Qualitative Engine Analysis Shootout: Theoria, PlentyChess, Stockfish, Obsidian

Discussion of anything and everything relating to chess playing software and machines.

Moderator: Ras

syzygy
Posts: 5897
Joined: Tue Feb 28, 2012 11:56 pm

Re: Qualitative Engine Analysis Shootout: Theoria, PlentyChess, Stockfish, Obsidian

Post by syzygy »

FireDragon761138 wrote: Tue Jan 27, 2026 11:59 am I use AI assistance to formulate some of my ideas into a coherent presentation, but the ideas themselves are mine, and the research is mine. It's not AI slop, and it's not philosophical nonsense.
It really is AI slop. It is full of nonsensical claims that have no basis in the single line that was trivially changed.

The only idea you seem to have is to use Lc0 training data to train SF's NNUE net.
SF's default NNUE net is already trained using Lc0 training data.
So your idea is not new.

That's it. No AI needed. And I didn't even need to suggest your approach is "deeply flawed".
FireDragon761138
Posts: 80
Joined: Sun Dec 28, 2025 7:25 am
Full name: Aaron Munn

Re: Qualitative Engine Analysis Shootout: Theoria, PlentyChess, Stockfish, Obsidian

Post by FireDragon761138 »

syzygy wrote: Wed Jan 28, 2026 4:47 am
FireDragon761138 wrote: Tue Jan 27, 2026 11:59 am I use AI assistance to formulate some of my ideas into a coherent presentation, but the ideas themselves are mine, and the research is mine. It's not AI slop, and it's not philosophical nonsense.
It really is AI slop. It is full of nonsensical claims that have no basis in the single line that was trivially changed.

The only idea you seem to have is to use Lc0 training data to train SF's NNUE net.
SF's default NNUE net is already trained using Lc0 training data.
So your idea is not new.

That's it. No AI needed. And I didn't even need to suggest your approach is "deeply flawed".
I checked the GPL. Calling it "my engine" doesn't violate the GPL, and is therefore my right. The GPL covers copying and distribution only. I put actual work into this project, and your attempts to minimize this do not go unnoticed.
syzygy
Posts: 5897
Joined: Tue Feb 28, 2012 11:56 pm

Re: Qualitative Engine Analysis Shootout: Theoria, PlentyChess, Stockfish, Obsidian

Post by syzygy »

FireDragon761138 wrote: Wed Jan 28, 2026 4:51 am
syzygy wrote: Wed Jan 28, 2026 4:47 am
FireDragon761138 wrote: Tue Jan 27, 2026 11:59 am I use AI assistance to formulate some of my ideas into a coherent presentation, but the ideas themselves are mine, and the research is mine. It's not AI slop, and it's not philosophical nonsense.
It really is AI slop. It is full of nonsensical claims that have no basis in the single line that was trivially changed.

The only idea you seem to have is to use Lc0 training data to train SF's NNUE net.
SF's default NNUE net is already trained using Lc0 training data.
So your idea is not new.

That's it. No AI needed. And I didn't even need to suggest your approach is "deeply flawed".
I checked the GPL. Calling it "my engine" doesn't violate the GPL, and is therefore my right. The GPL covers copying and distribution only. I put actual work into this project, and your attempts to minimize this do not go unnoticed.
It violates all ethical codes. Especially when you pretend to have developed wholly new approaches that avoid the "deeply flawed" techniques used in SF, and in reality you have just slapped a new name on an old engine and contributed nothing. You are lying to us by calling it "your" engine. It is not very nice to come to a forum and lie.

