Final Release of Ethereal, V12.75

[syzygy]

"Congrats to Stockfish for the incredible achievement. My middle finger to everyone else using it."

I think I tried hard to explain in many places why NNUE is not Minic official evaluation, how it is a copy/paste of code and technology, and how it is just an experiment for me do dig into the NN world. Maybe the good thing to do was to not try it at all. I feel very sorry for the trouble. "alea jacta est".

Stockfish has nothing to do with the achievement.
Nothing whatsoever.
Not only did they not invent it, they blindly attached the code, not even understanding that they were also pasting in a term for evaluation of drops from Shogi pieces in hand which has no place in standard chess.

Not.
Even.
That.

The shogi programmers are the ones who invented it and attached the code in a branch to the Stockfish Github they made and gifted to the computer chess community.

Wrong. They became part of the SF team.

[BrendanJNorman]

I don't understand this gloom about NNUE. It's just a new evaluation term. Search is still there. Even if all engines implement SF NNUE's the rankings will remain, you only have to add 300 to the current one.

It's not "a new evaluation term". A new evalution term doesn't discard the rest of the eval, it doesn't kill eval originality. It's a drop-in eval replacement.

The same as what NNUE is causing now has happened many times before. At some point it took great skill to create a multithreaded search. Now everybody does the "lazy" thing, as it turned that the most braindead way of doing SMP simply works best (for modern low-branching-factor engines and on current PC architectures, ignoring GPUs). The lazy SMP development really did take all the skill out of creating a multithreaded engine. One can be disappointed about that, but one could also just face reality and accept that what works best, works best.

Same happened with tuning by running thousands of games. Took all the skill out of picking values based on experience and intution.

Same happened with search extensions/reductions after Rybka got reverse engineered and of course now with Stockfish showing how it is should be done. Nobody is inventing the search themselves anymore.

(OK, Lc0 is an exception to all of this. And the NN idea has only just been introduced to computer chess, I"m sure there are discoveries waiting to be made and new unique engines to be created, even if only as a fork of SF.)

This is a good point too. Although Andrews's feeling is very understandable, it is also kind of like a caveman cursing the "new" use of fire for cooking.

If every engine is suddenly "dropping" Stockfish's eval in for "instant 200 Elo!!" - It will definitely cement my decision not to use the top engines except lc0, Stockfish, and sometimes even Ethereal for analysis.

[Albert Silver]

An hour after I make this post I'm filled with regret.

[1] Do people really want to see a bunch of NNUE engines, all running the same evaluation function, occupying the top slots of computer chess?
[2] Would people really find it interesting to have Ethereal use whatever "best" network is out there?
[3] Are people still interested when engines replace 50% of what made them unique?
[4] Do people want to continue to have Ethereal, acting as a Stockfish-lite, for new programmers?
[5] Is Chess doomed to go the route of Shogi and Go? We share the Networks, but have our own searches? Is that okay?
[6] Is this chess world okay with me plunking in a Network file and never changing it again?

If someone can convince me of all of those things, or at least most of them, then maybe there is still work to be done. The OCD in me likes the idea of refactoring all of the garbage in the NNUE libs. Likes to idea of coming up with a cleaner solution than what I did to hack the code in with such speed. I'de have to look at benchmarks. If NNUE is the "great equalizer", then its of little substance. But if NNUE is +160 elo flat across the board, then maybe its not the end of the world.

At the same time, NOT quitting goes against everything I've preached in this forum for the last few years. I don't even like people using Ethereal data to tune their engines. Alayan and I were going to release datasets to go with my tuning paper, but I did not like the idea of people converging to Ethereal. But now its okay that I drag and drop 2,000 lines from Stockfish? Its okay that I just run whatever Network Fishtest has brewing or Dkappe has brewing or someone else?

I'm not a net trainer. I won't do it, its not interesting. Ethereal's tuner has lots of things akin to net training. Lots of knobs to tweak. but it is a means to and ends, and that ends is putting human chess knowledge into the engine. The ends of Network training is to get a "better" black box.

I don't know if I'm ripping myself away from something I love for no reason. If I'm doing it because I see the pain in the future. If I'm doing it because I feel the pain now but am too daft to see the light at the end of the tunnel. I need someone to convince me that this hobby is still worth pursuing. That beating Leela and Allie and Scorpio is worth something. That doing so is not just a result of NNUE, but also a result of the ten thousand hours I've put into the project. That I'm not sacrificing my ideals and my principles by continuing on in this new era.

What a drama queen. I know this is going to make me sound singularly unsympathetic in this thread filled with messages of "please don't go!", but I really don't feel terribly sympathetic to all this grandstanding.

