Last Rybka Thread for now...

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

Moderators: hgm, Dann Corbit, Harvey Williamson

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
Post Reply
Alexander Schmidt
Posts: 1100
Joined: Thu May 10, 2007 12:49 pm

Last Rybka Thread for now...

Post by Alexander Schmidt » Fri Aug 29, 2008 11:59 am

...hopefully :)

There where many aggressive posts in the last days, nice to see that people calm down now (including me...). We have seen a lot of things, some hidden in the middle of a thread, so probably most people didn't recognize it.

I try to merge the facts.

I think people showed that there must be a relation between Fruit and Rybka. There where different attempts and both had the same result: Very strange similaries between Rybka 1.0 Beta and Fruit 2.1.

Based on decompiling Rybka and compareing with Fruit:

http://pagesperso-orange.fr/ct_chess/Fr ... rt_go.html

And based on the output of Rybka and Fruit:

http://talkchess.com/forum/viewtopic.ph ... 93&t=23029

The GPL simply doesn't allow to copy a single line of GPL code or to start a new project with GPL code. All this work must be released under GPL too. So the question is not: "how much or which code is taken" from a GPL programm, the question is: "Is there any code taken from a GPL program."

Of course, there are much differences between Rybka and Fruit. There are lots of possible tests, and regarding the chessplaying, Rybka and Fruit are different. Different Movegen, different search, different evaluation.

There is no doubt that Rybka is an awesome engine. Vas found something that makes Rybka stronger than all others. But the question is, did Vas take some help by Fruit, this would be illegal and would have given him an advantage compared to other programmers. And with that he would also have harmed the others.

Now everyone can decide for himself if the shown facts are enough. The only thing we can do now is waiting for more facts.

Let me add it is not fair to say, some are jealous because they cannot create such a strong engine. It's always a question of how many time you spend in it, and you must have the one genuine idea that makes the difference.

People supporting the Open Source idea try to defend it. Thanks to Zach and others for their hard work.

chrisw

Re: Last Rybka Thread for now...

Post by chrisw » Fri Aug 29, 2008 12:17 pm

Good try, but this data is no data at all to prove anything.

The first link is a spreadsheet with some parameter passing code from the UCI. All engines have to have this stuff. The code shown is is not parallel anyway between the two programs. It's just basic, simple get parameters from the user-interface, like the time control and ply depth blah-blah. Irrelevent, could have been sourced from anywhere, no indication of code copying, and certainly absolutely nothing at all to do with the engine AI.

The second link is to data passed back to the interface. All programs pass data backwards and forwards to the interface and in many ways the data that gets passed is forced, else the interface won't work. It also proves precisely nothing and has absolutely nothing at all to do with the engine AI.

Your analysis of the GPL is also faulty. GPL prohibits use of code 'portions' but it does not define what a 'portion' is. To go to court with the GPL in mind requires in any case a substantial and significant portion of meaningful code that does something important of consequence and can't possibly have been sourced elsewhere, neither by the alleged copying program nor by the original GPL program.

Indeed, everyone can decide for himself if the shown facts are enough.

No, they are neither enough nor in any way meaningful in the first place.

setjmp() ? The judge will tell you to go setjmp() yourself in a lake ;-)

Alexander Schmidt wrote:...hopefully :)

There where many aggressive posts in the last days, nice to see that people calm down now (including me...). We have seen a lot of things, some hidden in the middle of a thread, so probably most people didn't recognize it.

I try to merge the facts.

I think people showed that there must be a relation between Fruit and Rybka. There where different attempts and both had the same result: Very strange similaries between Rybka 1.0 Beta and Fruit 2.1.

Based on decompiling Rybka and compareing with Fruit:

http://pagesperso-orange.fr/ct_chess/Fr ... rt_go.html

And based on the output of Rybka and Fruit:

http://talkchess.com/forum/viewtopic.ph ... 93&t=23029

The GPL simply doesn't allow to copy a single line of GPL code or to start a new project with GPL code. All this work must be released under GPL too. So the question is not: "how much or which code is taken" from a GPL programm, the question is: "Is there any code taken from a GPL program."

Of course, there are much differences between Rybka and Fruit. There are lots of possible tests, and regarding the chessplaying, Rybka and Fruit are different. Different Movegen, different search, different evaluation.

