Will Shredder ever return ...

[AdminX]

Will Shredder ever return as the number 1 chess engine on the rating charts? If you were Stefan Meyer-Kahlen what would you change, ajust, or add to improve or surpass the current top of the crop group of chess engines? The latest CCRL 40/40 has Shredder at the number 6 spot. It's kind of sad that the closest thing to Rybka is only Naum. I hate seeing this former great engine getting it's head handed to him in what seems like game after game versus the top two Rybka and Naum engines.

[Dr.Wael Deeb]

Strange enough Ted,the very same thought crossed my mind a few days ago when I was looking at the updated CCRL rating list....It's sad indeed but hey....welcome to the real world....
BTW,there was a saying in Bulgaria when I was living there and it goes somehting like this:
Life is like a carrot....one day it's in your hand....ten days it's in........
Sorry,you have to guess it yourself

[Matthias Gemuh]

Life is like a carrot....one day it's in your hand....ten days it's in........

... a rabbit ?

Why not eat it yourself ? But if you then eat the rabbit, the carrot is in ...

Reminds me of Shakespeare :
- A man can fish with a worm that had eaten of a king
and eat of the fish that had eaten of that worm

... sounds like his Majesty ends up flushed down the toilet.


Matthias.

[Tord Romstad]

Will Shredder ever return as the number 1 chess engine on the rating charts? If you were Stefan Meyer-Kahlen what would you change, ajust, or add to improve or surpass the current top of the crop group of chess engines? The latest CCRL 40/40 has Shredder at the number 6 spot. It's kind of sad that the closest thing to Rybka is only Naum. I hate seeing this former great engine getting it's head handed to him in what seems like game after game versus the top two Rybka and Naum engines.

That's an unfair comparison, as Shredder is a very old program, while Rybka and Naum are much newer. When you keep working on an old program, you'll sooner or later arrive at a local optimum where improving the program further is very difficult. If SMK, with all his experience, were to write a new program today, he would probably easily be able to achieve a rating 100-200 Elo points above Shredder 11. Clearly, he has other priorities, like developing a complete chess program (that's another way your comparison is unfair, because Rybka and Naum are just UCI engines: They are really not competing in the same league as Shredder) and supporting multiple platforms. I think he's got his priorities right: Just look at the rating. Shredder 11 on a quad has a rating above 3000 (!) on the CCRL list. There aren't that many people in the world who needs a program stronger than that.

Tord

[AdminX]

Will Shredder ever return as the number 1 chess engine on the rating charts? If you were Stefan Meyer-Kahlen what would you change, ajust, or add to improve or surpass the current top of the crop group of chess engines? The latest CCRL 40/40 has Shredder at the number 6 spot. It's kind of sad that the closest thing to Rybka is only Naum. I hate seeing this former great engine getting it's head handed to him in what seems like game after game versus the top two Rybka and Naum engines.

That's an unfair comparison, as Shredder is a very old program, while Rybka and Naum are much newer. When you keep working on an old program, you'll sooner or later arrive at a local optimum where improving the program further is very difficult. If SMK, with all his experience, were to write a new program today, he would probably easily be able to achieve a rating 100-200 Elo points above Shredder 11. Clearly, he has other priorities, like developing a complete chess program (that's another way your comparison is unfair, because Rybka and Naum are just UCI engines: They are really not competing in the same league as Shredder) and supporting multiple platforms. I think he's got his priorities right: Just look at the rating. Shredder 11 on a quad has a rating above 3000 (!) on the CCRL list. There aren't that many people in the world who needs a program stronger than that.

Tord

You make some good points. I would just like to see his program return to the top of the lists. Rating is not the only thing, but it does influence choice.

[fern]

As always, you hit the target with a fair judgment. Me, I do not ask neither I need stronger engines, but perhaps better guis and features. There are many extra things to do in those fields, but till now, 99% of programmers, if not all, are engaged in the Elo race.
A really god teacher, 10 times better than even Fritz, is what I hope to get before my brain becomes so dumb I will forget the legal movements of the pawn.

My best
fern

[Dann Corbit]

To write any chess program is a difficult thing. If you can write a program as strong and robust as TSCP, then you are a very good programmer.

