I think many people in the past (100...150 years) applied all those HPD, NPD, BPD, OCD concepts to dreamers who were thinking about flights in atmosphere and space 
Chess will never be solved, here's why
Sort:
Mar 14, 2025
0
#15723
Elroch wrote:
Pedantical, the AI copy and paste is incorrect. To solve chess you need to know the initial state and the rules about moving (each of which maps a state to a state) and terminal states. Nothing else.
For example, when you are presented with a chess problem, you are being given an initial game state and asked to solve it, and you can, given knowledge of the rules. The only difference is that for solving chess you are given the initial position, and the analysis needed is a bit (ahem) larger.
Also, the reference to the game tree - it says
"...the 1012010^{120}10120 game tree. Solving chess means navigating that tree,"
is at best misleading (at worst simply wrong). You do NOT need to navigate the game tree, just a tiny fraction of it (very rough estimate 10^30 nodes). Again, this is like for solving a chess problem: if you guess the first move correctly, you only need look at lines that start with that first move - you have successfully ignored all the branches of the game tree that start with other moves. And this can happen at any stage in the solution, not just the first move.
And again, the same is true of solving chess, in principle (it's just impractically large).
For computers to solve chess - they don't need to make moves or play a game.
They don't need 'game tree' either.
You can link positions by addiing to the material and varying where the additional material is placed (not moved). Would work much better with won positions than with drawn positions.
And programmers will use computer jargon for chess-solving projects.
Do they have to use game theory jargon too to do that?
No but they'll do that too.
But to do the job right and get into alternative solving routes - they'll have to do better than asserting jargon.
--------------------
Long arguments defending game theory jargon ...
Instead - there's actual discussion to be had as alternatives to the hopeless 'wide game tree search' whether from the back (runs into John Tromp's number which is even more of an Avocado than the Avogadro number) or from the front (runs into the Shannon number of 10^120 which reminds me of the Hayflick limit in human life span. 120 years.
Almost nobody is getting anywhere near that 120 year life span.
But the Shannon number is even more hopeless. A lot more.
-----------------
Give it some scale. They couldn't brute-solve checkers either.
Too many positions.
So they alt-solved checkers instead.
You know if they can't brute-solve checkers - they ain't solvin' chess with 'wide game tree search' either.
----------------
Now Dubro (also calls himself AVRO) is going to plug these conversations into Grok and spam Grok's reaction here and try to get the forum in trouble while also trying to bury the posts of other posters? Or is he trying to annoy Elroch and MARattigan?
Or is his idea to try to make pseudo-intellectuals like the two O's and C-rat look good?
Could have all those effects ... whatever Dubro thinks he doing.
Dubro would be stopped by blocking. But there's no opening poster here to do that.
Chess.com ought to have a feature whereby the moderators and staff can selectively block whoever from a forum they're trying to destroy - when there's no opening poster account in existence.
Staff can hit their buttons much faster than Dubro can make new accounts.
How soon would people start beating each other discussing this question in real life face to face?
Mar 14, 2025
0
#15725
emilio1689 wrote:
How soon would people start beating each other discussing this question in real life face to face?
they'd be arguing about other things before this one.
Mar 14, 2025
0
#15726
@MARattigan
my suggestions aren't the same as tygxc's.
I already qualified how.
You missed it.
You continue to be wrong because you want to equate 1) e4 e5 Ba6 with the very different process of adding material to endgames you know are won endgames.
I talk about won positions - you talk about draws.
I talk about endgames - you talk about tygxc.
Invoking tygxc proves nothing.
------------------
Is Martin annoyed by the antics of Dubro/AVRO?
Martin prefers better for the forum. I'm with him there but not on everything.
Members can prefer that the spam-trolling by AVRO stops but its better a matter of preference than annoyance.
Like O - AVRO will work hard to disrupt the forum ... fact of life on chess.com and the internet.
Its almost like rain falling.
Don't let the rain get you down. There's a song about that too.
