connor_mcmonigle wrote: ↑Wed Jan 12, 2022 3:39 pm
You're testing books rather than engines.
1.-Just your opinion!
False, that is a fact. Now you are the one who has an opinion, that is basically untrue.
Sylwy wrote: ↑Wed Jan 12, 2022 4:09 pm
2.-Kaira1.1.bin is a small engine book; Perfect_2010.abk is a larger one accompanying a popular GUI:
3.-the book is only used at the root where it basically has nothing to do with the engine logic. Not in search;
Book size is not really a factor, it is unfair either way. In your case Kayra has two books, SF has one...
Sylwy wrote: ↑Wed Jan 12, 2022 4:09 pm
4.-SF hasn't polyglot book support. Why ?
Book support is a GUI concern not an engine one. A good GUI should be able to support different books for different engines. But still why not use the same polyglot book for both engines? Ah because you want to prove a point, sweet
If your GUI doesn't support polyglot, you are using the wrong tool, fix it
If we do the rematch of the match in a different condition , we can reach a clearer result.
Let Kayra 1.1 play with Perfect 10 book and let Stockfish Dev play with Kayra 1.1 book.
Arena Gui is the most downloaded chess program after Stockfish. It is strange that Arena still uses the Perfect 2010 opening book as the official book instead of Perfect 2021. And the standart chess engines came with Arena are 10- 20 year old chess programs. I don't think anyone uses these chess engines except those who love nostalgic chess programs.
mehmet123 wrote: ↑Thu Jan 13, 2022 7:29 am
Arena Gui is the most downloaded chess program after Stockfish. It is strange that Arena still uses the Perfect 2010 opening book as the official book instead of Perfect 2021. And the standart chess engines came with Arena are 10- 20 year old chess programs. I don't think anyone uses these chess engines except those who love nostalgic chess programs.
Yeah, Arena GUI is a great graphical interface with plenty of capabilities. Arena partner engines? Yes, a lot of nostalgia. Many have made the era. I have several hundred engines installed on the Arena 3.5.1 GUI! From all ages ...... and even from the multiverse.
The final result of the match that caused yesterday's hysteria of my "friends" ( & co):
+8 -0 =92 +28 Elo points
I just replayed the match on Hiarcs Chess Explorer Pro v1.0 using the same Hiarcs book for both engines. The result - so far - is still poor for SF.
Talkchess "scientists" have not yet answered point 3.-the book is only used at the root where it basically has nothing to do with the engine logic. Not in search.
I see the role of the book only for the variety of the opening, not something like a monster like Cerebellum. Just a few moves in the opening. The rest is engine logic. Otherwise it's just statistics.
I will come back with other matches. I'll see when!
Have a nice day !
mehmet123 wrote: ↑Thu Jan 13, 2022 7:29 am
Arena Gui is the most downloaded chess program after Stockfish. It is strange that Arena still uses the Perfect 2010 opening book as the official book instead of Perfect 2021. And the standart chess engines came with Arena are 10- 20 year old chess programs. I don't think anyone uses these chess engines except those who love nostalgic chess programs.
Yeah, Arena GUI is a great graphical interface with plenty of capabilities. Arena partner engines? Yes, a lot of nostalgia. Many have made the era. I have several hundred engines installed on the Arena 3.5.1 GUI! From all ages ...... and even from the multiverse.
The final result of the match that caused yesterday's hysteria of my "friends" ( & co):
+8 -0 =92 +28 Elo points
I just replayed the match on Hiarcs Chess Explorer Pro v1.0 using the same Hiarcs book for both engines. The result - so far - is still poor for SF.
Talkchess "scientists" have not yet answered point 3.-the book is only used at the root where it basically has nothing to do with the engine logic. Not in search.
I see the role of the book only for the variety of the opening, not something like a monster like Cerebellum. Just a few moves in the opening. The rest is engine logic. Otherwise it's just statistics.
I will come back with other matches. I'll see when!
Have a nice day !
If a book forces the engine to play inferior moves in the opening, of course the engine is going to perform worse. It's unclear why you find this surprising. Introducing another manipulated variable to any test makes drawing conclusions about the impact of any one of the manipulated variables impossible. This isn't my opinion, this is fact. If variety is all you're after, you should just use some large book of forced opening lines and insure that openings are repeated across game pairs to minimize noise as is well established best practice.
Needless to say, 100 games is still insufficient to draw any conclusions, but you're certainly making the situation even worse.
