Rybka's "Ra3"

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

Moderators: hgm, Rebel, chrisw

swami
Posts: 6640
Joined: Thu Mar 09, 2006 4:21 am

Rybka's "Ra3"

Post by swami »

I saw the Rybka vs Zappa game yesterday, a lot of people seem to think that Rybka's Ra3 move early in the opening is something that only computers can understand, and it can't be comprehended by human.

I think it is some kind of bug in a evaluation or search function,or that programs do make occasional moves that are without plans and are pointless according to strong human from GM standpoint or top correspondence player.

I think it is time that this kind of bug need to be fixed,I know that it is difficult to do but it is better that moves without plans need to be avoided.Only difference is that a lot of chess engines are pretty good at tactics and test suites whereas GM's are not,atleast not unless they are given more time.
But I get an impression that engines do not understand GM level positional chess where every move that GM's make are not without plans.

I would like to hear from chess programmers and some players what they think about it.

Regards.
User avatar
Mike S.
Posts: 1480
Joined: Thu Mar 09, 2006 5:33 am

Re: Rybka's "Ra3"

Post by Mike S. »

[D]1r3r1k/3q1ppp/p2p1n2/2p1pPB1/P5P1/1P1P3P/1P3P2/R2QR1K1 w - c6 0 23
The move 23.Ra3 in the 8th game looked unusual, but maybe it was not such a big mystery: If we look at the pawn b3 and the Qd1 before it was played: By Ra3, Rybka simply releases the wQ from defending the b3 pawn, allowing to play Qe2 or Qf3 later. (Also, the rook remained in a to support the passer which wasn't created too far ahead, by 28.Qxa6.)

So I think, it just looked unusual but had fairly simple reasons actually. If your engines don't want to play it, switch to multi-pv mode and it will probably be among the best 3, soon.

Another funny looking move happened in the 9th game:

[D]3r1k2/1bq2pp1/1p1rpn1p/p3N3/1PPb1P1B/P2R4/4Q1PP/1B2R2K b - - 0 26
26...Ba1 :mrgreen: Also not difficult: Rybka wanted to play Rd6-d4 and a1 was the bishop's only square to go to (Bxe5 was bad).
Regards, Mike
swami
Posts: 6640
Joined: Thu Mar 09, 2006 4:21 am

Re: Rybka's "Ra3"

Post by swami »

Good analysis.but Ra3 supporting that pawn letting other pieces move leaves the rook there with pawn till the end game.Most Grandmasters would have had better moves than Ra3.

Well there are some unusual or 'waiting' move from most engines in engine vs engine matches.There are none in grandmaster games,it's got to be something to do with the playing style.

Let's see what funny moves we encounter tonight :)

Regards.
bedouin

Re: Rybka's "Ra3"

Post by bedouin »

What about the Re3 that was played by Deep Fritz versus Kramnik? All that was missing was a mechanical voice saying "computer thinking". Computers do make such crude moves because they have AI.

Has someone tried the LK versions of Rybka to see whether they can play that move?