Mar 14, 2025
0
#15727
playerafar wrote:
@MARattigan
my suggestions aren't the same as tygxc's.
I already qualified how.
You missed it.
You continue to be wrong because you want to equate 1) e4 e5 Ba6 with the very different process of adding material to endgames you know are won endgames.
I talk about won positions - you talk about draws.
I talk about endgames - you talk about tygxc.
Invoking tygxc proves nothing.
------------------
Is Martin annoyed by the antics of Dubro/AVRO?
Martin prefers better for the forum. I'm with him there but not on everything.
Members can prefer that the spam-trolling by AVRO stops but its better a matter of preference than annoyance.
Like O - AVRO will work hard to disrupt the forum ... fact of life on chess.com and the internet.
Its almost like rain falling.
Don't let the rain get you down. There's a song about that too.
To make matters simple, can we agree that the following two extracts A and B from your original post #15543 are errors?
A.
The idea of computer projects seeking to solve chess by shortening the process with no further handling of positions that are obviously wins (corresponds to 'resign' in chess games) is a valid one.
...
tygxc took a little run at that but in a hopeless way.Like for example wanting to reject all further analysis after e4 e5 Ba6.It 'looks' valid. But the plus of a bishop isn't always enough to win.Even with 'ceteris paribus' factored in.
B.
In all the solved tablebase positions (in other words all 7-piece or fewer legal chess positions) that have been found to be wins for white or for black -
Each won position has a further algorithm run on it -
where adding more pieces to the winning side in all ways that do not interfere with that side's win - is considered - but with no further evaluation - they are simply counted and added and a number of such is determined.
A huge amount of work could then be saved.
That's a four way Y/N,Y/N answer that doesn't depend on what either of us might want to talk about.
Can you just give your answers please?
------------------
The only difference I see see between A and @tygxc's assertion you give is the size of the material difference. You initially invoked @tygxc. My referring to him is not meant to be a case of "four legs MARattigan good, two legs @tygxc bad", rather that the arguments I believe adequately disposed of the @tygxc argument you mention also dispose of A.
---------------------------------
Yes, Dubro/AVRO is a pain in the neck, but we've had (and have) others. He obviously has a personality disorder which probably renders his life quite unpleasant, so maybe we should have some compassion (while still hoping to see the back of him).
---------------------------------
If you were to amend B to save the actual positions and prime a tablebase generation process with the results, we could discuss the likely overall efficiency consequences if you're interested.
Mar 14, 2025
0
#15728
playerafar wrote:
... or from the front (runs into the Shannon number of 10^120 ...
That's an irrelevant figure.
Grok3 regularly includes it. Did it come from Grok or ChatGPT?
Best current lower bound on the number of possible games would be 10^29241 according to https://www.wismuth.com/chess/longest-game.html
(if 75M/5R rules are in effect).
Mar 14, 2025
0
#15729
@MARattigan
this from your post just now:
"The only difference I see see between A and @tygxc's assertion you give is the size of the material difference."
Why is that the only difference you see Martin?
Why don't you 'see' (by remembering what you know) that a 'won endgame' is very very different for 'solving chess' purposes from a blunder of a piece in the opening?
The differences are Gigantic.
The 'only difference' you want to see?
But instead of making previous semantics a conversation -
there's the idea of looking at the actual idea.
It may be that no force on earth will get you to look at the idea of adding pieces to the winning side that already has a won position.
---------------
The detail that its a win becaue of lopsided material advantage - is just that - a detail
That detail is necessary - because if the win was a positional/tactical one (including where the winning side is down a lot of material) rather than a lopsided material one - then adding pieces to the winning side could actually mess up the win.
For example by blocking his pieces that would otherwise be checkmating.
But now you've been diverting to the detail.
When are you going to address the main points Martin?
(I'm not holding my breath though. I'm on very solid ground here.)
------------------
What would a real opposition look like to the suggestion that adding more pieces to the winning side who is winning because of lopsided material advantage - would produce positions that are also winning provided that stalemate positions are excluded?
