Something very strange [Strelka]

[Christopher Conkie]

But I gather Vas's main concern was with "emater.c", which has a lot of data tables. If these also exist in Rybka I would be concerned, too.

Osipov words are:

If he means the data tables from emater.c file, there is no such data tables in Rybka -- I have checked it. And everyone can check it scanning Rybka's exe.

This does not prove nothing.
It is possible that these data tables are derived from rybka.
Uri

Can you prove it? It is also possible that I am Darth Vader. Or Princess Leia.

Vasik can prove it in the right place at the right time.

[Andrej Sidorov]

Vasik can prove it in the right place at the right time.

What an argument! :lol::lol:

[Jorge Garcia de Andres]

Hi all, it's my first post on this forum, although I readed frequently int the past as guest,

I've passed today working on the source of strelka, trying to compile and making changes to achieve it. Finally I got it, I do not know why but if you open the dsw in vc++ 6.0 gives a lot of compilation errors, although I edit the dsp and the dsw and it says that it's compiled with vc++ 6.0; anyway I make my own compilation and now it works.

I am very familiar with the toga code because I've passed many hours trying to enpower it. Looking at the code of strelka it feels very familiar for me, I really think that osipovich used fruit code as skeleton. Indeed strelka it's pretty elaborated, it's simplified I mean that it's only what it's necessary, and the board representation and the code relationed it's masked. So it's really faster, aproxymately doubled.

The evaluation is different, introduces the emater table really difficult to understand, a lot of values that you don't know what are doing, but one thing I feel interesting it's the way material calculation occurs

summ = wq +
bq * 2 +
wr * 2*2 +
br * 2*2*3 +
wb * 2*2*3*3 +
bb * 2*2*3*3*3 +
wn * 2*2*3*3*3*3 +
bn * 2*2*3*3*3*3*3 +
wp * 2*2*3*3*3*3*3*3 +
bp * 2*2*3*3*3*3*3*3*9;

I remember a post on kasparov chess forum, about a guy with alias rybkin
I do feel it's another Osipovich alias, he said that he decompiled rybka 1.0
and was talking about the way of indexing the famous supposed table of material imbalances.

Anyway, to sum up I really think strelka it's a clone pretty elaborated of fruit with other influences rybka and beowulf, for sure.

[Christopher Conkie]

Hi all, it's my first post on this forum, although I readed frequently int the past as guest,

I've passed today working on the source of strelka, trying to compile and making changes to achieve it. Finally I got it, I do not know why but if you open the dsw in vc++ 6.0 gives a lot of compilation errors, although I edit the dsp and the dsw and it says that it's compiled with vc++ 6.0; anyway I make my own compilation and now it works.

I am very familiar with the toga code because I've passed many hours trying to enpower it. Looking at the code of strelka it feels very familiar for me, I really think that osipovich used fruit code as skeleton. Indeed strelka it's pretty elaborated, it's simplified I mean that it's only what it's necessary, and the board representation and the code relationed it's masked. So it's really faster, aproxymately doubled.

The evaluation is different, introduces the emater table really difficult to understand, a lot of values that you don't know what are doing, but one thing I feel interesting it's the way material calculation occurs

summ = wq +
bq * 2 +
wr * 2*2 +
br * 2*2*3 +
wb * 2*2*3*3 +
bb * 2*2*3*3*3 +
wn * 2*2*3*3*3*3 +
bn * 2*2*3*3*3*3*3 +
wp * 2*2*3*3*3*3*3*3 +
bp * 2*2*3*3*3*3*3*3*9;

I remember a post on kasparov chess forum, about a guy with alias rybkin
I do feel it's another Osipovich alias, he said that he decompiled rybka 1.0
and was talking about the way of indexing the famous supposed table of material imbalances.

Anyway, to sum up I really think strelka it's a clone pretty elaborated of fruit with other influences rybka and beowulf, for sure.

You know something?

This is the most intelligent posting i've seen in a while. Your first post and you looked at much that has gone before no doubt. Interesting how the CCC is percieved right now from outside.

You mean this.....

http://kasparovchess.crestbook.com/view ... hp?id=1490

Thank you and on behalf of all sane people, welcome to the forum.

Regards

Christopher

[Gerd Isenberg]

But I gather Vas's main concern was with "emater.c", which has a lot of data tables. If these also exist in Rybka I would be concerned, too.

