POLL: Do you believe Rybka is a derivative work program?

[M ANSARI]

This is silly because how far do you go back. Maybe back to C+ routines or assembler routines or whatever the first successful programming language was. Every program on earth uses pieces of older programs just like a new car designer will use the already very useful idea of using a round rubber object for tires. More interesting is what creative input the last person doing the final project manages to put into it. Rybka is head over shoulders superior to any engine out there, so that is plenty enough creativity for me.

[MirceaH]

AGAIN???

Stop this, please...

[Vladimir Xern]

This has been brought up several times before, fruitlessly (pun somewhat intended). The general consensus always seems to be that even if Rybka were a direct clone, no one would really care. Once you reach a high-enough level, there appears to be a certain exemption status that even comes with token apologists.

[Sean Evans]

Still 100 Elo difference, which is more then can be explained by speedup.

Would it be possible to use the source code of several programs to create the base and then add the author's own code to create one Frankenstein monster of a program that no one can beat?

For instance if I took Fruit as the base, sprinkled some Crafty and Rebel in it and some other various programs, one would think I would have a very strong program!

Seems the logical way to create a high end chess program in modern times, there is plenty of chess coding knowledge publicly available in 2008.

Cordially,

Sean

[Dann Corbit]

Still 100 Elo difference, which is more then can be explained by speedup.

Would it be possible to use the source code of several programs to create the base and then add the author's own code to create one Frankenstein monster of a program that no one can beat?

For instance if I took Fruit as the base, sprinkled some Crafty and Rebel in it and some other various programs, one would think I would have a very strong program!

Seems the logical way to create a high end chess program in modern times, there is plenty of chess coding knowledge publicly available in 2008.

Cordially,

Sean

It would be more difficult than it sounds.
The various parts of a chess engine (move generation, evaluation, search) typically interact with one another. The underlying formulations (array based, bitboard, containers and structures) vary from engine to engine.

On the other hand, it would be relatively easy to read code from open source engines and use their ideas.

The transplant of organs won't work well, but the blueprints can transfer fairly easily.

In fact, I think that is why we see so many strong engines today.

Something else that people should consider is the testing methodology of Dr. Hyatt. Crafty's strength is skyrocketing over the past several months due to a careful and deliberate effort to test all factets of chess calculation. Clearly, this idea works very well. If Crafty can gain a couple hundred Elo, so can everyone else.

There is the difficulty (however) that not everyone has a huge Beowulf cluster at their beck and call.

[gerold]

Do you believe Rybka is a derivative work program?

imho, concerning 1.0 beta, i think the evidence speaks for itself:
http://www.chesslogik.com/controversies/

but there are many who don't share this viewpoint...
so i have a better poll:

who the hell really cares?

the fact is that modern Rybka is incredible...a fantastic piece of programming and software engineering, and Vas should be commended for that.

Hi Norman. I have read all the testing reports etc. None prove
whether Rybka is a clone. I was one of the first to test 1.0 beta.
it was and still is diff. than any other eng.

Best to you,

Gerold.

[gerold]

Still 100 Elo difference, which is more then can be explained by speedup.

Would it be possible to use the source code of several programs to create the base and then add the author's own code to create one Frankenstein monster of a program that no one can beat?

For instance if I took Fruit as the base, sprinkled some Crafty and Rebel in it and some other various programs, one would think I would have a very strong program!

Seems the logical way to create a high end chess program in modern times, there is plenty of chess coding knowledge publicly available in 2008.

Cordially,

Sean

It would be more difficult than it sounds.
The various parts of a chess engine (move generation, evaluation, search) typically interact with one another. The underlying formulations (array based, bitboard, containers and structures) vary from engine to engine.

On the other hand, it would be relatively easy to read code from open source engines and use their ideas.

The transplant of organs won't work well, but the blueprints can transfer fairly easily.

In fact, I think that is why we see so many strong engines today.

Something else that people should consider is the testing methodology of Dr. Hyatt. Crafty's strength is skyrocketing over the past several months due to a careful and deliberate effort to test all factets of chess calculation. Clearly, this idea works very well. If Crafty can gain a couple hundred Elo, so can everyone else.

There is the difficulty (however) that not everyone has a huge Beowulf cluster at their beck and call.

I have diff. copies of Crafty from the last 5 years. It has improve
'some. 200 elo i don't know about that.