According to this ranking, Shredder is the 5th strongest chess program in the world:

Code: Select all

Rank Name Rating Score Average Opponent Draws Games LOS  ELO + - 
1 Rybka 3 64-bit 4CPU 3233 (+7) +26 -26 78.7% -205.1 33.3% 559   
2 Naum 4 64-bit 4CPU 3152 (+2) +27 -27 63.0% -84.3 43.3% 409 
3 Deep Fritz 11 4CPU 3106 (+13) +31 -31 54.7% -30.0 55.6% 279 
4 Zappa Mexico II 64-bit 4CPU 3068 (-2) +23 -22 58.0% -53.3 48.3% 582 
5 Deep Shredder 11 64-bit 4CPU 3030 (+13) +22 -22 52.4% -14.7 43.2% 657 
6 Deep Sjeng WC2008 64-bit 4CPU 3028 (+1) +25 -25 45.3% +29.5 41.0% 503 
7 Toga II 1.4.1SE 4CPU 3014 (+11) +23 -22 49.3% +7.0 40.2% 612 
8 Thinker 5.4a Inert 64-bit 4CPU 3008 (+9) +27 -27 45.2% +35.2 44.9% 408 
9 Hiarcs 12 4CPU 3007 (+4) +21 -21 50.6% -0.4 42.9% 672 
10 Glaurung 2.2 64-bit 4CPU 2995 (+5) +26 -26 44.2% +38.2 41.3% 446 
11 Loop M1-T 64-bit 4CPU 2952 (-3) +22 -22 46.0% +24.3 43.2% 606 
12 Bright 0.3a 4CPU 2944 (+5) +28 -28 46.1% +29.6 41.1% 387 
13 Deep Junior 10 4CPU 2928 (+17) +26 -26 44.8% +37.3 33.1% 484 
14 Spike 1.2 Turin 2849 (-1) +19 -19 52.9% -19.3 38.4% 949 
15 Ktulu 8.0 2814 (+8) +19 -19 48.8% +9.4 37.5% 875 
16-17 Chess Tiger 2007.1 2807 (+3) +20 -20 54.5% -25.7 44.0% 723 
16-17 Delfi 5.4 2CPU 2807 (-8) +32 -32 57.3% -49.0 39.5% 301 
18 Twisted Logic 20080620 2796 (-5) +33 -33 51.9% -11.3 39.8% 266 
19 Frenzee Feb08 2789 (-4) +27 -27 52.0% -12.6 38.2% 416 
20 Booot 4.14.0 2774 (+6) +28 -29 51.1% -8.4 41.2% 364 
21 Movei 00.8.438 (10 10 10) 2763 (-6) +26 -26 49.9% +0.5 40.4% 451 
22 Cyrano 0.6b17 2762 (+7) +35 -35 55.2% -35.0 36.8% 250 
23 Chessmaster 11 2760 (-9) +27 -27 49.6% +1.2 33.4% 419 
24 Scorpio 1.91 2758 (-8) +29 -29 44.7% +35.5 37.8% 376 
25 E.T. Chess 13.01.08 2750 (+8) +25 -25 48.5% +8.2 40.0% 463 

Considering that there are ~7 billion people on the earth and that only 4 people have currently written a program that is stronger than his, I would say that puts him in a fairly strong position.

There is some strange cultural thing that says we have to strive for number one, and I don't think that is necessarily an awful thing. But (for instance) if an Olympic athlete does not win the gold medal, that is nothing to be ashamed of. And if the athlete does not even get to the podium that is nothing to be ashamed of. The women sprinters in the Olympics would kick my butt in a fearsome way. The women shotputters can throw a shotput farther than I can. They are great athletes. So I admire the Canadian perspective in sports reporting. They praise someone in a big way if they achieve a personal best.

Any chess program is a terrific achievement. If it is one of the world's strongest, then it is even more a thing of beauty. If it does not stand over all the others, its beauty is not diminshed.

IMO-YMMV.

[mclane]

yes-. true. but how ? there are 200 ELO points between them.

[Dann Corbit]

yes-. true. but how ? there are 200 ELO points between them.

Shredder has picked up about 200 Elo points since Shredder 9.0.