Osipov words are:

If he means the data tables from emater.c file, there is no such data tables in Rybka -- I have checked it. And everyone can check it scanning Rybka's exe.

Hi Andrej,

On the matter of Strelka - I see no reason to don't trust Vasik's statement. Vast parts, as he stated, does of course not imply all of the Strelka source. "Rybkin" explained some disassembled internals about Rybka in in the Kasparov chess programming forum. Some time later Osipov's Strelka appeared with all the consequences until now.

Despite of legal issues, I have strong objections against reverse engineering commercial (or other closed source) programs (clone checking is another issue of course) to extract ideas - even early betas, without explicit license and copyright state. The source is published now - and I am torn between the great Strelka source and the possible damage against a fellow commercial colleague.

Does Osipov argue Strelka does not contain ideas based on reverse engineered parts of Rybka?

Thanks,
Gerd

[Dann Corbit]

Evgenii, your question (implying vas is unethical)

Joseph,

what my question is implying is a pure speculation of yours. That was a re-action of untrue statement of Gerold Daniels /that the reason of "negativity" to Vas is that he's the best/.
From your post I see you asked yourself this question but, I think, you didn't liked the answer you gave.

Regards,
Geno

Actually, I thought his post was very pertinent. Vas hasn't done anything unethical, yet you imply he has, or that there is 'negativity' attached to his name. There isn't other than the the grumbling of his amazing domination of the computer chess scene, and the sometimes excessive posts regarding Rybka.

As an aside, you mention SMK, and while it is absolutely true he was also number one for years, he never came close (no disrespect to him and his formidable program) to the sheer difference in Elo between his number one and the number two, much less increased or held such a huge edge for so much time.

Albert

I disagree that Vas did nothing unethical
Displaying misleading information about nodes per second is considered by me as something unethical.

Note that I do not think that he did something illegal and I do not think that he stole code from fruit.

Uri

I think that 'unethical' is too strong of a word here.
Vas can choose any definition for node that he likes. Junior uses different ideas for reporting depth, for instance. If the definition is a little different it would be nice for him to publish his definition, but I do not think that he has to do that either.

[Uri Blass]

Evgenii, your question (implying vas is unethical)

Joseph,

what my question is implying is a pure speculation of yours. That was a re-action of untrue statement of Gerold Daniels /that the reason of "negativity" to Vas is that he's the best/.
From your post I see you asked yourself this question but, I think, you didn't liked the answer you gave.

Regards,
Geno

Actually, I thought his post was very pertinent. Vas hasn't done anything unethical, yet you imply he has, or that there is 'negativity' attached to his name. There isn't other than the the grumbling of his amazing domination of the computer chess scene, and the sometimes excessive posts regarding Rybka.

As an aside, you mention SMK, and while it is absolutely true he was also number one for years, he never came close (no disrespect to him and his formidable program) to the sheer difference in Elo between his number one and the number two, much less increased or held such a huge edge for so much time.

Albert

I disagree that Vas did nothing unethical
Displaying misleading information about nodes per second is considered by me as something unethical.

Note that I do not think that he did something illegal and I do not think that he stole code from fruit.

Uri

I think that 'unethical' is too strong of a word here.
Vas can choose any definition for node that he likes. Junior uses different ideas for reporting depth, for instance. If the definition is a little different it would be nice for him to publish his definition, but I do not think that he has to do that either.

The main problem is with giving misleading information.

I do not think that he has to report true nodes per seconds but I think that it is better even not to give that information or give information that
is not misleading(for example writing constant number of nodes that is not dependent on time position or hardware) and not
to give misleading information.

In the case of Junior there was no problem of misleading information because it was obvious from the start that Junior's depth is different.

Junior is the only engine that I know that does not increase the depth by 1 every iteration and the iteration that it does are dependent on the time control.

It starts by increasing the depth by 3 every iteration and only in the last iterations it can increase the depth by 2 or by 1.

Uri
Uri

[Gerd Isenberg]

Hi all, it's my first post on this forum, although I readed frequently int the past as guest,

I've passed today working on the source of strelka, trying to compile and making changes to achieve it. Finally I got it, I do not know why but if you open the dsw in vc++ 6.0 gives a lot of compilation errors, although I edit the dsp and the dsw and it says that it's compiled with vc++ 6.0; anyway I make my own compilation and now it works.

