Chess will never be solved, here's why

Martin and others are kind of 'confirmed' when it comes to AI being 'not good' for some things.
Just now I was able to get a clear description from it contrasting glymphatic system from lymphatic system and where the exit points are. With two different meanings of 'node' there.
But when I went to the subject of 'nodes' in chess computing -
(might get 'reaction' on this like 'you see ??') Lol.
it suggested that the y-axis of the graph is the depth (number of plies) and the x-axis is the breadth of numbers of moves available.
And the node expresses the 'game state' (position of pieces along with things like whose move and en passant status - 50 move rule and so on)
I remarked that since the depth of moves played would determine the number of moves available then the traditional roles of x-axis and y-axis would be reversed - with x dependent on y.
Did I 'trick it' into agreeing on that?
I then tried another 'trick'. I suggested that the system would work better with a 3d 'graph'.
Maybe I tricked it there too. Into agreeing that adding a z-axis would work better.
Its clear that the discussion is moving on into a discussion of nodes.
But using AI to define the terms - maybe I'll get 'I told you so!' ...
--------------------------
I was actually looking into the timeline of Miescher (discovered nuclein in 1869 apparently) and the coining of the term 'nucleic acid' to replace the term 'nuclein'. Seemed to turn out that the acid action (not flowery terms there) of phosphate groups in nuclein is essential to life.
So then the conversation moved from there into buildup of monophosphates versus triphosphates in the nervous system and what that causes and then the glymphatic system.
-------------------------------------
Anyway - 'node' representing the 'game state' (which is more than than 'diagram of position') as a point on a graph with depth in plies as vertical y-axis and breadth of number of moves available as horizontal x-axis.
Doesn't sound right though.

Yes, that had crossed my mind. I'll rerun it and look at the pv to see if the explanation is viable.

That explains why Stockfish has trouble evaluating common opening lines and specifically openings that lead to alot of imbalance of chaotic games, such as the kings gambit/Danish gambit/Traxler. If a deeper opening book could somehow be incorporated or merged into the typical searching and evaluation code, would that help improve that, or is any kind of computer code functional merging impossible given the down votes on my last post?

@MARattigan
Descartes pioneered graphs in the 1600s.
(Not mentioned for debate purposes. Simply to describe the timeline.)
Apparently Euler introduced 'nodes and edges' in the 1700s but did not use those terms to describe same.
'Nodes and edges' terminology supposedly first appeared between World War I and II but that terminology grew as computer science grew - after World War II.

chatgpt timed out on me. So I had to use copilot.
So now we have another term 'game tree graph'.
And 'Euler's type of graph'.

I think the opening book is probably in NNUE's head and if you want a deeper opening book you'ld want to ask SF/NNUE, so it's already incorporating. I could be wrong.

@Elroch could maybe confirm.

We're generally only talking about Euler's type of graph., so you don't need to use the term. If it's got axes then you know it's the other type.

And any tree is a graph (unless it's @Optimissed talking about his mother in law's apple tree, which you can ignore) so you don't need 'game tree graph', just 'game tree' will do.

I was more curious about the concept of "merging" different computer programs. A while back that crossed my mind regarding video games. Like if 2 totally different graphic intensive games were somehow interlinked where it would be a wild, random, chaotic mix between the two running in a single source code, but in such a way that it wouldn't crash and the codes combining wouldn't be incompatable. Features from both games would appear simultaneously in a new, merged game in a fun unpredictable way combining the individual features (physics, behaviors, objects..etc). I actually asked chatgpt about this once and it said that it might just be possible to some small extent. Even between 2 totally different programs like 2 video games. But here we're talking both chess programs. The different engine methodologies could run in parallel, not even like a fusion of 2 different source codes.

But presumably one with pages or a different engine.

But didn't your version of SF play e5 after e4 in your example. How would it avoid the king's gambit as Black?

And then you could have a third one to choose which to believe and maybe a fourth one to do the same and a fifth one too choose which of the second two to believe. Sounds like quite a good idea.

chess might one be solved not by an engine but by math because theoreticcaly Chess is math so by being the number 1 algebra person in math you might be able to solve every move by making algorithms and other stuff to figure out the actual best move every game because even stockfish isn't perfect

No what I mean is interactive circuit breakers so that if within a random typical pruning search, it stumbles upon a position that matches the starting position of whatever obscure puzzle, then it can pull the known full analysis of lines from that database, and know what obscure moves to play out, and thus terminate other existing searches of "more relevant" moves. As if the position was an already solved tablebase, but just that 1 position, or those 100 unique positions, or however many we manage to program in manually. The database would be hooked up with the search algorithm and step in to stop it, if one of those database positions is matched exactly. I don't know how this would be coded though.

Partially incorrect. Chess is only a mathematical game in terms of individual piece movement geometry and geometric constraints of the board itself. For example, why the number of moves a king has located in a centralized location can always be represented as (3^n - 1), with n being the dimensionality of the board (2 for standard chess), and similar type of expressions for rooks (7n) and other pieces. But there are no equations to explain why some 10-20 move obscure solution to a composed puzzle "works". Other games like "lights on" can actually be mathematically solved with linear algebra. Even things related to the chessboard itself, such as the 8 queens problem/knights tour problem/piece domination problems can be somewhat represented mathematically (mostly combinatorics) The pieces and board and some interactions between the 2 can be represented mathematically, but not the game of chess itself. The game uses mathematically constructs but doesn't have mathematical solution.

If you can understand this video, you will start to understand why "math" as you are looking at it isn't going to work out. Even if you posited some completely new algorithms that are some giant leap forward, you can't test them without solving chess first. You have to therefore create tiny incremental advances from already established parts of chess already "solved" by "math". You can prove easy stuff, like tablebases do,, by brute force, but when was the last new absolute "proof" of anything new in chess? Not in the past 50 years...we've solved pretty much all we are going to this way,.

Even if you somehow got this going, you still face the problem of needing more and more computing power as you keep adding new layers, because you still have to prove your new advances work once those advances outstrip the capabilities of GMs to handle. Note that just like any other science, if you can't prove and observe your results, you can't really go forward to the next step, because then your efforts are built on X number of assumptions, where X keeps increasing until it becomes impossible.

This is also why the Stockfish discussion going on right now might be fun, but it doesn't do anything towards solving chess.

This is an example of someone inexcusably assuming everyone else is as ignorant as himself. If you had read much of the discussion.you could perhaps have learnt that many two player zero sum games have been solved. Better would have been to have read anything on the subject - say a couple of wiku articles.

Thanks. This was of course correct - it was a very silly and inappropriate comment by Dub.

https://www.chess.com/forum/view/general/knights-tour-1?page=1#last_comment

Here's a good old thread that I just updated, which shows one aspect of "chess" that's purely mathematical. But only strictly due to geometry of individual piece movement. Nothing to do with the game strategy itself.