Code: Select all

Rank Name Rating Score Average Opponent Draws Games LOS  ELO + - 
6 Deep Shredder 11 64-bit 4CPU 3017 +13 -13 50.8% -6.0 43.5% 1969 
  Deep Shredder 10 64-bit 4CPU 2936 +17 -17 50.4% -1.7 39.9% 1135 
  Deep Shredder 9 2CPU 2853 +18 -17 56.9% -44.9 36.3% 1087 

Unfortunately for the competition, Rybka has picked up 300 Elo in about the same time frame:

Code: Select all

Rank Name Rating Score Average Opponent Draws Games LOS  ELO + - 
1 Rybka 3 64-bit 4CPU 3226 +19 -18 76.9% -190.1 36.5% 1105   
  Rybka 2.2n2 64-bit 4CPU 3110 +30 -29 73.3% -157.9 38.6% 404 
  Rybka 2.1 64-bit 2CPU 3073 +26 -25 73.3% -166.0 34.4% 572 
  Rybka 1.2 64-bit 2997 +22 -21 74.4% -174.4 32.9% 821 
  Rybka 1.1 64-bit 2986 +24 -24 76.1% -185.1 34.6% 665 
  Rybka 1.0 64-bit 2918 +16 -16 68.6% -126.0 35.0% 1388 

On the other hand, if we step back and look at the long term picture, SSDF data shows that Shredder has picked up 3043-2290=753 Elo, whereas Rybka has only picked up 3227-2902=325 Elo:

Code: Select all

     THE SSDF RATING LIST 2009-01-16   %118048 games played by  311 computers
                                           Rating   +     -  Games   Won  Oppo
                                           ------  ---   --- -----   ---  ----
   1 Deep Rybka 3 x64  2GB Q6600 2,4 GHz     3227   36   -31   673   86%  2924
   6 Deep Shredder 11  2GB  Q6600  2,4 GHz   3043   31   -29   564   64%  2941
  10 Rybka 2.3.1 Arena 256MB Athlon 1200 MHz 2925   24   -24   842   56%  2884
  11 Fritz 11  256MB Athlon 1200 MHz         2919   33   -33   453   58%  2865
  12 Deep Shredder 11  256MB  Athlon 1200    2913   34   -34   410   49%  2920
  14 Rybka 1.2  256MB Athlon 1200 MHz        2902   25   -24   893   71%  2747
  15 Shredder 8 MP 2GB  Q6600  2,4 GHz       2899   33   -34   441   43%  2948
  24 Shredder 10 UCI  256MB Athlon 1200 MHz  2824   20   -20  1246   58%  2767
  27 Shredder 9.0 UCI  256MB Athlon 1200 MHz 2809   20   -20  1280   66%  2696
  28 Shredder 8.0 CB  256MB Athlon 1200 MHz  2802   22   -21  1121   71%  2648
  29 Shredder 7.04 UCI 256MB Athlon 1200 MH  2794   20   -20  1269   66%  2680
  32 Rybka 1.0 beta 128MB K6-2 450 MHz       2774   64   -69   115   38%  2858
  35 Shredder 7.0  256MB Athlon 1200 MHz     2762   26   -25   841   69%  2626
  48 Shredder 6.0  256MB Athlon 1200 MHz     2710   29   -29   587   60%  2640
  54 Shredder 7.0 UCI 128MB K6-2 450 MHz     2684   24   -24   834   43%  2736
  62 Shredder 5.32  256MB Athlon 1200 MHz    2656   25   -25   779   50%  2657
  73 Shredder 6.0 UCI 128MB K6-2 450 MHz     2614   43   -43   264   52%  2603
  77 Shredder 5.32  128MB K6-2 450 MHz       2602   20   -20  1242   42%  2663
  81 Shredder 5.0  128MB  K6-2 450 MHz       2578   16   -16  1818   44%  2620
  83 Shredder 4.0  128MB  K6-2 450 MHz       2568   23   -22   986   58%  2508
 116 Shredder 3.0  64MB P200 MMX             2417   20   -20  1262   48%  2429
 120 Shredder 2.0  58MB P200 MMX             2402   19   -19  1314   44%  2441
 143 Shredder 1.0 Pentium 90 MHz             2290   55   -55   165   50%  2292

So maybe we just have to wait a little longer.


OK, OK. A lot of that is hardware.

[Dann Corbit]

Oops, make that +453 for Rybka.