And yes I certainly will call out your work for what it is: empty.
FireDragon761138
Posts: 80
Joined: Sun Dec 28, 2025 7:25 am
Full name: Aaron Munn

Re: Qualitative Engine Analysis Shootout: Theoria, PlentyChess, Stockfish, Obsidian

Post by FireDragon761138 »

syzygy wrote: Wed Jan 28, 2026 4:59 am
FireDragon761138 wrote: Wed Jan 28, 2026 4:51 am
It violates all ethical codes.
No it doesn't... and don't speak to me of ethics if you want to be taken seriously, as you probably have zero grounding for any kind of coherent ethical system in the first place. The GPL doesn't create a community obligation in terms of ongoing relationship, that's legal nonsense. I don't owe you or your community fealty or devotion. Deeper principles such as natural law and the rights of a man to his own labor come into play and override any kind of legal penumbra you think the GPL could cast. And in today's political-legal climate, dominated by many Originalists in the US, you won't stand a chance with that argument in court.
Especially when you pretend to have developed wholly new approaches that avoid the "deeply flawed" techniques used in SF, and in reality you have just slapped a new name on an old engine and contributed nothing.
I said fundamentally flawed, and i meant that in a technical sense. And it is true, Stockfish isn't some kind of oracle, and numerous examples of how it misjudges positions are discussed on this forum.
sscg13
Posts: 24
Joined: Mon Apr 08, 2024 8:57 am
Full name: Chris Bao

Re: Qualitative Engine Analysis Shootout: Theoria, PlentyChess, Stockfish, Obsidian

Post by sscg13 »

Certain actions could be legal, but being legal doesn't necessarily mean that everyone will think highly of said actions.
Many people do not think you give enough credit to the countless developer hours and 19000+ CPU years that went into the Stockfish search that you are able to freely use, and hence find stripping "(the Stockfish developers)" from the UCI author accreditation to be distasteful.

I would also like to remind you that AI, by design, is biased towards agreeing with and glazing whoever it talks to.
misjudges positions
From what I can make of your philosophy, a "true oracle" that would play and evaluate chess perfectly would be even more opaque. Most engine developers focus on getting closer to "perfect chess", it is a natural consequence that higher level chess is harder to interpret.
Last edited by sscg13 on Wed Jan 28, 2026 12:06 pm, edited 1 time in total.
Ciekce
Posts: 206
Joined: Sun Oct 30, 2022 5:26 pm
Full name: Conor Anstey

Re: Qualitative Engine Analysis Shootout: Theoria, PlentyChess, Stockfish, Obsidian

Post by Ciekce »

FireDragon761138 wrote: Wed Jan 28, 2026 5:27 am The GPL doesn't create a community obligation in terms of ongoing relationship, that's legal nonsense. I don't owe you or your community fealty or devotion. Deeper principles such as natural law and the rights of a man to his own labor come into play and override any kind of legal penumbra you think the GPL could cast. And in today's political-legal climate, dominated by many Originalists in the US, you won't stand a chance with that argument in court.
you're very obsessed with defending the legality of your efforts

people aren't saying what you're doing is illegal, they're accurately describing it as cringe
FireDragon761138 wrote: Wed Jan 28, 2026 5:27 am I said fundamentally flawed, and i meant that in a technical sense.
you might want to read up on the dunning-kruger effect
syzygy
Posts: 5897
Joined: Tue Feb 28, 2012 11:56 pm

Re: Qualitative Engine Analysis Shootout: Theoria, PlentyChess, Stockfish, Obsidian

Post by syzygy »