Look, stop trying to impose your values and your views of how to program a chess engine on everyone else. It is just lame. If you enjoy chess programming a certain way, no one, and I do mean no one, is stopping you. No one will criticize you, or judge you, or tell you you are the Amish of computer chess programmers. So stop telling everyone else how they must do it, and if they don't do it exactly the way you approve then you will leave.

If you derive pleasure from your hobby, and/or find it rewarding and cool, then stay. However, if you enjoy it but demand everyone do it your way or you won't play... Bah.

[Albert Silver]

"Congrats to Stockfish for the incredible achievement. My middle finger to everyone else using it."

I think I tried hard to explain in many places why NNUE is not Minic official evaluation, how it is a copy/paste of code and technology, and how it is just an experiment for me do dig into the NN world. Maybe the good thing to do was to not try it at all. I feel very sorry for the trouble. "alea jacta est".

Stockfish has nothing to do with the achievement.
Nothing whatsoever.
Not only did they not invent it, they blindly attached the code, not even understanding that they were also pasting in a term for evaluation of drops from Shogi pieces in hand which has no place in standard chess.

Not.
Even.
That.

The shogi programmers are the ones who invented it and attached the code in a branch to the Stockfish Github they made and gifted to the computer chess community.

Wrong. They became part of the SF team.

No, Tanuki (AKA Nodchip) did not become part of the SF team.

[syzygy]

first we lost Gull and Fizbo... after that, Andscacs... it seems Xiphos is not coming back either... and now Ethereal...

people get scared too easily from this neural monsters like leela and SF NNUE... This shows that more effort needs to be put in eval functions, and not only in perfecting search...

Don't you think it was SF rather than SF+NNUE that most got scared off? How can anyone compete with the whole SF team plus fishtest and its huge computational resources?

By changing their paradigm. Program for playability/style, not strength.

Have a competition based on who can "fool" a human playing in a blind test as follows:

Strongish (even my "low" 2200ish strength) player plays 3 blitz or rapid games, barely wins all 3 (maybe even loses one), and doesn't know which ONE game was against the engine.

I think this would be a worthwhile goal. And with strength ranges going from 2200 all the way up to GM, all passing this "Turing Test" of sorts.

I play against engines a lot, and I can say that no engine has managed this yet, although some are close.

I believe this is the type of thing Thorsten always laments about.

I agree there are various directions one could go in once one has given up on trying to write the #1 chess engine. What I meant is that the existence of SF+fishtest+resources is already good enough a reason to give up on trying to write the #1 engine.

But if one wants to create the #1 engine, just start with SF and find a way to do NNUE better. Then submit your modifications to the official SF tree or wait for someone else to do it. If enough people do this, there will probably be incompatible competing forks which cannot all be integrated into SF and variety is back.

[BrendanJNorman]

Thinker, the shining beacon in a dark world:
http://www.talkchess.com/forum3/viewtop ... =2&t=52325

Witch trial, as is usual in the computer chess world.

Key words: appears and convinced.

Thinker “The versions examined here are Thinker 5.4d Inert (March 2009), and Thinker 5.5.4, entered into Leo Dijksman’s WBEC tournament in January 2011. The first appears to derive from Strelka 2 (thus Rybka 1, and relating to Fruit 2.1). The second appears to derive from IvanHoe (thus partially Rybka 3).”
...
Based on the above-mentioned reports by Mark Watkins the ICGA is convinced that, at the very least, both Fritz Reul and Kerwin Medina have a case to answer. Depending on how Reul and/or Medina respond to these allegations the ICGA might decide to conduct further investigations and/or take some form of strong sanctioning action against the programmers.

Seems to me that Thinker was "guilty until proven innocent".

http://wbec-ridderkerk.nl/html/premier.htm
Outperforming stockfish.
Possibly could beat Rybka.
All in a program the size of a grain of sand.

I do have two wishes:
1. A cotton-picking, gravy-licking pv
2. For Lance and {someone else is working on it now whose name escapes me} to write a book on how to write a chess program.

Just to keep things in perspective, feel free to try to explain to me how one actually develops a chess engine that does _not_ display the PV or score? How does one test and debug such an engine?

food for thought, for those that actually have inquiring minds, rather than being mindless followers...

There is logical, and there is irrational, one chooses their camp.

Bob Hyatt:

He is active poster in Computer Chess Forums, served as moderator of CCC, and as member of the Secretariat of ICGA Investigations.

Just sayin', there are probably 50+ versions of Thinker, and because two are accused, a great programmer quit.

The norm in this community, it seems.

Anyway, back to Ethereal...

[syzygy]

"Congrats to Stockfish for the incredible achievement. My middle finger to everyone else using it."