I am very familiar with the toga code because I've passed many hours trying to enpower it. Looking at the code of strelka it feels very familiar for me, I really think that osipovich used fruit code as skeleton. Indeed strelka it's pretty elaborated, it's simplified I mean that it's only what it's necessary, and the board representation and the code relationed it's masked. So it's really faster, aproxymately doubled.

The evaluation is different, introduces the emater table really difficult to understand, a lot of values that you don't know what are doing, but one thing I feel interesting it's the way material calculation occurs

summ = wq +
bq * 2 +
wr * 2*2 +
br * 2*2*3 +
wb * 2*2*3*3 +
bb * 2*2*3*3*3 +
wn * 2*2*3*3*3*3 +
bn * 2*2*3*3*3*3*3 +
wp * 2*2*3*3*3*3*3*3 +
bp * 2*2*3*3*3*3*3*3*9;

I remember a post on kasparov chess forum, about a guy with alias rybkin
I do feel it's another Osipovich alias, he said that he decompiled rybka 1.0
and was talking about the way of indexing the famous supposed table of material imbalances.

Anyway, to sum up I really think strelka it's a clone pretty elaborated of fruit with other influences rybka and beowulf, for sure.

Hi Jorge Garcia,

welcome to CCC!

Yes, this is a beautiful piece of code. I was not aware of that math.
I also wonder who the inventor of these index mapping is.
Less populated, but more valuable pieces use the less significant bits.

I still use five bytes as 10 nibble-array as part of my position structure - incremental updated during make/unmake - and use some 64-bit modulo (by imul) to get an index inside a relative small material-hashtable, which seems not that bad either.

Cheers,
Gerd

[Dann Corbit]

My question is about concept:

Is only the same evaluation code in two engines sufficient to say:
"one of these engines is a clone" ?

Is only the same search function code in two engines sufficient to say:
"one of these engines is a clone" ?

I have never heard about such criterion.

For most people, a "clone" is an engine where the code or parts of the code is copied. So if one took parts of code, the word "clone" is not quite correct.

It is well known that many engines share the same move generation code, and no one claim they are clones.

If one took "only" the movegeneration of an open source engine he still has to follow the licence. That means if you use only the movegen of Fruit, you have to publish your engine under GPL with sources.

For the use of Crafty code you need the permission of Robert Hyatt, to publish the engine or participate on tournaments afaik.

Alex

There is also an important distinction between a legal clone and an illegal one.

Toga is a perfectly legal clone of Fruit, and the license of Toga follows the intentions of Fabian.

Other clones (e.g. BIONIC) never got permission and violated copyright and/or license agreements and so are actually law violations. These are clearly invalid or illegal clones.

I think that there is another area which is "grey" or "fuzzy" clones. Did they do something illegal? Did they violate some license? We are not sure.

Perhaps a 4th area might by "dirty" clones. They have no legal violation in the final analysis but they did something we don't like.

In the example of Toga or other clones where the author is in full compliance with the license agreement, I see nothing negative about it.
For the other types, there is something we don't like about it, but it may be only opinions in some cases without legal basis or we may be unsure about the legality of something.

[Marek Soszynski]

There is also an important distinction between a legal clone and an illegal one.

Toga is a perfectly legal clone of Fruit, and the license of Toga follows the intentions of Fabian.

Other clones (e.g. BIONIC) never got permission and violated copyright and/or license agreements and so are actually law violations. These are clearly invalid or illegal clones.

I think that there is another area which is "grey" or "fuzzy" clones. Did they do something illegal? Did they violate some license? We are not sure.

Perhaps a 4th area might by "dirty" clones. They have no legal violation in the final analysis but they did something we don't like.

In the example of Toga or other clones where the author is in full compliance with the license agreement, I see nothing negative about it.
For the other types, there is something we don't like about it, but it may be only opinions in some cases without legal basis or we may be unsure about the legality of something.

All the legal complications are compounded by the fact that laws apply in particular jurisdictions only. Particular laws in the European Community may differ subtly from those in North America which may in turn differ greatly from those in some other parts of the world. I hesitate to add that the fine details of ethical codes are not universal either.

What the chess computer community needs to agree on is a definition and classification of clones. Whatever a biological clone is, or a copy of a work of art is, is somebody else's concern. Our concern is with chess engines. We should lay down our own moral code in this, our own area. And to do that we must start by defining terms. Let there be a Cloning Commission whose task is to look deeply into the subject, and then produce a report and recommendations.