'Deus X' Unveiled as FAT FRITZ PROJECT...

[ChickenLogic]

Actually I think it has not much sense to discuss with you or Silver, you showed already the same behaviour
after you were criticized after DeusX for not acknowleding the effort of LC0.

Why should I have to acknowledge anything about DeusX or lc0? DeusX has nothing to do with me, other than that i’ve also trained alternate and unusual nets (and contributed in small ways to the source code of the project) and feel a great deal of sympathy for A Silver whenever the mean girls turn out in force.

If CB and A Silver sell this engine and it turns out to be Leela they can get into legal trouble unless they take some steps from stopping that. E.g. revealing that it is leela in the first place, publicly releasing any difference in the code etc.

TCEC rules are actually very good so it doesn't get cluttered up with a hundred Leelas with some different net. The problem with A. Silver I have is that he just said it is HIS ENGINE which is a lie straight from hell. Only after the interview with Cato it became clear what was going on. People were legitimately upset with him. Maybe TCEC rules would be different if Silver wasn't such a secretive person.

Edit: There's a great example on how to behave: Ethereal.
There's a list of all things that are ideas from other engines even though it is in fact a unique engine. That is what open source should be like - complete transparency and not being ashamed that you implemented other peoples ideas while at least acknowledging their hard work and genius.

[chrisw]

You dreamed that bit of libelous BS up. TCEC cooked up some seriously odd rules, though, to prevent third party nets from entering with the lc0 net container.

What is "Lc0 net container"?

"DeusX" was Lc0 binary, Lc0 neural network architecture and implementation, trained with Lc0 scripts (so the only thing different is NN weights, it's not even "different neural net" as some people say).
I don't know what "Fat Fritz" is, but I expect it's the same.

Whether it's fine to commercialize is debatable, but what I don't like the most is that what "Fat Fritz" really is is hidden from people paying money for it.

After a great deal of arguing, centred on the promotional video TCEC-A Silver-ZeusX/ TCEC changed the description of ZeusX, supposedly with the agreement of LC0 Authors and A Silver, from "program written by A Silver" to:

“DeusX is a neural network, trained by Albert Silver from unique non-Leela data, exclusively from human played games. The engine that executes this network is Lc0, developed by the Leela Chess Zero team. The code for training the neural network, so that it can run in the Lc0 binary, also comes from the Leela Chess Zero project. Deserved credit to the Lc0 team will also be given in any citation of the engine in the following form, “DeusX Neural Network by Albert Silver, powered by Lc0″(c)”

Just now, according to the Chessbase article written by A Silver, DeusX was/is fact in September 2018 a program again:

A Silver wrote: Back in late September 2018, I had been working on a program called “Deus X”, which was designed to explore and challenge the idea that learning from the best humans and engines, other than itself, would lead to a worse result.

and

A Silver wrote:
“Deus X” (the Fat Fritz prototype)

Did I get it right back in 2018? That TCEC-Silver-Nelson video was done in full knowledge and support for subsequent ChessBase commercial use of "ZeusX", a set of weights produced allegedly by A Silver, but in reality produced inside LC0 development by D Kappe?

On 2-August-2018 Chris Whittington wrote:

My opinion? That video was a pre-planned opening shot for ChessBase to distribute Leela only with their own mega-hyped-super-duper-Albert-net (or weights as they are better known), calling it by a new name "ZeusX", "developed by ASilver" bla-di-bla and generally spreading as much BS as possible to convince the public to part with some money whilst simultaneously being fed BS hype articles as to "ownership, development, new NN, human style" and all the other BS we'll be reading. The video was to launch the BS, TCEC is designed to be the launch party, much extra BS will be produced about "on-going AI development, new ideas bla bla" and the Leela Authors will have been taken as the mugs who actually do all the creative work and get minimal credit. That's it.

[Guenther]

If CB and A Silver sell this engine and it turns out to be Leela they can get into legal trouble unless they take some steps from stopping that. E.g. revealing that it is leela in the first place, publicly releasing any difference in the code etc.

