Did you let it play the first eight moves as well? And why another "omg engines are stupid look I have deeper human insights" thread. The Internet is already full of these. Who cares :p
Rybka is Dumb!
Sort:
That's why engines come with opening books. Their specialty isn't openings or endgames. Use the book it came with.
I'll take Rybka's analysis over convential wisdom. It may well be that for the average player, h3 weakens the pawn structure with no compensation, but with computer-level play, it must be the best move.
Ripping open the kingside would be via a sacrificial attack, right? So if Rybka, being the chess engine that it is, could counter any attacks following the sacrificial attack, then it would come out slightly ahead I guess?
Once you've installed and set the opening book that was designed for the Rybka that you purchased, you will either feel that:
1. you just spent the best $70 of your life
2. you'd wish that you never heard about opening books and tablebases for chess engines that can run on multiprocessor computers to handle search depth=21 easily
Perhaps a better idea would be to let Rybka play the move h3 and you continue the game, showing us and your engine how you can rip off the king side. I suspect Rybka has its own plans :)
For those of you that believe in Rybka's suggestion -- you don't understand how computers play chess. It has to do with horizons and evaluations. It may not be a game losing blunder, but it's certainly not a move you could recommend, especially in practical terms.
Nov 11, 2010
0
#9
Run Rybka 4 for a longer time during the opening phase. Also, the Rybka 4 Opening Book by Jiri Dufek is extremely good.
http://www.wholesalechess.com/chess/rybka_4_opening_book_by_jiri_dufek?ac=wscmerch
The strongest, free chess engine, Houdini 1.03a/2 CPU found 9.h3 at Depth = 16, then found 9.Bg5 at Depth = 19.
9.Bg5 gives White the best winning chances at 48.6%, acording to our Game Explorer. So, Houdini found the correct, best response for White. Here's Houdini's continuation at Depth = 22.
Rybka does not understand the death star
rdecredico wrote:
orangehonda wrote:
For those of you that believe in Rybka's suggestion -- you don't understand how computers play chess. It has to do with horizons and evaluations. It may not be a game losing blunder, but it's certainly not a move you could recommend, especially in practical terms.
I understand it just fine, having twice as many years experience with them as you have been on the planet.
There are several totally respectable lines in which white plays an early h3. It usually comes in a move or two before White commits his light-squared bishop and often the knight in these systems goes to e2 instead of f3, but they often overlap and h3 at this juncture (as given by the engine) is completely playable as it can transpose into known variation of the Makogonov system.
And for those of you that think they know better than other people they have never met and now nothing about, we have a set of words for what they are and a saying that aplies directly: Unwarranted arrogance.
"Youth is wasted on the young."
Easy fellah! it's only chess that's being discussed here.
Strangely, my engine (DeepRybka4 SSE42x64 running on a beast) seems to prefer 9. a3 (+0.46 @ depth21).
Seems like a difficult position for engines to evaluate, certainly I wouldn't want to play a3 or h3 at this point.
9.h3 is certainly not a common move here, my database has only 58 games with it -as opposed to more than 5000 with 9.b4 and nearly 5000 with 9.Ne1. The Black has scored more than 50 % wins in those games, the White's wins and draws together being less than 50 %. I don't think it is an outright losing move but it is not a good one.
Atos wrote:
9.h3 is certainly not a common move here, my database has only 58 games with it -as opposed to more than 5000 with 9.b4 and nearly 5000 with 9.Ne1. The Black has scored more than 50 % wins in those games, the White's wins and draws together being less than 50 %. I don't think it is an outright losing move but it is not a good one.
Unfortunately, when it comes to a machine as powerful as Rybka the results of past games makes no difference. Rybka 4 is strong enough to beat any player in the world consistently and every machine around 8/10 or 8.5/10 times, and would not play the move for no reason: it must have something which will assure it can comfortably survive all of the coming attacks, or which will allow it to launch its own powerful plan. Do not assume that this plan cannot exist just because humans have been incapable of coming up with it. I'd like to see you continue through to a win for Black from this against Rybka 4 (in fact, I'm sure that no computer in the world would be able to do better than 1.5/10 against Rybka 4 in this position).
That's irrelevant. Rybka can beat an amateur with 1.f3 but that will not be because 1.f3 is a good opening move. The move is strategically naive even if it doesn't have a forced tactical refutation.
Atos wrote:
That's irrelevant. Rybka can beat an amateur with 1.f3 but that will not be because 1.f3 is a good opening move. The move is strategically naive even if it doesn't have a forced tactical refutation.
How do you know (I think I can assume here that Rybka is better at calculation than you - if I'm wrong feel free to correct me)? It is possible that Rybka is well aware that it can handle any oncoming attack and believes that 9. h3 prepares it for its own attacking plan which humans have yet to come up with, or which adequately and most easily enables it (Rybka only - other people may have a harder time finding a defense) to defend against any and all oncoming attacks.
Chessgod123 wrote:
Atos wrote:
That's irrelevant. Rybka can beat an amateur with 1.f3 but that will not be because 1.f3 is a good opening move. The move is strategically naive even if it doesn't have a forced tactical refutation.
How do you know (I think I can assume here that Rybka is better at calculation than you - if I'm wrong feel free to correct me)? It is possible that Rybka is well aware that it can handle any oncoming attack and believes that 9. h3 prepares it for its own attacking plan which humans have yet to come up with, or which adequately and most easily enables it (Rybka only - other people may have a harder time finding a defense) to defend against any and all oncoming attacks.
Rybka doesn't think in such terms, it thinks in terms of calculating lines and evaluations. The Black's attack is still too far off for Rybka to be able to calculate all the lines effectively. (Or for us, but we can see that the move is likely to prove a weakening.)
h3 is a book move. I would have played h3 as well. It's as good as Ne1
Chuck Norris could win it. With rook odds.
I'm sure those of you who play the KID will recognize this position:
So if your familiar with theory, you know there are many moves for White here: b4, Ne1, Be3, etcetera. However, Rybka, at 11 knodes a second, at 14 depth, says that White should play ... (wait for it) ...
9.h3!! +(0.44)
Am I missing something here? Any beginner KID player knows that h3 simply makes Black's job of ripping open the kingside EASIER. And it weakens the pawn structure around the king for no good reason. Why did I spend $70 on Rybka when it comes up with moves like this?