About strelka1.8 sources

[Tord Romstad]

Well, _no one_ has it yet compared to the released version!

Huh? Dann has done so in this very thread. The scores, the PV and the node counts were identical for the released version and the binary Dann compiled himself.

Tord isn't it possible to get more logical again? Danns post which only
covers the start position BTW was _after_ my post.

OK, my mistake.

Second if we would deduce from one position it is the same as the
released 1.8 version months or whatever time ago, also fine, but what
about all the _hundreds positions_ which showed the _same_ output
than a certain Rybka version just with an added depth +2 and an exact
score added?

If you look back at my first message in this thread, you will see that I already said that I can't make any comments about similarities to Rybka. I have never seen Rybka's source code, I don't own a copy of Rybka (it doesn't exist for my OS of choice), and in fact I don't even think I have ever seen Rybka play.

I still stand by everything I have written in this thread. Strelka is not a clone of any chess program I know, and I feel confident that the source code I have is the real Strelka 1.8.

Tord

[jesper_nielsen]

<<<SNIP!>>>

I don't think DanChess belongs in that group. DanChess did borrow the SEE from crafty (albeit with changes to accomodate his data structures). However, that was an unintentional violation of what Dr. Hyatt considered acceptable. The other instances were total program snatches and stamping a new name on them (El Chinito did have some interesting ideas injected into it, but it was a clear 'program snatch').

Now, to step back, DanChess stepped over the line. But it was one toe over the line. The other two programs mentioned were 29' broad jumps.

Now you have me worried!

In Pupsi i use the SEE algorithm from Crafty.

I state in the readme, that this is so: "SEE evaluation. (As seen in Crafty)"

I was under the impression, that this was OK, but maybe i was wrong?!

Where does the line go?

Kind regards,
Jesper

[geots]

Chris, you're a great guy and i consider you a friend. But Hoffman, Dann, Uri, Alessandro and Tord have all said Strelka is not a clone. Hoffman and Dann, at least, were including Rybka in this also. So forgive me, but what ever your find is completely irrelevant to me. I shall go with what they say and keep testing it. I think all that can be said has been said- and im satisfied with what those 5 (i would call experts) say. Have a nice day and best regards.

George, only because you wrote it twice meanwhile, I want to tell you that
Alessandro does _not_ have Strelkas' source and he never gave that
impression in that thread.

Guenther

I stand by what i said, by your count for the 3rd time. I will continue to test Strelka. You have proven that people have tunnel vision and are zeroed in on Strelka to the point it is ridiculous. I mention that a commercial version will score worse in most tests than Strelka, and you completely choose to ignore that. Not even curious? Come, now?

It escapes me what my post has to do with your answer?
(moreover what has strength now to do with all of it? That is completely
irrelevant. As for the tunnel vision I guess I will add a post from yourself
3 months ago to show your motivation for it
I simply don't care and that's why I am just neutral on it, no more no less.

Code: Select all

George Speight July 10 2007:
----------------------------------
One thing for sure- no matter which side of the fence you are on, its a very interesting issue. IF and thats IF- it were an illegal clone, i stand to lose more than many- as im consuming a great amount of time running a Strelka Gauntlet. So naturally i have to assume it is NOT a clone. I dont have the expertise to do anything but read as much as possible and I assume it is NOT a clone. Again, just an assumption. The best i can do.
But i still dont see how Hoffman or anyone can make any assumption about it and Rybka without having Rybkas source code. No one has yet explained that to me. 

Guenther

Guenther, dont tell me that is the best you can do. You disappoint me. I certainly said that then, but a lot has been explained to me since then that i was not aware of. And, sigh, Jesus, I am not saying it is or is not a clone. You prove it is illegal- any more so than Toga. While you are proving all this- for the 4th time now by your count- i will continue to test it.

George, did anyone ever say you cannot test what you want?
Nothing in your post again has something to do with my post. I never
proved anything nor do I see proofs for the opposite.
I would wish people would write less emotional and step back to
a more logical approach, this helps to read what is between the lines not
what is not between them.


