Stoofvlees II d2 is Killing Stockfish in TCEC

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

Moderators: hgm, Rebel, chrisw

User avatar
towforce
Posts: 11751
Joined: Thu Mar 09, 2006 12:57 am
Location: Birmingham UK

Re: Stoofvlees II d2 is Killing Stockfish in TCEC

Post by towforce »

Ras wrote: Wed May 01, 2024 9:42 pmWith a bigger board, the balance between short-range and long-range mobility (e.g. knights vs. bishops) will be different, and that difference will increase with board size. King mobility in the endgame likewise. It will still be a strategy game, but it won't be chess, and the knowledge required to succceed at that game is different from chess knowledge.

Which also makes it uninteresting for people to watch, because they would watch a game that they don't understand, and as I said, hardly anyone cares about chess variants, so there's no motivation for people to drop chess in favour of something else just for enjoying engine matches. These matches would be meaningless match result numbers.

Btw., the draw death was also talked about ten years ago. Then GPUs got powerful enough to run NNs, and then NNUE was invented. There are still positions that even NNUE engines don't quite understand, so there's still room for improvement. Improvement that will not happen if instead, we abandon chess.
Uri Blass wrote: Wed May 01, 2024 9:46 pmbigger board is a different game and is not chess.

It is also not clear which engine is better with bigger boards because engine A may be better with 12*8 board when engine B may be better with 20*8 board.

Ok - I take the point! :)

However, I still see great value in 1-ply engines: let's find out which engine knows the most about the game of chess! :twisted:
The simple reveals itself after the complex has been exhausted.
User avatar
towforce
Posts: 11751
Joined: Thu Mar 09, 2006 12:57 am
Location: Birmingham UK

Re: Stoofvlees II d2 is Killing Stockfish in TCEC

Post by towforce »

Uri Blass wrote: Wed May 01, 2024 9:52 pmI think chess knowledge is clearly relevant for bigger board.
The technique to win KRK is still the same and a lot of ideas are the same.

It is not chess but I disagree that people do not understand the game.

I expect top GM's to win in case there is a competition with big prize money with a bigger chess board.

Where it would be especially difficult is that on a bigger board with more pieces, positions would be a lot more complicated with more interactions between groups of pieces. Human players would have to move more quickly (or games would last too long), so with complicated positions and time pressure, I would expect even the best human players to make a lot of blunders.
The simple reveals itself after the complex has been exhausted.
chesskobra
Posts: 215
Joined: Thu Jul 21, 2022 12:30 am
Full name: Chesskobra

Re: Stoofvlees II d2 is Killing Stockfish in TCEC

Post by chesskobra »

towforce wrote: Wed May 01, 2024 4:39 pm
Let's take the symmetry of pairs of pieces as an example (there are many other types of symmetry), as this is what I think you're alluding to.

If the number of pieces is n, then the number of pairs of pieces is triangle(n - 1) (can also calculate it as c(n,2)).

So if there are 10 pieces, then there are 45 pairs of pieces, hence 45*8=360 symmetries just based on piece pairs alone.

Triangle number calculator - link.
Combinations calculator - link.
I didn't allude to anything. I want to know what you are alluding to. I only asked if you meant rotational or reflexive symmetries. What does the number c(n,2) have to do with symmetries, and further what does 8c(n,2) actually mean? (I can calculate it, but I want to know what it means for symmetries of or in a position.) The only symmetries I understand are the rotational and reflexive symmetries. For example, if KRk on a1,b1,h8, respectively, is a win for white, then so is KRk on a8,a7,h1, because the latter position is obtained by rotation through 90 degrees, so once you solve the first position, all the calculation for the first position can be applied a 90 deg clockwise rotation to obtain the calculation for the second position. Everybody knows this; it is not my discovery. And even these symmetries will be killed if pawns are involved. But I don' t see what the number 360 has to do with symmetries in 10 pieces. I would also like to see a precise definition of symmetries that you are using.

Anyway, I would like to see at least one example of a new pattern that you have found that leads to even minor simplification. Otherwise, it is generalities like "solve Riemann hypothesis by finding deep patterns - that is how mathematics is done".
jefk
Posts: 693
Joined: Sun Jul 25, 2010 10:07 pm
Location: the Netherlands
Full name: Jef Kaan

Re: Stoofvlees II d2 is Killing Stockfish in TCEC

Post by jefk »

Chesskobra wrote
"solve Riemann hypothesis by finding deep patterns"

ah, some fundamental math again huh ? :)

well besides finding a 100 pct sound logical proof for the conjecture
which every primary school kiddy can agree on, there
nowadays are other approaches to the Riemann stuff:
this guy claims he has a proof, unconventional in method but
still a proof, and he's offering 10k if you can refute his reasoning
https://www.0bq.com/RS


Also mr syzygy might enjoy it (and on beforehand already
state the proof is incorrect or something like that).

NB this guy also is working on a Riemann proof (see his FB timeline)
https://www.facebook.com/james.koukas

PS so with this more modern math approach (which i like actually)
maybe I should offer 10k for anyone who can prove chess is *not* a
draw, by showing us (a) line(s) for White which always win (no
matter which defense for black i would give as counterexample).
But that's another subject again (than Riemann) isn't it :)
chesskobra
Posts: 215
Joined: Thu Jul 21, 2022 12:30 am
Full name: Chesskobra

Re: Stoofvlees II d2 is Killing Stockfish in TCEC

Post by chesskobra »

You and the poster above me should start a thread to discuss profound ideas (mods can make it a sticky thread). The two of you should collaborate. I think you would find a common ground. His ideas on symmetries in deep patterns in chess and its generalisations might quickly prove that chess is a draw - because anything other than a draw would mean a violation of symmetry. Others would learn a thing or two from your discussions.
jefk
Posts: 693
Joined: Sun Jul 25, 2010 10:07 pm
Location: the Netherlands
Full name: Jef Kaan