I think I tried hard to explain in many places why NNUE is not Minic official evaluation, how it is a copy/paste of code and technology, and how it is just an experiment for me do dig into the NN world. Maybe the good thing to do was to not try it at all. I feel very sorry for the trouble. "alea jacta est".

Stockfish has nothing to do with the achievement.
Nothing whatsoever.
Not only did they not invent it, they blindly attached the code, not even understanding that they were also pasting in a term for evaluation of drops from Shogi pieces in hand which has no place in standard chess.

Not.
Even.
That.

The shogi programmers are the ones who invented it and attached the code in a branch to the Stockfish Github they made and gifted to the computer chess community.

Wrong. They became part of the SF team.

No, Tanuki (AKA Nodchip) did not become part of the SF team.

https://github.com/official-stockfish/S ... ORS#L1-L12

SF doesn't have a closed team of developers. Dozens of people have contributed to it, and the NNUE authors are among them.

[BrendanJNorman]

first we lost Gull and Fizbo... after that, Andscacs... it seems Xiphos is not coming back either... and now Ethereal...

people get scared too easily from this neural monsters like leela and SF NNUE... This shows that more effort needs to be put in eval functions, and not only in perfecting search...

Don't you think it was SF rather than SF+NNUE that most got scared off? How can anyone compete with the whole SF team plus fishtest and its huge computational resources?

By changing their paradigm. Program for playability/style, not strength.

Have a competition based on who can "fool" a human playing in a blind test as follows:

Strongish (even my "low" 2200ish strength) player plays 3 blitz or rapid games, barely wins all 3 (maybe even loses one), and doesn't know which ONE game was against the engine.

I think this would be a worthwhile goal. And with strength ranges going from 2200 all the way up to GM, all passing this "Turing Test" of sorts.

I play against engines a lot, and I can say that no engine has managed this yet, although some are close.

I believe this is the type of thing Thorsten always laments about.

I agree there are various directions one could go in once one has given up on trying to write the #1 chess engine. What I meant is that the existence of SF+fishtest+resources is already good enough a reason to give up on trying to write the #1 engine.

But if one wants to create the #1 engine, just start with SF and find a way to do NNUE better. Then submit your modifications to the official SF tree or wait for someone else to do it. If enough people do this, there will probably be incompatible competing forks which cannot all be integrated into SF and variety is back.

Agree on both counts.

[Albert Silver]

"Congrats to Stockfish for the incredible achievement. My middle finger to everyone else using it."

I think I tried hard to explain in many places why NNUE is not Minic official evaluation, how it is a copy/paste of code and technology, and how it is just an experiment for me do dig into the NN world. Maybe the good thing to do was to not try it at all. I feel very sorry for the trouble. "alea jacta est".

Stockfish has nothing to do with the achievement.
Nothing whatsoever.
Not only did they not invent it, they blindly attached the code, not even understanding that they were also pasting in a term for evaluation of drops from Shogi pieces in hand which has no place in standard chess.

Not.
Even.
That.

The shogi programmers are the ones who invented it and attached the code in a branch to the Stockfish Github they made and gifted to the computer chess community.

Wrong. They became part of the SF team.

No, Tanuki (AKA Nodchip) did not become part of the SF team.

https://github.com/official-stockfish/S ... ORS#L1-L12

SF doesn't have a closed team of developers. Dozens of people have contributed to it, and the NNUE authors are among them.

Perhaps we have a different understanding of what that means.

NNUE was created by the shogi programmers. They began use of it in 2018. The paper cites AlphaZero even. Then one of them ported it for use in computer chess and made a branch in Stockfish that supported it. They code had tons of problems at first (there are a number of quite funny comments in it still, including outright apologies...), but by no means did he 'become a part of the SF team' as a result.

Around July we saw it coming together, a strong net was made, and the results were clearly spectacular. The playing part of the code (not the learning/training) was merged to the official Stockfish. If that makes Nodchip 'part of the SF team', then so be it. But despite having announced he felt his work was done and he would be leaving having accomplished the goal of porting NNUE to SF, he has stayed due to issues and loose ends still present in the learning/training code, which is where he continues to supervise and code fixes.

[Tony P.]

The shogi programmers are the ones who invented it and attached the code in a branch to the Stockfish Github they made and gifted to the computer chess community.

Their decision to add that code to SF and neither add it to another chess engine nor publish code snippets under a less restrictive license can be regarded as a gift to the SF community and to those willing to derive from SF (and be at odds with the originality rules of competitions; how enforceable they are is a separate topic ), not the whole CC community. SF has earned that semi-exclusivity by being the strongest CPU engine at the time when shogi programmers decided to borrow search ideas from chess, and staying such when the time came for them to give back.