[Dann Corbit]

Still 100 Elo difference, which is more then can be explained by speedup.

Would it be possible to use the source code of several programs to create the base and then add the author's own code to create one Frankenstein monster of a program that no one can beat?

For instance if I took Fruit as the base, sprinkled some Crafty and Rebel in it and some other various programs, one would think I would have a very strong program!

Seems the logical way to create a high end chess program in modern times, there is plenty of chess coding knowledge publicly available in 2008.

Cordially,

Sean

It would be more difficult than it sounds.
The various parts of a chess engine (move generation, evaluation, search) typically interact with one another. The underlying formulations (array based, bitboard, containers and structures) vary from engine to engine.

On the other hand, it would be relatively easy to read code from open source engines and use their ideas.

The transplant of organs won't work well, but the blueprints can transfer fairly easily.

In fact, I think that is why we see so many strong engines today.

Something else that people should consider is the testing methodology of Dr. Hyatt. Crafty's strength is skyrocketing over the past several months due to a careful and deliberate effort to test all factets of chess calculation. Clearly, this idea works very well. If Crafty can gain a couple hundred Elo, so can everyone else.

There is the difficulty (however) that not everyone has a huge Beowulf cluster at their beck and call.

I have diff. copies of Crafty from the last 5 years. It has improve
'some. 200 elo i don't know about that.

Here is some data from:
http://www.computerchess.org.uk/ccrl/40 ... t_all.html

Code: Select all

Rank Name Rating Score Average Opponent Draws Games LOS  ELO + - 
Crafty 23.0 64-bit 4CPU 2852 +26 -26 41.8% +56.0 33.9% 499 49.7% 
Crafty 22.8 64-bit 2792 +56 -57 43.4% +42.4 35.7% 98 56.3% 
Crafty 22.4 32-bit 2691 +38 -38 48.0% +14.1 35.0% 220 49.8% 
Crafty 22.1 32-bit 2659 +28 -28 49.0% +5.7 30.1% 439 50.1% 
Crafty 21.6 32-bit 2647 +34 -34 48.1% +11.3 30.5% 295 50.3% 
Crafty 21.5 32-bit 2638 +33 -33 46.5% +26.5 28.7% 324 48.4% 
Crafty 20.14 32-bit 2628 +34 -34 41.9% +56.9 32.4% 296 54.7% 
Crafty 20.13 32-bit 2618 +36 -36 46.5% +26.0 35.4% 254 52.5% 
Crafty 20.11 32-bit 2594 +36 -36 48.2% +14.1 29.4% 272 50.7% 

And from CEGT:

Code: Select all

no Program Elo + - Games Score Av.Op. Draws 
245 Crafty 23.0 x64 2CPU 2734 19 19 750 48.1% 2748 42.0% 
278 Crafty 22.9 x64 2CPU 2690 27 27 447 40.6% 2756 32.4% 
300 Crafty 22.0 x64 2CPU 2670 32 32 331 41.5% 2729 28.7% 
310 Crafty Cito 1.4.2 x64 2CPU 2663 29 29 380 50.9% 2656 30.8% 
328 Crafty 21.6 x64 2CPU 2648 29 29 392 44.0% 2690 30.4% 
342 Crafty 20.14 x64 2CPU 2632 33 33 310 42.1% 2688 26.8% 
498 Crafty 17.14 2470 71 71 70 40.7% 2536 27.1% 

The latest crafty versions also scale remarkably well with CPU count.
So if you throw a big pile of CPUs at Crafty 23.0, you will get a serious powerhouse.

[Kirk]

How can it be derivative if it is the best in the world?

If there is a similar base to the building, it certainly looks different at the top

[Dann Corbit]

Still 100 Elo difference, which is more then can be explained by speedup.

Would it be possible to use the source code of several programs to create the base and then add the author's own code to create one Frankenstein monster of a program that no one can beat?

For instance if I took Fruit as the base, sprinkled some Crafty and Rebel in it and some other various programs, one would think I would have a very strong program!

Seems the logical way to create a high end chess program in modern times, there is plenty of chess coding knowledge publicly available in 2008.

Cordially,

Sean

It would be more difficult than it sounds.
The various parts of a chess engine (move generation, evaluation, search) typically interact with one another. The underlying formulations (array based, bitboard, containers and structures) vary from engine to engine.