A 'nimble' opposition ...
A) Hey player - even if you can take a nice big bite out of John Tromp's number with that - that still leaves too big a number!
(multiple replies to that: One of them is 'but it still would leave a smaller number - provided the computer counts up all the obvious wins thus generated - and subtracts them)
B) Player - if that idea's good how come nobody thought of it before?
(multiple replies possible including: 'They would have. Because its not a complex idea.
They would have thought of that before and after computers got developed. Where 'they' refers to a lot of people.
C) Player - if 'they' thought of that - how come they haven't published about that?'
(again - multiple possible replies including: 'Whoever's going to try and publish something usually considers whether they can get paid for it. Both writers and publishers.'
D) OK - but then why didn't they just go ahead and do something like that?
(Multiple replies again .... including the immortal
😎
Mar 14, 2025
0
#15730
You seem to have inordinate difficulty with the word "yes" and "no".
Mar 14, 2025
0
#15731
MARattigan wrote:
You seem to have inordinate difficulty with the word "yes" and "no".
No difficulty at all.
Are you having a problem saying 'yes' to 'won endgames are different from piece blunders in openings'?
'inordinate difficulty' in seeing the difference?
(I could fall into a trap here ... a 'pot' that chessplayers often fall into of seeing many conversations as Bobby Fischerism contests.)
-------------------------------
Martin - with due respect to you - if you want a better forum - then do better.
I'll keep the 'follow' button turned on - just for you Martin?
But try to do better than trying to equate two things that you keep forgetting are different M.
--------------------
Regarding the AVRO account - maybe the moderators will shut him down.
But it surely would be much better if the site administration figured out a method of designated staff having a block button for designated forums with no opening poster.
But - can't have everything.
Its a great website. Imperfect like most other things in an imperfect world.
> Regarding the AVRO account - maybe the moderators will shut him down.
For what?
Mar 14, 2025
0
#15733
@playerafar
You still seem to have inordinate difficulty with the word "yes" and "no".
I would certainly say "yes" to, "Is a won endgame different from piece a piece blunder in openings?" (put as a yes/no question). The one is play from a position or class of positions. The other is probably a single move or a short sequence.
Mar 14, 2025
0
#15734
MARattigan wrote:
You still seem to have inordinate difficulty with the word "yes" and "no".
None at all.
You slipped up with the word 'seem'.
See my post 15651 you're having 'inordinate difficulty' with.
And see you Later.
-------------------
emilio if you know that Dubrovnik-1950 was muted and he cancelled in retaliation -
and you know its obvious that AVRO is the same person -
then you'll understand 'maybe'.
Maybe.
See you later 2.
I know neither Dubrovnik-1950 nor AVRO
Mar 14, 2025
0
#15736
playerafar wrote:
MARattigan wrote:
You still seem to have inordinate difficulty with the word "yes" and "no".
None at all.
You slipped up with the word 'seem'.
See my post 15651 you're having 'inordinate difficulty' with.
And see you Later.
-------------------
...
OK then. You do have inordinate difficulty with the words "yes" and "no".
You've posted lots of stuff in response to my #15647 but no "yes" or "no" to the straightforward questions I asked.
And I did answer "yes" or "no" to the question in #15651. Did you not notice?
By the way, you've made a fair number of posts on the subject of specific people being incapable of admitting their mistakes in the past. By avoiding a straightforward answer to my #15647 questions, you wouldn't be guilty of that yourself, by any chance?
Mar 14, 2025
0
#15737
playerafar wrote:
Elroch wrote:
Pedantical, the AI copy and paste is incorrect. To solve chess you need to know the initial state and the rules about moving (each of which maps a state to a state) and terminal states. Nothing else.
For example, when you are presented with a chess problem, you are being given an initial game state and asked to solve it, and you can, given knowledge of the rules. The only difference is that for solving chess you are given the initial position, and the analysis needed is a bit (ahem) larger.