Guenther

In that case you are neutral and have no idea if Strelka is a clone or not. And if if is, you have no idea if it is illegal or not. You are just a neutral bystander. My apologies. I misinterpreted what you were saying. I am in the same boat. I have no idea either. That is why i rely on experts like Hoffman, Dann and Tord.

Best Regards

[Uri Blass]

<<<SNIP!>>>

I don't think DanChess belongs in that group. DanChess did borrow the SEE from crafty (albeit with changes to accomodate his data structures). However, that was an unintentional violation of what Dr. Hyatt considered acceptable. The other instances were total program snatches and stamping a new name on them (El Chinito did have some interesting ideas injected into it, but it was a clear 'program snatch').

Now, to step back, DanChess stepped over the line. But it was one toe over the line. The other two programs mentioned were 29' broad jumps.

Now you have me worried!

In Pupsi i use the SEE algorithm from Crafty.

I state in the readme, that this is so: "SEE evaluation. (As seen in Crafty)"

I was under the impression, that this was OK, but maybe i was wrong?!

Where does the line go?

Kind regards,
Jesper

The problem with DanChess was not only SEE based on my memory.

I remember that
DanChess took some evaluation code from Crafty(not as copy and paste but I remember cases when he used the same evaluation numbers
so it was not only copying the idea).
I remember that DanChess also used the same type of bitboards as Crafty.


Uri

[Dann Corbit]

<<<SNIP!>>>

I don't think DanChess belongs in that group. DanChess did borrow the SEE from crafty (albeit with changes to accomodate his data structures). However, that was an unintentional violation of what Dr. Hyatt considered acceptable. The other instances were total program snatches and stamping a new name on them (El Chinito did have some interesting ideas injected into it, but it was a clear 'program snatch').

Now, to step back, DanChess stepped over the line. But it was one toe over the line. The other two programs mentioned were 29' broad jumps.

Now you have me worried!

In Pupsi i use the SEE algorithm from Crafty.

I state in the readme, that this is so: "SEE evaluation. (As seen in Crafty)"

I was under the impression, that this was OK, but maybe i was wrong?!

Where does the line go?

Kind regards,
Jesper

The problem with DanChess was not only SEE based on my memory.

I remember that
DanChess took some evaluation code from Crafty(not as copy and paste but I remember cases when he used the same evaluation numbers
so it was not only copying the idea).
I remember that DanChess also used the same type of bitboards as Crafty.


Uri

Here is the DanChess controversy, relived in all its glory:
http://www.stmintz.com/ccc/index.php?id=349222

In summary the bones of contention were:
1. Bitboard used the same numbering scheme (A1 = 0...)
2. King tropism tables were identical
3. SEE used the same decisions.

Subsequent rewrites of DanChess removed every shred of crafty code and his latest programs are very un-crafty like.

In Daniel's case it was (IMO) innocent ignorance as to what constitues fair use. He did not try to hide the crafty association (but he did fail to give credit initially). He was a micron away from giving up chess programming completely because of his bitter dissappointment.

I guess that the whole DanChess misunderstanding would have been completely avoided by sending an email to Dr. Hyatt when Daniel first started the project and also by giving credit where credit is due in the readme file. (I think that Dr. Hyatt is a very fair person and also that he was really sick and tired of things like BIONIC {which I nicknamed 'Believe It Or Not It's Crafty!'} and Voyager [along with several others] which increased the strength of his response due to his irritation).

The bottom line (as I see it) is to keep things open and above board and give credit where credit is due. Ask permission before you do *anything* with someone else's code. These simple steps will prevent CCC tornados. At least (with my wishful thinking) I hope so.

[Graham Banks]

In Daniel's case it was (IMO) innocent ignorance as to what constitues fair use. He did not try to hide the crafty association (but he did fail to give credit initially). He was a micron away from giving up chess programming completely because of his bitter dissappointment.

I, for one, am very glad he didn't give up.
Scorpio is a very fine engine that goes from strength to strength.

Regards, Graham.