage with this option is that the games will take quite a bit longer to finish.
I am considering the options here if people really want to see the games played out all the way.
Wasn't game 28 adjudicated earlier than it should (according to the "early" definition)? There were no 6 consecutive plies above 4.8.
.