hgm wrote: ↑Fri Jul 23, 2021 4:59 pm
...
I suppose there still are the issues of:
1) Whether a NN is a computer program, and the weights are the corresponding machine code.
2) Whether this program counts as linked with a conventional executable when the latter uses the output of the NN it interprets for further calculation
3) What would count as the 'source code' of the NN in the GPL sense.
It will be interesting what the court decides on this.
I can imagine an outcome where (1) and (2) are decided as 'yes, it does', and (3) would be the set of training positions.
How I read the SF blog article it is also about false marketing, how to give proper credit to an GPL project, would be interesting to get a court decision on this too.
hgm wrote: ↑Fri Jul 23, 2021 5:22 pm
Is 'giving credit' a condition of the GPL?
From the GPL FAQ:
I want to get credit for my work. I want people to know what I wrote. Can I still get credit if I use the GPL?
You can certainly get credit for the work. Part of releasing a program under the GPL is writing a copyright notice in your own name (assuming you are the copyright holder). The GPL requires all copies to carry an appropriate copyright notice.
Fat Titz by Stockfish, the engine with the bodaciously big net. Remember: size matters. If you want to learn more about this engine just google for "Fat Titz".
As a chess player I fully understand where this guy is coming from.
"Good decisions come from experience, and experience comes from bad decisions."
__________________________________________________________________
Ted Summers
hgm wrote: ↑Fri Jul 23, 2021 6:35 pm
But I assume the Stockfish source at GitHub to which they refer does contain such a copyright notice.
The comment seem to be about giving credit during marketing. I don't hink the GPL puts any conditions on how you market the program.
To paraphrase the gentleman who earlier referred to “chess community standards,” he didn’t care if ChessBase’s behavior was technically legal if it went against his conception of correct behavior. This at least was honest, even if he got his facts a bit scrambled.
The issue in this discussion, and frankly any time the GPL is raised on this board, is that people try to make the GPL fit as a tool for enforcing their concept of correct behavior. This misconception isn’t unique to chess. Google “redhat gpl violation.” What you’ll get back is a bunch of opinions, most well over a decade old, fulminating against this commercial interloper on their turf.
Fat Titz by Stockfish, the engine with the bodaciously big net. Remember: size matters. If you want to learn more about this engine just google for "Fat Titz".
AlexChess wrote: ↑Fri Jul 23, 2021 9:24 am
Since it's impossible to reach Stockfish 14 (+100 ELO over commercials Dragon 2.0 and Fat Fritz 2.0, +200 ELO over Ethereal 13 NNUE) the only way to survive for Chessbase is to include Powerbooks 2022 & up to 7-man online tablebases on Fritz 18 if they want to sell their 20 years old GUI for 99€, not only the SF modified source code! How is possible that they cannot hire a top level developers team to create their own engine instead of simply rename "Fritz" the work of others???
You assume customers care only about elo.
It is possible to sell a new engine because it has some other relative advantages.
For example faster in finding mates relative to stockfish and relative to all other engines based on analysis of many winning positions
or can tell users if there is a forced draw for one side that is something I would like to have in an engine (0.00 evaluation is not equivalent to a forced draw).
Another option is being best in getting points against weak engines with some big handicap that is again something that is possible to test and maybe it is possible to have some engine that is 200 elo better than other engines when it start without queen d1(of course no possible points against strong engines but possible points maybe against tscp or weaker engines).
For that there are programs like SparkChess (perfect for 99% of chess players) . But a lot of people here, that like me are interested in engines matches, are looking only for the strongest ones
The question if there are tests that are not chess games when sparkChess is the strongest.
I would like to see the distance to mate against part of the army and the question is what is the best engine to do it and if some commercial engine can be signoificantly faster than what we have today based on tests with many positions.
examples:
[d]4k3/ppp3pp/8/8/8/8/PPPPPPPP/RNBQKBNR w KQ - 0 1 bm 1.e4 mate in 8
[d]4k3/pppp2pp/8/8/8/8/PPPPPPPP/RNBQKBNR w KQ - 0 1 bm 1.e4 mate in 9(do not know if there are more solutions but it is one of the solutions).
Edit:I do not know if sparkchess is better than stockfish10 in finding the shortest mate(not sure what is the best version of stockfish for finding mates but it is certainly not stockfish14)
I also do not know if there is a commercial program that is simply better than every free program in finding mates
but I see no reason to pay for something that is 50 elo weaker than stockfish and I think that people may prefer even to pay for something that is 500 elo weaker than stockfish if it is proved to be at least twice faster than all the other programs in finding mates based on tests.
dkappe wrote: ↑Fri Jul 23, 2021 8:06 pm
...he didn’t care if ChessBase’s behavior was technically legal if it went against his conception of correct behavior.
You can even almost ignore the "correctness" thing it just casts doubt about the decision makers most basic competence. If you rely on the work of foss developers and release a product that is overwhelmingly built on their achievement it stands to reason you should give them appropriate credit and fairly represent the product even if not strictly speaking legally required. Just because you don't want to piss off the people you rely on to do all the heavy lifting for you... for free.
It's just plain stupid behavior, on an institutional level. If no one of the many people that must be in the decision chain raised concerns about this it speaks volumes about the culture of the company. Being a software company that disrespects the work of software developers is a really bad look.
Last edited by Jakob Progsch on Sat Jul 24, 2021 12:51 am, edited 1 time in total.
dkappe wrote: ↑Fri Jul 23, 2021 8:06 pm
...he didn’t care if ChessBase’s behavior was technically legal if it went against his conception of correct behavior.
You can even almost ignore the "correctness" thing it just casts doubt about the decision makers most basic competence. If you rely on the work of foss developers and release a product that is overwhelmingly built on their achievement it stands to reason you should give them appropriate credit even if not strictly speaking legally required. Just because you don't want to piss off the people you rely on to do all the heavy lifting for you... for free.
It's just plain stupid behavior, on an institutional level. If no one of the many people that must be in the decision chain raised a red flag about this it speaks volumes about the culture of the company.
This has already been covered. Stockfish was prominently mentioned in the marketing material.
Fat Titz by Stockfish, the engine with the bodaciously big net. Remember: size matters. If you want to learn more about this engine just google for "Fat Titz".
