I do NOT believe that Chess is getting close of being solved at all.
Is LCZero better than Stockfish positionally ?
Moderators: hgm, Rebel, chrisw
-
Chessqueen
- Posts: 5685
- Joined: Wed Sep 05, 2018 2:16 am
- Location: Moving
- Full name: Jorge Picado
Is LCZero better than Stockfish positionally ?
Russian believe that they are special, that they can Kill thousands of Ukranian civilians, but cry like babies when a few Russian...
-
jefk
- Posts: 694
- Joined: Sun Jul 25, 2010 10:07 pm
- Location: the Netherlands
- Full name: Jef Kaan
Re: Is LCZero better than Stockfish positionally ?
sigh; he talks about TCEC games which are restricted regarding the openings (for both sides,
in other words the openings are 'cooked' ie. a strictly defined opening sequence, despite
the almost hysterical comments by mr Gotham already during the opening stage), the vid
is 6 months old btw (everything ok with you mr CQ, hope you're doing well (***);
Here's a more recent vid
similar topic, and here SF seems better again (differences imo are small, it's
comparing apples and oranges and depends on position, search dept etc.
PS what you're talking about is a strong solution for chess, which i admit will
not be found any time soon.
https://academickids.com/encyclopedia/i ... e_vignette
The 'weak' solution (unrestricted openings and perfect/adequate play) is known by now
(although many suggest it's not proven (yet (*) ), it's a Draw as we see eg. in top ICCF games.
And it's also apparent now with higher rated engine games on playchess (**) (where i'm currently
testing -and further developing an unconventional/gambit book; less advantage for White
in the opening (which usually is a minuscule 'advantage'anyway) but sharper play and a bit less
draws, although again, final outcome are draws unless opponent makes (opening) mistake;
a different exercise than playing TCEC games, because my opponent(s) still are free
to choose whatever defense/opening, eg. avoiding the gambit or whatever.
(*) i've finally started with a (draft) paper on this topic, with the intention to submit it
to a journal before end August.
(**) some engines now simply play boring opening lines with e3 d3 and then later d3d4 or so,
while i first was fooled with my (rather small) book with suboptimal lines , i now easily
achieve draws, even with a slower computer, especially if the engine can connect
to the chinese database (if it's online/available, which seems only 1/3 of the time or so)
(***) if we're going on like this it becomes more a topic for the 'kindergarten' subforum
in other words the openings are 'cooked' ie. a strictly defined opening sequence, despite
the almost hysterical comments by mr Gotham already during the opening stage), the vid
is 6 months old btw (everything ok with you mr CQ, hope you're doing well (***);
Here's a more recent vid
similar topic, and here SF seems better again (differences imo are small, it's
comparing apples and oranges and depends on position, search dept etc.
PS what you're talking about is a strong solution for chess, which i admit will
not be found any time soon.
https://academickids.com/encyclopedia/i ... e_vignette
The 'weak' solution (unrestricted openings and perfect/adequate play) is known by now
(although many suggest it's not proven (yet (*) ), it's a Draw as we see eg. in top ICCF games.
And it's also apparent now with higher rated engine games on playchess (**) (where i'm currently
testing -and further developing an unconventional/gambit book; less advantage for White
in the opening (which usually is a minuscule 'advantage'anyway) but sharper play and a bit less
draws, although again, final outcome are draws unless opponent makes (opening) mistake;
a different exercise than playing TCEC games, because my opponent(s) still are free
to choose whatever defense/opening, eg. avoiding the gambit or whatever.
(*) i've finally started with a (draft) paper on this topic, with the intention to submit it
to a journal before end August.
(**) some engines now simply play boring opening lines with e3 d3 and then later d3d4 or so,
while i first was fooled with my (rather small) book with suboptimal lines , i now easily
achieve draws, even with a slower computer, especially if the engine can connect
to the chinese database (if it's online/available, which seems only 1/3 of the time or so)
(***) if we're going on like this it becomes more a topic for the 'kindergarten' subforum