Forums

Rybka cant slove this one

Sort:
j1m1

Yesterday i played  2 hour game against chessmaster's personality Buck and we reached this position. I played very nice move wich wins pretty much on the spot. Later, after the game, i tried to check it with rybka and it also doesnt find it. So try to find it. It will be well worth your time. ( no cheaters on this one :) )

j1m1

I'll post the solution in few days.

Lord-Chaos

not e6 is it? then if the f pawn moves or takes, Nxg6 but im not sure what that will do... havnt checked it on board yet, visualized it =)

Lord-Chaos

actually got it, e6 if he takes then we get a nice fork on queen and rook, so e6 f6 bxf6 Bxf6 Nxf6! but then he takes my bishop =( something to do with the B file i presume lol.

RyanMK

Nxh5!

GenericZebra

Ok, jim.  Lets see the solution.

j1m1

Yep. Nxh5 followed by Qe2. There is no good denfence to this idea.

kosmeg

And rybka couldn't find that?

j1m1

Nope. You can try for yourself. Main variation is a bit longer but its just winning outright.

kosmeg

Yeah...I don't have Rybka but I checked it with Fritz 10. You're right it can't find it. Quite weird such a strong machine can't find that as it's the first thing that comes into mind, but that's why I like human chess and not computer chess...Intuition creates, calculation just executes...

kosmeg

I believe this sacrifice is something intuitive, that's why most engines are not able to solve it. Engines are weak when it comes to ideas...That's why sometimes the are weak in endgames and most of the time they're unable to solve studies. Rybka is rated around 3100 if I'm right and my coach is a player playing for 2100-2200. I've seen them both trying to solve a study and he needed something like 15 minutes, Rybka needed 3,5 hours...

PurplePuppy
RainbowRising wrote:

I thought Rybka and such worked by just looking at every move and calculating all the variations?


It doesn't look at ALL the moves. Doing that would take forever on even the state of the art computers. It just chooses a few moves which it believes to be "reasonable" and analyzes the lines resulting from it.

LordJones3rd
PurplePuppy wrote:
RainbowRising wrote:

I thought Rybka and such worked by just looking at every move and calculating all the variations?


It doesn't look at ALL the moves. Doing that would take forever on even the state of the art computers. It just chooses a few moves which it believes to be "reasonable" and analyzes the lines resulting from it.


 the difference between reasonable and rubbish is...?