Raubfisch

[kramnik]

I am still unanswered. I want to point out that I am NOT the developer of this chess engine and that I host it on my website https://chess.massimilianogoi.com only because I found Raubfisch a very strong engine.
Besides I don't understand all these whinings about being "illegal" while Kraken (it should be the author's nickname) does not get profit at all from it... I've written Max Fehler and Karl Werner on the Raubfisch page just because they were cited in the readme file of its package (and they still are). I'm not a German, so I don't know to which words joke you are referring to about the names.

[Dann Corbit]

The tricky part to understand about it being illegal is that they are breaking the law. I don't know why it is tricky but it seems to be hard to comprehend.
That's all you need to understand.
The engine is a GPL engine.
The license for the use of this engine says that you must share the code changes you make.
They authors did not do that.

You might think that is petty, but it is not.
There are hundreds of workers who have toiled for thousands of hours writing and improving Stockfish.
Have a look at the fishtest framework and the group called "fishcooking" and you will see that there is a whole army dedicated to it.

And they only thing they as back for this multi-million dollar effort (literally) is that you honor their agreement if you are going to use the free tool.

The Raubfish people did not do that. That is literally a crime (not just some opinion of mine, they violated the agreement and could be arrested, fined, jailed, whatever) and they would deserve it.

[Dann Corbit]

I've written Max Fehler and Karl Werner on the Raubfisch page just because they were cited in the readme file of its package (and they still are). I'm not a German, so I don't know to which words joke you are referring to about the names.

Fehler is "mistake" in German

[Dann Corbit]

I've written Max Fehler and Karl Werner on the Raubfisch page just because they were cited in the readme file of its package (and they still are). I'm not a German, so I don't know to which words joke you are referring to about the names.

Fehler is "mistake" in German

Reminds me of a gag pulled on my by Martin Borriss, the author of Gullydeckel. He told me it was his brother's name, but it means "Manhole cover".

[jdart]

The Raubfish people did not do that. That is literally a crime (not just some opinion of mine, they violated the agreement and could be arrested, fined, jailed, whatever) and they would deserve it.

One of several reasons I am not that fond of GPL is that it is not readily enforceable in cases like this. I don't think the Stockfish team has the resources to take the Raubfisch authors to court, especially if they are foreign nationals. Plus they probably couldn't recover damages. It is not clear what monetary harm they suffered (I don't condone license violations and don't say they suffered no harm; but how to quantify it, for a non-commercial project?).

--Jon

[Stephen Ham]

Dear gents,

There's a lot not to like, and more to be suspicious about Raubfisch, including fake/joke names of the author(s) and engine versions, plus violation of the GPL.

Nonetheless, my match between the latest iterations of Raubfisch (RF) and Stockfish (SF) (see previous posts for match specs) has extended to 110 games. So far, RF leads with 9 victories to 7. So, just on the basis of performance, RF is more than holding its own against the latest SF after 110 games.

Again, one could argue that RF's success is due to actually being SF. But RF is larger - or is the additional size just meaningless "fluff" to disguise this? However, it has a slightly different move selection than SF and a clearly different evaluation function, which I think is more stable and accurate.

I'll let the match continue a bit further and will report the result here.

All the best,
-Steve-

[Dann Corbit]

Dear gents,

There's a lot not to like, and more to be suspicious about Raubfisch, including fake/joke names of the author(s) and engine versions, plus violation of the GPL.

Nonetheless, my match between the latest iterations of Raubfisch (RF) and Stockfish (SF) (see previous posts for match specs) has extended to 110 games. So far, RF leads with 9 victories to 7. So, just on the basis of performance, RF is more than holding its own against the latest SF after 110 games.

Again, one could argue that RF's success is due to actually being SF. But RF is larger - or is the additional size just meaningless "fluff" to disguise this? However, it has a slightly different move selection than SF and a clearly different evaluation function, which I think is more stable and accurate.

I'll let the match continue a bit further and will report the result here.

All the best,
-Steve-

They may actually have some innovation that adds value. However, it takes about 1000 games to get a feel for which engine is really stronger. I have had test runs where a change started at +100 Elo after a few hundred games and ended at -100 Elo (roughly) after a thousand.

If the date on your Raubfisch is really new, it might essentially be Stockfish. It is also possible that they really fixed an important bug or added an important eval term or improved a search parameter.

[kramnik]

In order to legally and officially claim that a program belongs to GPL you must prove it. It's not enough the ...fisch suffix. And to do so you should have its code, that is a paradox since it wasn't released. Raubfisch is at the head of downloads on my website, with more than 4.000 downloads. This means that is liked by folks. At the second place there is Stockfish Polyglot, which source lay on GitHub, then Cfish, on GitHub too and Brainfish, which code lay on the developer's site https://www.zipproth.de. I'm always very careful to follow rules/licenses/laws and requests when I upload something on my website.

About Raubfisch, what should we do, should we prevent any forum (i.e. https://mzchessforum.altervista.org/vie ... =75&t=1806) to mirror it too? Should we prevent Dropbox to host its packages?..... For what regards a potential economical damage, well it's equal to 0, since the Stockfish source is free, provided that Raubfisch derives from Stockfish. The only issues here could be ethical, but to the accusers the duty to prove that Raubfisch derives from Stockfish. Legally, a simple match of their analysis lines is not enough.

[Guenther]

In order to legally and officially claim that a program belongs to GPL you must prove it. It's not enough the ...fisch suffix. And to do so you should have its code, that is a paradox since it wasn't released. Raubfisch is at the head of downloads on my website, with more than 4.000 downloads. This means that is liked by folks. At the second place there is Stockfish Polyglot, which source lay on GitHub, then Cfish, on GitHub too and Brainfish, which code lay on the developer's site https://www.zipproth.de. I'm always very careful to follow rules/licenses/laws and requests when I upload something on my website.

About Raubfisch, what should we do, should we prevent any forum (i.e. https://mzchessforum.altervista.org/vie ... =75&t=1806) to mirror it too? Should we prevent Dropbox to host its packages?..... For what regards a potential economical damage, well it's equal to 0, since the Stockfish source is free, provided that Raubfisch derives from Stockfish. The only issues here could be ethical, but to the accusers the duty to prove that Raubfisch derives from Stockfish. Legally, a simple match of their analysis lines is not enough.

yawn...it is since long proven that it is illegally derived from SF, no code needed, there are other tools.
You are a typical denier, already added you to my ignore list since post one.

BTW it is on a ridiculous level to think people here conclude anything from a 'fish' suffix.

[kramnik]

yawn...it is since long proven that it is illegally derived from SF, no code needed, there are other tools.

Proven by what? Suppositions are worthless in a trial.