Chess will never be solved, here's why - Chess Forums - Page 728

playerafar

Feb 21, 2025

0

#14541

Which is false.

playerafar

Feb 21, 2025

0

#14542

power_9_the_people wrote:

The time to get chess solved is never

The time to make up your mind about people , on the other hand....

From some AI folks:

""It's not currently possible to solve chess within a practical time frame, and it's unlikely that it ever will be.
Explanation
The complexity of chess is so great that it's not possible for modern computers to solve it.
In 1950, Claude Shannon calculated that it would take 1090 years for a computer to make its first move in chess.
Computer analysis has solved chess for almost every position with up to seven pieces.
However, solving all eight-piece positions is still ongoing.
The complexity of the game and its infinite possibilities make it unlikely...."

The possibilities aren't infinite.
But that number from John Tromp formulated in the year 2000 is much too tough for today's computers to have a hope of solving chess anytime soon.
But computers and solving methods will improve.
Have been improving.
When they were building the 'chunnel' under the english channel the two teams tunneling from France and England met in the middle.
Is it possible in theory that a table base project analyzing from endgames - will meet a 'game tree wide forward search' in the middle?
Arguably happens every time a game reaches a book ending.

playerafar

Feb 21, 2025

0

#14543

Book ending. Not book.
I can further qualify 'book ending'.
I was referring to 'solved' book endings.

DiogenesDue

Feb 21, 2025

0

#14544

playerafar wrote:

The possibilities aren't infinite.
But that number from John Tromp formulated in the year 2000 is much too tough for today's computers to have a hope of solving chess anytime soon.
But computers and solving methods will improve.
Have been improving.
When they were building the 'chunnel' under the english channel the two teams tunneling from France and England met in the middle.
Is it possible in theory that a table base project analyzing from endgames - will meet a 'game tree wide forward search' in the middle?
Arguably happens every time a game reaches a book ending.

Tromp's 4.8 x 10^44 is from 2021.

DiogenesDue

Feb 21, 2025

0

#14545

The-New-Maximum wrote:

fire trucks are just watertrucks

Every so often some person gets it into their head to post a small explosion of one-liners they think are clever. Ostensibly this is to derail the thread, but it's more often a cry for help, driven by a resentment of not being noticed or included in things in their lives, which they transfer to the smaller microcosm of a forum thread with many participants.

DiggyDug04

Feb 21, 2025

0

#14546

Magnus Carlsen on Joe Rogan

DiogenesDue

Feb 21, 2025

0

#14547

Optimissed wrote:

Could be to irritate you.

That would fall under "derailing the thread".

DiogenesDue

Feb 21, 2025

0

#14548

Optimissed wrote:

DiogenesDue wrote:

Optimissed wrote:

Could be to irritate you.

That would fall under "derailing the thread".

Not logically speaking. It could have been to derail you? And preserve the thread?

You're doing no better than he did on that score. You never do.

playerafar

Feb 21, 2025

0

#14549

Dubrovnik versus Opto?

'Dub' would be doing better ...

playerafar

Feb 21, 2025

0

#14550

DiogenesDue wrote:

playerafar wrote:

The possibilities aren't infinite.
But that number from John Tromp formulated in the year 2000 is much too tough for today's computers to have a hope of solving chess anytime soon.
But computers and solving methods will improve.
Have been improving.
When they were building the 'chunnel' under the english channel the two teams tunneling from France and England met in the middle.
Is it possible in theory that a table base project analyzing from endgames - will meet a 'game tree wide forward search' in the middle?
Arguably happens every time a game reaches a book ending.

Tromp's 4.8 x 10^44 is from 2021.

Well then I got it wrong. I read it was from work J. Tromp did in 2000-2002.
Checking again just now he apparently came up with his first number of 10^45.8 chess positions as a lower bound in 1998.
https://www.chessprogramming.org/John_Tromp

Elroch

Feb 21, 2025

0

#14551

Dubrovnik-1950 wrote:

You fools just need to stop. You are trying to hang your hat on a Wikipedia definition that does not apply to chess. It is for games like Star Craft, and other type games where everything can happen at the same time.

Wikipedia even says by name Chess is a perfect information game.

-------------------------------------

In economics, perfect information (sometimes referred to as "no hidden information") is a feature of perfect competition. With perfect information in a market, all consumers and producers have complete and instantaneous knowledge of all market prices, their own utility, and own cost functions.

In game theory, a sequential game has perfect information if each player, when making any decision, is perfectly informed of all the events that have previously occurred, including the "initialization event" of the game (. the starting hands of each player in a card game).[1][2][3][4]

Perfect information is importantly different from complete information, which implies common knowledge of each player's utility functions, payoffs, strategies and "types". A game with perfect information may or may not have complete information.