There is no doubt that Rybka is an awesome engine. Vas found something that makes Rybka stronger than all others. But the question is, did Vas take some help by Fruit, this would be illegal and would have given him an advantage compared to other programmers. And with that he would also have harmed the others.

Now everyone can decide for himself if the shown facts are enough. The only thing we can do now is waiting for more facts.

Let me add it is not fair to say, some are jealous because they cannot create such a strong engine. It's always a question of how many time you spend in it, and you must have the one genuine idea that makes the difference.

People supporting the Open Source idea try to defend it. Thanks to Zach and others for their hard work.

rebel777

Re: Last Rybka Thread for now...

Post by rebel777 » Fri Aug 29, 2008 12:24 pm

Alexander,

This won't do. Please provide:

1) the disassembler you used, preferable with download link;

2) The offsets in both executables of the alleged similar code chunks so a few of us can check ourselves.

3) List the assembler code of both chunks. Are they 100% identical? I would like to see with my own eyes.

Also consider:

1) Have you checked Fruit with other engines? Maybe such similarities are not so uncommon at all? What if you find similar chunks in Fritz, Shredder, Tiger, Hiarcs etc. ?

2) The fact that you don't know if there possible is an agreement between Vas and Fabien that allows Vas to use Fruit code and that both gentlemen also decided to keep that silent. Maybe Vas even paid Fabien for the service. Seems such a natural explanation and if true you better stop your research. Why not research this as first?

Bottom line, you are far from having proofed anything.

Awaiting regards,

Ed

Alexander Schmidt
Posts: 1100
Joined: Thu May 10, 2007 12:49 pm

Re: Last Rybka Thread for now...

Post by Alexander Schmidt » Fri Aug 29, 2008 12:42 pm

rebel777 wrote:1) Have you checked Fruit with other engines?
Regarding the output: Yes. Wach at my link.
rebel777 wrote:2) The fact that you don't know if there possible is an agreement between Vas and Fabien
Yes, I said this some days ago. It's possible. We have to wait what happens now :)

Uri Blass
Posts: 8948
Joined: Wed Mar 08, 2006 11:37 pm
Location: Tel-Aviv Israel

Re: Last Rybka Thread for now...

Post by Uri Blass » Fri Aug 29, 2008 12:47 pm

<snipped>
Alexander Schmidt wrote:
I think people showed that there must be a relation between Fruit and Rybka.
Of corse there is a relation
People learn from idea of other people.

There is a relation between most modern chess programs and fruit
There where different attempts and both had the same result: Very strange similaries between Rybka 1.0 Beta and Fruit 2.1.
Nothing strange
Alexander Schmidt wrote: There is no doubt that Rybka is an awesome engine. Vas found something that makes Rybka stronger than all others. But the question is, did Vas take some help by Fruit, this would be illegal and would have given him an advantage compared to other programmers. And with that he would also have harmed the others.
I believe that the answer is at least partly negative
I believe that part of the other also took from gpl programs but had not success like Vas.

Uri

Alexander Schmidt
Posts: 1100
Joined: Thu May 10, 2007 12:49 pm

Re: Last Rybka Thread for now...

Post by Alexander Schmidt » Fri Aug 29, 2008 12:51 pm

chrisw wrote:The first link is a spreadsheet with some parameter passing code from the UCI. All engines have to have this stuff.
Yes, all engines have it. But at least in a different order. This looks very similar. If this is not enough for you, OK.
chrisw wrote:and certainly absolutely nothing at all to do with the engine AI.
I gonna repeat myself :) The GPL simply doesn't allow to copy a single line of GPL code or to start a new project with GPL code. All this work must be released under GPL too. So the question is not: "how much or which code is taken" from a GPL programm, the question is: "Is there any code taken from a GPL program." It is not relevant which or how much code from a GPL program is taken
chrisw wrote:The second link is to data passed back to the interface. All programs pass data backwards and forwards to the interface and in many ways the data that gets passed is forced, else the interface won't work.
I think I elaborately explained why the similaries are suspicious. I even showed other engineoutput. Take some time and read it. If it is not enough for you, OK.
chrisw wrote:Your analysis of the GPL is also faulty.
No :)

chrisw

Re: Last Rybka Thread for now...

