Mexico - Zappa 5.5 v Rybka 4.5- Zappa wins $10,000

[Anthony C]

Thanks all.

I was amazingly happy to draw the 10th game. Rybka had a very good position very quickly and Zappa then proceeded to kill whatever counterplay it had on the queenside with b4?. At that point its score slowly started rising, but Rybka played very poorly. At one point in particular Zappa was expecting Kh1 + Rg1 with a very nasty attack. Black cannot do anything at all. Fortunately Rybka played a3, giving Zappa its counterplay on the queenside again, and Zappa managed to draw.

I would hardly say Zappa crushed Rybka in the match, but at least it was close enough that with a bit of luck it was able to win.

anthony

[VP]

Thanks all.

I was amazingly happy to draw the 10th game. Rybka had a very good position very quickly and Zappa then proceeded to kill whatever counterplay it had on the queenside with b4?. At that point its score slowly started rising, but Rybka played very poorly. At one point in particular Zappa was expecting Kh1 + Rg1 with a very nasty attack. Black cannot do anything at all. Fortunately Rybka played a3, giving Zappa its counterplay on the queenside again, and Zappa managed to draw.

I would hardly say Zappa crushed Rybka in the match, but at least it was close enough that with a bit of luck it was able to win.

anthony

Hi Anthony,
Thanks for giving the chess community such a strong engine!!
Congratulations to Zappa team on the wonderful result.
Please reconsider your retirement from chess programming, as I am sure you can give something to the chess community which no one else can.

We need more competition at the top, and while other authors are also trying, you are probably the most likely with Zappa to challenge Vas and Rybka.
As the games demonstrated, Rybka is far from being a perfect engine, and no other engine has been able to exploit its shortcomings like Zappa.

We all know that Vas is going to bounce back, fix the short comings and create a super strong Rybka 3.0, and would request you to start working on Zappa 3.0 as well

Best regards,
VP

[George Tsavdaris]

[fen]7B/5k2/4p3/5pR1/4r1p1/8/3Kb3/8 b - - 0 71[/fen]

I just tried little Palm Hiarcs on this position and running on my handheld it knew Ke7 was the correct move and 71...f4??? was a blunder.

I don't have Palm Hiarcs but does Palm Hiarcs shows a draw score for 71...f4?

Because if it shows a negative score(that favors black) for 71...h4? it does not understand that f4? draws so it does not say that 71...f4 was a blunder.

I mean if a program shows Ke7 as -5.50 and f4 as -3.50, this program is valuable for us because it understands that the better move is Ke7, but this program also does not show that f4 is a blunder since it thinks it is winning also.
This program shows indeed that f4 is inferior to Ke7, but not that f4 is a blunder as you say, for the Hiarcs Palm case....

[DarkSide]

HI ALL!! I´m happy to log in this forum!! Zappa is one of my favorites engine, congrats to all the team, Antony great job , I think this match showed the best Zappa and the wrongs Rybka issues, I must say that Rybka disappointed to me, I always thought that Ryb was a heavy eval engine and not a deep search engine, but well it has a good algorithm that allows to select him a number very reduced of variants and search it very deep. Unfortunately it seems that Zappa will not be able to even reach to Rybka in CEGT and CCRL test ( I would really like this) probably in this level (4CPU) the deep search of Rybka is enough to him to follow in the leadership... I hope that Hiarcs is going to be the next Fish´s assassin come on Mark the force is with you

[Mark Uniacke]

George,

I gave Palm Hiarcs just a few seconds on each move and it thought Ke7 was -6.08 and f4??? as -2.93.

While I agree with you it does not show a draw score, it does show a score dropping over 3 pawns in value in one move, it is clearly a blunder and risks the game result changing which is in fact what happened.

Remember the comparision here is between Rybka running on 8 latest Intel processor 2.66Ghz cores with lots of memory and 3/4/5/6 piece endgame tablebases compared to an old slow mobile device with no endgame databases.

Rybka used a computer more than 10,000 times faster than my Palm!

[Dylan Sharp]

if Rybka loses (which appears likely.).i am expecting a mass exodus of loyal Rybka fans from the Rybka forum to the Hiarcs Forum

Maybe if you change the board name to "ZAPPA Chess Forums",give them the book Zappa used on the match, and buy everyone a 8 core machine, they will consider moving...

[Mark Uniacke]

I guess you'll have to wait and watch HIARCS eating Sushi instead then...

[Mark Uniacke]

We have MOVED the technical discussion to the new Programming Discussions section here:
http://www.hiarcs.net/forums/viewtopic.php?t=402