Games where some aspect of play is hidden from opponents – such as the cards in poker and bridge – are examples of games with imperfect information.[5][6]

------------------------

Examples
[edit]
Backgammon includes chance events, but by some definitions is classified as a game of perfect information.
Poker is a game of imperfect information, as players do not know the private cards of their opponents.
Chess is an example of a game with perfect information, as each player can see all the pieces on the board at all times.[2] Other games with perfect information include tic-tac-toe, Reversi, checkers, and Go.[3]

I have to say, I am not that keen on the wording of the definition in that wikipedia article. What matters is full knowledge of the state, where the state is defined to determine either the value (if it is a terminal state) or the legal moves (where a legal move is characterised by the state it reaches and nothing else. There is no need to have knowledge of the past beyond the state that has been reached: it just happens that in a large class of games including chess the history of moves determines the state.

[For clarity in chess you also need some information about the history for the repetition draw rule, but this history is only needed back to the last irreversible move. Of course castling rights and any en passant possibility are part of the state. Thus a full PGN suffices, but generally has superfluous information - the moves before the last irreversible move].

MARattigan

Feb 21, 2025

0

#14552

Elroch wrote:

Let me emphasise that the pure game of chess - not the unimportant and separate mechanisms for negotiating a result, which are a separate topic which only exists for convenience - is unambiguously a game of perfect information.

That's fine. There are many versions of chess. Pure chess, had we but a set of rules for it, may well be unambiguously a game of perfect information. FIDE rules versions of chess on the other hand are unambiguously not games. of perfect information, the problem being that mechanisms for negotiating a result are not in fact separate.

It is difficult to imagine that any valid stronger definition could exclude it (I have no reason to believe one exists), and it satisfied the first definition of this term (I believe the strongest definition was the first).

If you're talking about the definitions of "perfect information" in Wikipaedia, I think it would be weakest in terms of your earlier definition. The later concepts impose extra constraints, so a game satisfying those would imply it also satisfies the first.

For this purpose, chess is just like tictactoe from a theoretical point of view. The only difference is that the game tree is bigger and the practical significance of the difference in complexity is not of theoretical importance. Anyone who would dispute chess being a game of perfect information because of oddities of the negotiation of a draw should make the same claim of tictactoe.

When I learned noughts and crosses there were no rules governing resignation and agreed draws. There is no need for them, you can obviously just do them anyway.. There should also be no need for them in the FIDE laws, but be that as it may they are there.

Note that in the PGN of a game and reporting on it, none of the extraneous negotiation usually appears. Chess players understand that from theoretical point of view what matters is the moves played and the result (which is generally the "right" one, though there are anomalous losses on time (something game theory has little interest in) and mistaken agreed draws and resignations).

Not so relevant to the question of which versions of chess have perfect information. That follows only from the definitions of perfect information and the rules of the particular version.

I don't even think of these practical things as being part of chess as it is of interest to game theory, any more than for checkers or connect4. It's all about legal sequences of moves and results.

Legal sequences of moves of the pieces or actions changing the game state? I could be wrong, but I think connect4 is a sequential game and the two things are the same.

Elroch

Feb 21, 2025

0

#14553

MARattigan wrote:

Elroch wrote:

Let me emphasise that the pure game of chess - not the unimportant and separate mechanisms for negotiating a result, which are a separate topic which only exists for convenience - is unambiguously a game of perfect information.

That's fine. There are many versions of chess. Pure chess, had we but a set of rules for it, may well be unambiguously a game of perfect information. FIDE rules versions of chess on the other hand are unambiguously not games. of perfect information, the problem being that mechanisms for negotiating a result are not in fact separate.

It is difficult to imagine that any valid stronger definition could exclude it (I have no reason to believe one exists), and it satisfied the first definition of this term (I believe the strongest definition was the first).

If you're talking about the definitions of "perfect information", I think it would be weakest in terms of your earlier definition. The later concepts impose extra constraints, so a game satisfying those would imply it also satisfies the first.

Nope, definitely what I said. Versions that permit either (1) chance, or (2) simultaneous moves (as referred to in the wiki article) are clearly weaker conditions, encompassing a larger set of games. Chess is a game with no chance and no simultaneous moves (like tictactoe, checkers, etc).

For this purpose, chess is just like tictactoe from a theoretical point of view. The only difference is that the game tree is bigger and the practical significance of the difference in complexity is not of theoretical importance. Anyone who would dispute chess being a game of perfect information because of oddities of the negotiation of a draw should make the same claim of tictactoe.

When I learned noughts and crosses there were no rules governing resignation and agreed draws. There is no need for them, you can obviously just do them anyway.

Yes, it's informal as far as I aware. I doubt there is an equivalent of FIDE for tictactoe . But people do agree draws and resign.

There should also be no need for them in the FIDE laws, but be that as it may they are there.

As I said, to game theory they are probably best thought of as a process of negotiation that is separate to the game itself - players agreeing on the value of the (pure) game although it hasn't actually been decided.

Note that in the PGN of a game and reporting on it, none of the extraneous negotiation usually appears. Chess players understand that from theoretical point of view what matters is the moves played and the result (which is generally the "right" one, though there are anomalous losses on time (something game theory has little interest in) and mistaken agreed draws and resignations).

Not so relevant to the question of which versions of chess have perfect information. That follows only from the definitions of perfect information and the rules of the particular version.

Chess has perfect information, once you separate the game of interest from the negotiation of a value.