...

I don't think so, as long as it runs as a plain cloud service, which seems to be the plan.

[dkappe]

It’s interesting how little people understand about open source and the GPL in particular. If CB sells access (SaaS) to lc0+some net (there are a dozen or so lc0 offerings in their cloud), that’s fine by the GPL. If you want to close the SaaS “loophole,” you need to use the AGPL. Even then it only requires that you make the source available to users.

As far as crediting the Leela Chess project, it does so 4 times in TFA, including stating that it is the foundation of Fat Fritz. People need to read TFA before popping off.

[ChickenLogic]

I have read it carefully. It says that DeusX initially was based on Leela, that is true. But it also states:

"But that was not the main objective of the development — to make a program to compete against other chess engines. What we were aiming for was a new analytical tool that chess players could use to reevaluate their openings, and their general strategic ideas in the game."

and

"For a few days now we have silently launched the neural network program Anand is talking about on the ChessBase Engine Cloud. We decided to call it Fat Fritz, a tongue-in-cheek name that is easy to remember and talk about. Here's a summary description of the engine:

Fat Fritz is an advanced AlphaZero-type artificial intelligence chess engine that has been trained on billions of positions from human and computer-vs-computer games, self-play games, and endgame tablebases."

Whatever "new analytical tool" means, to me it still sounds like they are indeed trying to make this a product to be sold maybe not only in the cloud but also within the CB GUI. And if you have read carefully they do not say that Fat Fritz is Leela but they say it is their own engine which is a) either true and there's nothing wrong with it or b) (which seems more likely) a Leela clone that they call Fat Fritz and try to squeeze money out.

If option b is correct they should've said: Fat Fritz is using Lc0 as the engine with the incredibly strong net made by A. Silver.
If option a is correct then I take everything back

[Joost Buijs]

Have you ever written a chess engine? I assume not. Writing a chess engine from scratch takes several months, writing a strong engine takes several years. Training a network by supervised learning is something anyone with decent hardware can do in a couple of days when the software is already there. This is of a totally different order.

Yes, I have written a few chess engines, back in the dark ages before null move pruning. If you look around, you can find one of them bundled in a blindfold training tool, and the other one is leela_lite which, despite its name, has nothing to do with lc0. It uses T9 nets and a mcts/puct algorithm implemented in less than 100 lines of python to play chess.

Joost, clearly you’ve never trained a net. Otherwise you’d know that the leela SL tools are very limited (policy in particular is weak) and won’t ever get you to 3000. I’m also pretty sure you’ve never built a nn engine either.

I never claimed that I have written a NN engine, and I never will. My opinion is that leela_lite as you call it is just a wrapper, all the chess logic comes from py-chess which I assume is not yours, but maybe I'm mistaken.

[ChickenLogic]

Even then it only requires that you make the source available to users.

That is all I ever asked for.

[smatovic]

Ups, did the clone wars started again?

Some people should take some time to think about what the GPL philosophy stands
for, what was the intention of an programmer or an community if they put their
stuff under GPL, and - respect that.

--
Srdja

[dkappe]

I never claimed that I have written a NN engine, and I never will. My opinion is that leela_lite as you call it is just a wrapper, all the chess logic comes from py-chess which I assume is not yours, but maybe I'm mistaken.

Yes, you are mistaken. The money part of the engine is mcts/puct, which is certainly not provided by python-chess. Did I not want to go through the hassle of implementing bitboards, pgn parsing, etc., yet again? You betcha. Python-chess is great and I encourage everyone working with python and chess to use it. If you’re into needless pain, get in touch with Joost who will make sure your effort is pure enough.

[dkappe]

Ups, did the clone wars started again?

Some people should take some time to think about what the GPL philosophy stands
for, what was the intention of an programmer or an community if they put their
stuff under GPL, and - respect that.

--
Srdja

When you put your software under the GPL you are granting anyone who wants to use the software rights, including selling access to it as a SaaS offering. The GPL is not some magical commercial chastity belt.