Also, the reference to the game tree - it says
"...the 1012010^{120}10120 game tree. Solving chess means navigating that tree,"
is at best misleading (at worst simply wrong). You do NOT need to navigate the game tree, just a tiny fraction of it (very rough estimate 10^30 nodes). Again, this is like for solving a chess problem: if you guess the first move correctly, you only need look at lines that start with that first move - you have successfully ignored all the branches of the game tree that start with other moves. And this can happen at any stage in the solution, not just the first move.
And again, the same is true of solving chess, in principle (it's just impractically large).
For computers to solve chess - they don't need to make moves or play a game.
The first is an irrelevant distinction. A solver needs to PROCESS moves. Processing a move involves either - in forward analysis - generating the state that is reached from a given state and a given move, or - in retrograde analysis, used to generate tablebases - generating the previous state that would lead to a given state via a given move. In the former case, a complementary step is generating the set of legal moves from a state, in the latter a complementary step is generating the set of legal moves that would lead to a state.
They don't need 'game tree' either.
Well, conceptually what is needed is a graph (don't argue about the semantics - this is the correct word and has been for a century or so) of states with moves connecting them.
In implementation, what you need is a database of states added to by processing moves. Once you have generated a state AND the set of moves that lead to other states (in one direction or the other) you are no longer interested in where it is in the tree. i.e. you can discard part of the info about the tree, leaving only what needs processing.
Interesting fact, when checkers was solved storage was not as plentiful as now, so it was impractical to store all the states needed at the same time. This posed a challenge, since naively you want the entire list of already visited states in the database to see if a reached state is already in the list. But they realised that in order for the large majority of states that were relevant to still be there, they could work through the tree in a way that discarded most of the previous work, even if sometimes it would involve processing the same state from scratch more than once. These days, checkers would be solved with the entire analysis preserved - it's only about a terabyte needed.
By analogy, it could well be that if computational power increased to permit the solution of chess, the same issue would occur - that there wasn't room to store the whole solution and it would be done in a piecemeal fashion, discarding previous analysis to make room for new branches.
You can link positions by addiing to the material and varying where the additional material is placed (not moved). Would work much better with won positions than with drawn positions.
I am not sure how useful this is. Firstly, there are many ways it could go wrong, limiting the number of positions added. Secondly the fraction of total positions involved would likely be tiny.And programmers will use computer jargon for chess-solving projects.
Do they have to use game theory jargon too to do that?
No but they'll do that too.
When you look at a field from outside, it is appropriate to accept the terminology - it works and you have no objection other than subjective preference.
But to do the job right and get into alternative solving routes - they'll have to do better than asserting jargon.
--------------------
Long arguments defending game theory jargon ...
Productive people in this field don't spend their time defending the terms used (and certainly not attacking the terms used), they use them. And it works.
I have the green background and my eyes bleeding when I read the red text above it
Mar 14, 2025
0
#15739
Pick a colour you prefer - I just know red is good on normal white background.
Mar 14, 2025
0
#15740
Elroch wrote:
playerafar wrote:
...
You can link positions by addiing to the material and varying where the additional material is placed (not moved). Would work much better with won positions than with drawn positions.
I am not sure how useful this is. Firstly, there are many ways it could go wrong, limiting the number of positions added. Secondly the fraction of total positions involved would likely be tiny.
...
With @playerafar's proposed implementation - not useful at all. He proposes only counting the number of positions after his proposed piece additions without in any way processing the positions themselves.
He doesn't propose doing anything with the count when he's found it, so I would say the only thing that could go wrong is the counting process ties up the machine and prevents useful work.
I analysed it once. First posted on https://www.chess.com/forum/view/fun-with-chess/5-1-draw-contest which has the start of the analysis. (I got waylaid into a complete analysis of the king, 31 bishops and any number of knights v king endgame down to losing one bishop, which I finished, but it was very long and I reckoned everyone had stopped reading anyway.)