I don't even think of these practical things as being part of chess as it is of interest to game theory, any more than for checkers or connect4. It's all about legal sequences of moves and results.

Legal sequences of moves of the pieces or actions changing the game state?

You recall that I defined a move as a pair of initial state and terminal state and nothing else. What's on the board is just a sort of encoding.

I could be wrong, but I think connect4 is a sequential game and the two things are the same.

Yes, with the abstraction that appeals to me, they are. I feel chess players who are not so mathematical would think more of the concrete pieces, the board and so on, but abstractly it's just state transitions.

MARattigan

Feb 21, 2025

0

#14554

Elroch wrote:

MARattigan wrote:

Elroch wrote:

Let me emphasise that the pure game of chess - not the unimportant and separate mechanisms for negotiating a result, which are a separate topic which only exists for convenience - is unambiguously a game of perfect information.

That's fine. There are many versions of chess. Pure chess, had we but a set of rules for it, may well be unambiguously a game of perfect information. FIDE rules versions of chess on the other hand are unambiguously not games. of perfect information, the problem being that mechanisms for negotiating a result are not in fact separate.

It is difficult to imagine that any valid stronger definition could exclude it (I have no reason to believe one exists), and it satisfied the first definition of this term (I believe the strongest definition was the first).

If you're talking about the definitions of "perfect information", I think it would be weakest in terms of your earlier definition. The later concepts impose extra constraints, so a game satisfying those would imply it also satisfies the first.

Nope, definitely what I said. Versions that permit either (1) chance, or (2) simultaneous moves (as referred to in the wiki article) are clearly weaker conditions, encompassing a larger set of games. Chess is a game with no chance and no simultaneous moves (like tictactoe, checkers, etc).

I did say according to your earlier definition, viz.

If condition A implies condition B, A is stronger than B. To put it another way the set of entities with property A is a subset of those with property B.

The games with the extra constraints are a subset of the games in the first definition. Note that the first definition permits chance, a later comment implies some definitions do not.

I just noticed it also says some definitions allow simultaneous moves as well as saying such games are not generally considered games of perfect information, so I think we're both wrong. The first definition is neither weakest nor strongest.

I have to also retract my earlier comment that FIDE rules versions of chess are unambiguously not games. of perfect information. They're just not generally considered so. (Lack of ambiguity appears to be difficult in this context.)

For this purpose, chess is just like tictactoe from a theoretical point of view. The only difference is that the game tree is bigger and the practical significance of the difference in complexity is not of theoretical importance. Anyone who would dispute chess being a game of perfect information because of oddities of the negotiation of a draw should make the same claim of tictactoe.

When I learned noughts and crosses there were no rules governing resignation and agreed draws. There is no need for them, you can obviously just do them anyway.

Yes, it's informal as far as I aware. I doubt there is an equivalent of FIDE for tictactoe . But people do agree draws and resign.

There should also be no need for them in the FIDE laws, but be that as it may they are there.

As I said, to game theory they are probably best thought of as a process of negotiation that is separate to the game itself - players agreeing on the value of the (pure) game although it hasn't actually been decided.

I did at one point consider posting sample soluble sets of rules based as closely as possible on the FIDE laws and proposed to simply excise the resignation and agreed draw arts. from the basic rules and everything but the 3/5R and 50/75M arts from the competition rules.

I gave up because I think the game described on an objective reading is nothing like chess as she is played (I'll come back to that later) and in any case everyone would just ignore it.

It's a shame that game theorists, instead of just nebulously talking about "chess" don't produce some actual rules for what they are talking about (which is never the FIDE laws). Then we might have some links to the versions of "pure chess" you refer to. (No use posting them here - @playerafar an @Optimissed . will ensure you'll never find them again after a few days.)

Note that in the PGN of a game and reporting on it, none of the extraneous negotiation usually appears. Chess players understand that from theoretical point of view what matters is the moves played and the result (which is generally the "right" one, though there are anomalous losses on time (something game theory has little interest in) and mistaken agreed draws and resignations).

Not so relevant to the question of which versions of chess have perfect information. That follows only from the definitions of perfect information and the rules of the particular version.

Chess has perfect information, once you separate the game of interest from the negotiation of a value.

I don't think that's true of FIDE competition rules chess. (There is more than one version of chess, more than one version discussed on the thread, even.)

I don't even think of these practical things as being part of chess as it is of interest to game theory, any more than for checkers or connect4. It's all about legal sequences of moves and results.

Legal sequences of moves of the pieces or actions changing the game state?

You recall that I defined a move as a pair of initial state and terminal state and nothing else. What's on the board is just a sort of encoding.

I could be wrong, but I think connect4 is a sequential game and the two things are the same.

Yes, with the abstraction that appeals to me, they are. I feel chess players who are not so mathematical would think more of the concrete pieces, the board and so on, but abstractly it's just state transitions.

Rayfamily

Feb 21, 2025

0

#14555

edited moderator AndrewSmith

Spam

Rayfamily

Feb 21, 2025

0

#14556

ill do a pushup for every comment

Elroch

Feb 21, 2025

0

#14557