FireDragon761138 wrote: Wed Jan 28, 2026 5:27 am
syzygy wrote: Wed Jan 28, 2026 4:59 am
FireDragon761138 wrote: Wed Jan 28, 2026 4:51 am
It violates all ethical codes.
No it doesn't... and don't speak to me of ethics if you want to be taken seriously, as you probably have zero grounding for any kind of coherent ethical system in the first place. The GPL doesn't create a community obligation in terms of ongoing relationship, that's legal nonsense. I don't owe you or your community fealty or devotion. Deeper principles such as natural law and the rights of a man to his own labor come into play and override any kind of legal penumbra you think the GPL could cast. And in today's political-legal climate, dominated by many Originalists in the US, you won't stand a chance with that argument in court.
Your LLM must have been smoking something if this is what it produced.
Ethical codes are indeed not legal obligations. It is not illegal to come here and lie and deceive. But it is very much frowned upon.
Especially when you pretend to have developed wholly new approaches that avoid the "deeply flawed" techniques used in SF, and in reality you have just slapped a new name on an old engine and contributed nothing.
I said fundamentally flawed, and i meant that in a technical sense. And it is true, Stockfish isn't some kind of oracle, and numerous examples of how it misjudges positions are discussed on this forum.
You simply slapped a new name on an old engine. Do you not see the problem?
And all your claims about flaws are totally wishy washy and those alleged flaws are obviously just as present (or not existing in the first place) in SF-1.16 aka "Theoria 0.1".
Are you suggesting that "Theoria 0.1" does not misjudge any position? Oh man...
FireDragon761138
Posts: 80
Joined: Sun Dec 28, 2025 7:25 am
Full name: Aaron Munn

Re: Qualitative Engine Analysis Shootout: Theoria, PlentyChess, Stockfish, Obsidian

Post by FireDragon761138 »

sscg13 wrote: Wed Jan 28, 2026 11:56 am Certain actions could be legal, but being legal doesn't necessarily mean that everyone will think highly of said actions.
Many people do not think you give enough credit to the countless developer hours and 19000+ CPU years that went into the Stockfish search that you are able to freely use, and hence find stripping "(the Stockfish developers)" from the UCI author accreditation to be distasteful.
That's OK, I found Stockfish developers on Discord, implying Robert Hyatt was lazy, to be distasteful, just because he's never "updated Crafty with an NNUE". But I'm not going to accuse Stockfish of "theft" or being "unethical", just because they refuse to recognize Glaurung/Stockfish depends on advancements in computer chess made in the 1980's and 90's (which many of them discredited as based on misguided principles or theory).

Taste has nothing to do with morality, it's merely preference. Actual community involve relational responsibility and recognition, not merely assertions of normativity or legality. And I have received none of that from the Stockfish community. So I owe them nothing. I'm not obligated to show fealty to my abusers.
syzygy
Posts: 5897
Joined: Tue Feb 28, 2012 11:56 pm

Re: Qualitative Engine Analysis Shootout: Theoria, PlentyChess, Stockfish, Obsidian

Post by syzygy »

FireDragon761138 wrote: Wed Jan 28, 2026 9:38 pmTaste has nothing to do with morality, it's merely preference. Actual community involve relational responsibility and recognition, not merely assertions of normativity or legality. And I have received none of that from the Stockfish community. So I owe them nothing. I'm not obligated to show fealty to my abusers.
Asking yourself WHY you are not receiving your desired recognition seems to be beyond you.

I still think your persona is merely an elaborate joke.
FireDragon761138
Posts: 80
Joined: Sun Dec 28, 2025 7:25 am
Full name: Aaron Munn

Re: Qualitative Engine Analysis Shootout: Theoria, PlentyChess, Stockfish, Obsidian

Post by FireDragon761138 »

syzygy wrote: Wed Jan 28, 2026 10:05 pm
FireDragon761138 wrote: Wed Jan 28, 2026 9:38 pmTaste has nothing to do with morality, it's merely preference. Actual community involve relational responsibility and recognition, not merely assertions of normativity or legality. And I have received none of that from the Stockfish community. So I owe them nothing. I'm not obligated to show fealty to my abusers.
Asking yourself WHY you are not receiving your desired recognition seems to be beyond you.

I still think your persona is merely an elaborate joke.
No, I am not a joke. I'm a person.

You may not understand this, but I happen to think people deserve the benefit of the doubt as a right, until proven otherwise. Treating people outside the community or the consensus as a pariah to be the butt of jokes is a violation of a person's dignity and is fundamentally immoral.