On the other hand, it would be relatively easy to read code from open source engines and use their ideas.

The transplant of organs won't work well, but the blueprints can transfer fairly easily.

In fact, I think that is why we see so many strong engines today.

Something else that people should consider is the testing methodology of Dr. Hyatt. Crafty's strength is skyrocketing over the past several months due to a careful and deliberate effort to test all factets of chess calculation. Clearly, this idea works very well. If Crafty can gain a couple hundred Elo, so can everyone else.

There is the difficulty (however) that not everyone has a huge Beowulf cluster at their beck and call.

I have diff. copies of Crafty from the last 5 years. It has improve
'some. 200 elo i don't know about that.

Here is some data from:
http://www.computerchess.org.uk/ccrl/40 ... t_all.html

Code: Select all

Rank Name Rating Score Average Opponent Draws Games LOS  ELO + - 
Crafty 23.0 64-bit 4CPU 2852 +26 -26 41.8% +56.0 33.9% 499 49.7% 
Crafty 22.8 64-bit 2792 +56 -57 43.4% +42.4 35.7% 98 56.3% 
Crafty 22.4 32-bit 2691 +38 -38 48.0% +14.1 35.0% 220 49.8% 
Crafty 22.1 32-bit 2659 +28 -28 49.0% +5.7 30.1% 439 50.1% 
Crafty 21.6 32-bit 2647 +34 -34 48.1% +11.3 30.5% 295 50.3% 
Crafty 21.5 32-bit 2638 +33 -33 46.5% +26.5 28.7% 324 48.4% 
Crafty 20.14 32-bit 2628 +34 -34 41.9% +56.9 32.4% 296 54.7% 
Crafty 20.13 32-bit 2618 +36 -36 46.5% +26.0 35.4% 254 52.5% 
Crafty 20.11 32-bit 2594 +36 -36 48.2% +14.1 29.4% 272 50.7% 

And from CEGT:

Code: Select all

no Program Elo + - Games Score Av.Op. Draws 
245 Crafty 23.0 x64 2CPU 2734 19 19 750 48.1% 2748 42.0% 
278 Crafty 22.9 x64 2CPU 2690 27 27 447 40.6% 2756 32.4% 
300 Crafty 22.0 x64 2CPU 2670 32 32 331 41.5% 2729 28.7% 
310 Crafty Cito 1.4.2 x64 2CPU 2663 29 29 380 50.9% 2656 30.8% 
328 Crafty 21.6 x64 2CPU 2648 29 29 392 44.0% 2690 30.4% 
342 Crafty 20.14 x64 2CPU 2632 33 33 310 42.1% 2688 26.8% 
498 Crafty 17.14 2470 71 71 70 40.7% 2536 27.1% 

The latest crafty versions also scale remarkably well with CPU count.
So if you throw a big pile of CPUs at Crafty 23.0, you will get a serious powerhouse.

WBEC data:

Code: Select all

Rank Name                 Elo    +    - games score oppo. draws
Crafty 23.0-x64-2cpu      2641   73   73    52   65%  2552   54%
Crafty 22.8-x64-2cpu      2617   80   80    52   38%  2693   38%
Crafty 20.14-x64BH        2567   60   60    92   46%  2594   29%
Crafty 20.14BH-x64        2522   95   95    32   53%  2495   44%
Crafty 22.1-x64           2499  101  101    28   43%  2542   43%
Crafty-19.13-64           2493   57   57    92   56%  2452   47%
Crafty 19.20 x64BH        2485   42   42   183   43%  2536   32%
Crafty 20.13BH-x64        2481   59   59    88   62%  2406   40%
Crafty 19.20BH-x64        2476   97   97    32   50%  2481   38%
Crafty 19.12              2469   68   68    64   57%  2424   42%
Crafty 22.0-x64           2450   66   66    92   26%  2630   24%
Crafty 19.03              2434   59   59    84   59%  2376   46%
Crafty 19.15              2419   67   67    76   48%  2437   28%
Crafty 19.19-64BH         2416   59   59    92   39%  2490   37%
Crafty 19.01              2412   59   59   100   60%  2344   26%
Crafty 19.06              2381   66   66    68   46%  2407   43%
Crafty 18.12              2370   61   61    88   58%  2316   30%
Crafty 18.15              2324   67   67    76   53%  2308   26%