TalkChess.com

I think it is okay to have a NNUE to have a huge elo gain, technology is evolving everyday, it is not like yesterday how like HCE engine used to be, things different now. If the author's want their engine to be NNUE then let it be, like every NNUE engine with their own Implementation and idea to make their engine strong and to had a different style of playing. Making own net is fine because it preserves the original engine's gameplay with HCE of its own engine. But using a other's much stronger net is not fine. Copying code with authorization from its owner is fine, but Copying code without authorization is not fine. So, let's just let the author of every engine wants they want to do with their engine not by judging them. We should respect each other as we respect of each chess engine. This is just my opinion

When null move pruning came out, it was null move era.

When LMR came out, it was LMR era.

The same thing for every other innovation.

Funny, nobody cried wolf when every one else added those things to their programs, though.

Dann Corbit wrote: ↑Mon Jan 18, 2021 1:25 pm When null move pruning came out, it was null move era.
When LMR came out, it was LMR era.
The same thing for every other innovation.
Funny, nobody cried wolf when every one else added those things to their programs, though.

You see the difference though I hope? NMP is an _addition_ to your search. It was a few lines of code into your existing functionality. NNUE is _replacing_ your entire eval, which is typically the most unique thing in a given engine?

The NMP / LMR arguments don't really hold water. Better off arguing that all evaluations are nearly the same anyway, so why not make them exactly the same? Plus, this is not everyone reimplementing NMP to fit their engine. Its everyone copy pasting NNUE _into_ their engine. People were very angry about taking LMR tables back in the day -- Robert Hyatt had entire forum posts refuting allegations from Rebel author -- and now we take tables of 10million values?

There is absolutely nothing wrong with taking tables of a billion data points.
1. Math belongs to everyone
2. There is no originality whatsoever in the tables (work is not originality)\

Dann Corbit wrote: ↑Mon Jan 18, 2021 2:00 pm There is absolutely nothing wrong with taking tables of a billion data points.
1. Math belongs to everyone
2. There is no originality whatsoever in the tables (work is not originality)\

My entire program can be boiled down to 0s and 1s, and one could arrange that into a table.
Should you be able to take my entire table? There are arguments for NNUE, but that is also not one of them.

I suggest you examine copyright law.
I am right
You are wrong'
Simple as that.

Your argument has been tried and failed.
A telephone book cannot be copyrighted, despite the enormous work needed to create it.
Find out why.

Dann Corbit wrote: ↑Mon Jan 18, 2021 2:03 pm I am right; You are wrong'; Simple as that.

Typically a strong starting point for debate.
Unfortunately one cannot add a "moderator or administrator" to their ignore list.

I think you have the willpower to ignore me even though I frighten you with facts

I agree with Andrew. Evaluation is the soul of the engine. LMR, Null move, etc is nothing without eval.

By my saying, "I want the right answer" I was not implying that all programs should be NNUE or NN based.

I use Sting, which has no NNUE nr NN whatsoever. It just recently solved a problem that none of the NNUE based programs could solve.
That is genuinely useful.

I guess it is like baseball bats. I do not necessarily admire the grain of the wood or the sheen of a properly boned bat.
In fact, even if it makes a disgusting 'TINK' when it hits, I do not despise the metal bat.
Even a corked bat is OK to hit the ball out of the park, so long as it is just batting practice.

I need the bat to be useful, not beautiful