Post by chrisw » Fri Aug 29, 2008 1:42 pm

Alexander Schmidt wrote:
chrisw wrote:The first link is a spreadsheet with some parameter passing code from the UCI. All engines have to have this stuff.
Yes, all engines have it. But at least in a different order. This looks very similar. If this is not enough for you, OK.
chrisw wrote:and certainly absolutely nothing at all to do with the engine AI.
I gonna repeat myself :) The GPL simply doesn't allow to copy a single line of GPL code or to start a new project with GPL code. All this work must be released under GPL too. So the question is not: "how much or which code is taken" from a GPL programm, the question is: "Is there any code taken from a GPL program." It is not relevant which or how much code from a GPL program is taken
chrisw wrote:The second link is to data passed back to the interface. All programs pass data backwards and forwards to the interface and in many ways the data that gets passed is forced, else the interface won't work.
I think I elaborately explained why the similaries are suspicious. I even showed other engineoutput. Take some time and read it. If it is not enough for you, OK.
chrisw wrote:Your analysis of the GPL is also faulty.
No :)
If any code is taken it is certainly relevent what and how much.

Defence lawyer will demand the following:

1. Prove the code in the GPL program is original and not available anywhere else. For the UCI code or substantial chunks of it that is not going to be easy or possible.

2. Demonstrate the code is non-trivial. What means non-trivial?

3. Demonstrate the code chunk is substantive. What means substantive?

4. Account for the differences contained within the code chunks (you posted already comments that Fruit does so and so, Rybka not within the code chunks)

5. Account for the massive non-similarities, 99.99999% of the programs

6. Explain why you complaint is not vexatious. Defence will declare it is, and provide plenty of CCC posts to demonstrate vexatiousness.

7. Explain why your case should not be thrown right out of court with demand for costs.

Alexander Schmidt
Posts: 1100
Joined: Thu May 10, 2007 12:49 pm

Re: Last Rybka Thread for now...

Post by Alexander Schmidt » Fri Aug 29, 2008 1:57 pm

chrisw wrote:7. Explain why your case should not be thrown right out of court with demand for costs.
I don't plan to go to court with it. I think you missunderstood my first post, everyone shall decide for himself if it is enough or not. I will not convince you, and you will not convince me.

chrisw

Re: Last Rybka Thread for now...

Post by chrisw » Fri Aug 29, 2008 2:02 pm

Alexander Schmidt wrote:
chrisw wrote:7. Explain why your case should not be thrown right out of court with demand for costs.
I don't plan to go to court with it. I think you missunderstood my first post, everyone shall decide for himself if it is enough or not. I will not convince you, and you will not convince me.
Fair enough, I understand. But.

If you talk GPL licence, you talk legal. If you talk legal then it's fair to step ahead into a possible legal process. I argue that the thought experiment of stepping into legal process is going to be enough to make anyone recoil from it. Hence the GPL licence talk is weakened to zero as the GPL, is basically unenforcable even if the breaches you argue have taken place.

bnemias
Posts: 373
Joined: Thu Aug 14, 2008 1:21 am
Location: Albuquerque, NM

Re: Last Rybka Thread for now...

Post by bnemias » Fri Aug 29, 2008 2:06 pm

chrisw wrote:If any code is taken it is certainly relevent what and how much.
Only to the extent of obtaining needed proof. Ethically, "what" and "how much" are irrelevant.
Defence lawyer will demand the following:

1. Prove the code in the GPL program is original and not available anywhere else. For the UCI code or substantial chunks of it that is not going to be easy or possible.
I doubt it. By the time he gets to his defense, it will have presumably already been established that the defendant actually has copied code, presumably GPL'd code. If he borrowed it from elsewhere, it's up to the defense to show that.
6. Explain why you complaint is not vexatious. Defence will declare it is, and provide plenty of CCC posts to demonstrate vexatiousness.
Very funny. If the defense uses CCC posts to demonstrate anything, then it has already lost. Now, I looked up Vexatious litigation because I didn't know what it was. I think you should too, because you clearly don't.
7. Explain why your case should not be thrown right out of court with demand for costs.
Again, the judge would have already thrown out the case if this had merit. This would be argued before and after the case, not during.

Post Reply