Re: Stoofvlees II d2 is Killing Stockfish in TCEC

Post by jefk »

a bit off topic probably, but yep (earlier) thought of responding to Towforce, regarding some of
his innovative and original ideas regarding chess evaluation Not long ago a neural net alfazero
style without search already was able to reach GM level (but not 3400 or so like SF);
https://arxiv.org/abs/2402.04494

But no matter how you reason about it (eval) nobody can deny there also must be a search tree in chess,
more prevalent in chess problems, deep tactics, etc, but often also in normal positions.
Cleary indicated by minimax theorem by Von Neumann. see also the well known tree concept
and described eg. by Kotov in his (think like a GM) book eg.
https://www.chess.com/forum/view/genera ... hod?page=2

Symmetry ? well yep, but in addition, and this is more important, it's about the 'degrees of freedom'
(for Black). No matter what White plays from the (historical) 'normal' positions , there always remain
several options for Black to escape situations leading to worse positions. This is different then a situation
as eg. in Four in a row, where gradually the options for the second player become more and more restricted,
and thus the first player indeed can force a win.

Such a high degree of freedom (for Black) is apparent because eg. contrary to checkers, pieces in
chess can move back, there is the 3 pos repetition draw rule etc. And the 50 move (no pawn move)
draw rule, this all leads to a relatively large draw margin in chess. And to an obvious conclusion.
You suggested last year that i should write down my reasoning but i although I made notes during
our discussion(s) i didn't start (yet?) with a coherent and concrete article(*). Maybe because prove the 'draw
conjecture'' rigorously without a -narrowed- 'brute force' calculation (which unless Riemann should be
possible for chess) indeed is not a simple task; contrary to what towforce seems to think imo it will involve
lots of practical examples, concrete reasoning, etc. (and experience helps to get some hunches
about how this works in practical chess, and why White cannot find a forced winning line;
my conviction is more a result of (large and deep) experience than reasoning about symmetry, and
without a deep understanding about some math concepts as 'winning strategies' discussions about
the topic (and skeptical opinions) remain vague and inconcrete. A more suitable game to
explore deductive methods for proving a draw is maybe conventional draughts; it hasn't
been numbercrunched (yet) like checkers, yet most experts now are convinced it's a draw.

So it's complicated indeed, to illustrate with an example: in the position after 1.g4 i cannot
exclude that there's a forced win for Black. So in similar style, when i started my computer
chess opening research more than twenty years ago, i would'nt exclude a possibility that after
e.g 1.e4, d4, or 1.Nf3 (or possibly 1.c4) White might be able to gradually increase it's
positional advantage. And maybe for one or two positions of the chess960 this still could
be possible (unlikelly i admit, but you never know, and i didn't check it). But for the
normal chess position i made it topic of research starting even before the year 2000,
and there's *no* forced win for White as i found in practice(**), already around 2008)
https://superchess.blogspot.com/2008/01 ... rfect.html
https://superchess.blogspot.com/2008/03 ... y-etc.html

(*) but like i wrote, i offer 10 k ($) for the one who finds a forced winning line for White
(**) whereby i went much further than simple stuff as eg. the basic Hert tree (from Pohl)
something you suggested some time ago; this Hert tree is way too small, and thus
has deficicies btw,for example in the Scotch after e4 e5 Nf3 Nc6 d4 exd4
Nxd4 Nf6 Nc3 the move Bb4! (instead of ...d6?!) is the best (drawing) move
but it's not in the Hert500.pgn...

PS as for the Riemann examples i posted, i'm sure that from academic circles there
will be criticism about these 'proofs' or their methods (in particular the FB posting).
The youtube 'proof' (more some sort of inductive reasoning with lateral thinking and thus
probably not a rigorous proof according to some academis) is by a (serious and probably
highly talented) mathematician (and FB contact of me) working in Los Alamos btw.
In other words not a crank (like i know that for the Fermat last theorem there are
many crackpot proofs, also on FB). But despite such academic criticism i think that
such reasoning(s) gradually are making it more likely that the Riemann conjecture
is true, in other words a gradual process like it often is in theoretical physics research.
And although math isn't physics, also in the development of proofs, you often see some
gradual process; the (now accepted) proof by Andrew Wiles for Fermat's last theorem was
a long arduous process, whereby he ofcourse built on earlier works; in other words,
a gradual process, and not a sudden -once in a lifetime- (and required) event by
eg. some lonely genius the kind of Gregori Perelman type or so.
chesskobra
Posts: 215
Joined: Thu Jul 21, 2022 12:30 am
Full name: Chesskobra

Re: Stoofvlees II d2 is Killing Stockfish in TCEC

Post by chesskobra »

I am glad that, thanks to Zuckerberg and others, nowadays there are many venues where independent researchers can publish their unconventional proofs or disseminate knowledge. They don't have to plead to the gatekeepers at Harvard or Princeton or Cambridge for acceptance. The playing field has become even. We don't need Royal Society or similar exclusive clubs.
User avatar
towforce
Posts: 11751
Joined: Thu Mar 09, 2006 12:57 am
Location: Birmingham UK

Re: Stoofvlees II d2 is Killing Stockfish in TCEC

Post by towforce »

chesskobra wrote: Wed May 01, 2024 10:13 pm...generalities like "solve Riemann hypothesis by finding deep patterns - that is how mathematics is done".

Prime Numbers do seem to have some sort of pattern - here's a "Ulam Spiral":

Image

The Riemann Hypothesis is related to the distribution of prime numbers, but I don't understand it well enough to say whether finding an underlying pattern in Prime Numbers would enable a proof of it.
The simple reveals itself after the complex has been exhausted.