Chess will never be solved, here's why - Chess Forums - Page 770

TheChessIntellectReturns · 2022-01-16T22:50:19-08:00

Imagine a chess position of X paradigms. Now, a chess computer rated 3000 solves that position. All well and good. Could another computer rated a zillion solve that position better than Rybka? No, because not even chess computer zillion could solve the Ruy Lopez better than a sad FIDE master coul

DiogenesDue

Mar 4, 2025

0

#15381

Is crazed rat muted or just taking a break? Been wondering when Wynd would make good on his post of last week. If he is muted, I would have to say it does little good for setting a new tone if the censures happen in private. Rather, it should be public and the specific behavior that caused the action should be explained. I note that there are still casual "id*ot" and "mor*n" references being tossed around anyway, so...

playerafar

Mar 5, 2025

0

#15382

DiogenesDue wrote:

Is crazed rat muted or just taking a break? Been wondering when Wynd would make good on his post of last week. If he is muted, I would have to say it does little good for setting a new tone if the censures happen in private. Rather, it should be public and the specific behavior that caused the action should be explained. I note that there are still casual "id*ot" and "mor*n" references being tossed around anyway, so...

There appear to be at least three accounts associated with C-rat.
With the ibrust777 account having disappeared and the crazedrat1001 still existing unmuted and the crazedrat1000 account muted. Both rat accounts indicate 'Ian - United States' and both started in 2014 at different points in that year.
--------------------
And obviously Opto wants to be obnoxious enough so that people reply in kind and then he and various people shilling then report the well-meaning people who get baited into doing so.
And worth mentioning from time to time that Opto was muted for three months by chess.com recently. Why? So that members are informed that the staff knows something about him and that his issues go back several years.
------------------------
Forum subject:
leads to a lot of different conversations.
And its educational at times.
Set theory - game theory - limits of computers - history of math - quantum computers -
and of course the much vaunted and much defended 'nodes per second' which is associated with intense jargon.
Hey its funny at times !
Like technical jargon in law and medicine.

playerafar

Mar 5, 2025

0

#15383

"Tech 1: “You know that if we use an API to resolve this distributed link transactional situation, it could cause compromising of both the stack pointer and the serial bus, leading to over-activity in the NIC. That would conflict with nodes-per-second throughput in the CPU, causing a significant bottleneck. We might even end up with a memory leak in the kernel.”

Tech 2: “Yes, I had thought of that, but I think we can use a server-side buffer in the SWAP file to neutralize overflow into the environmental variables. We’ll just need to ensure that the I/O wait time doesn’t spike during peak load.”

Tech 1: “Gotcha. If we configure that correctly, we might even be able to virtualize this with some SDK script, coupled with legacy compatibility within the driver repository state. We could potentially offload the processing to a microservice architecture and distribute the load across the containers.”

Tech 2: “Exactly! And with some careful load balancing and throttling on the API calls, we could mitigate any chance of a race condition in the multi-threaded execution. That should also help with latency on the async callbacks.”

Tech 1: “Perfect. Let’s implement a rollback function on the transactional layer just in case we hit a deadlock—no need for us to trigger a kernel panic over this.”

Elroch

Mar 5, 2025

0

#15384

Dubrovnik-1950 wrote:

Grok's logic and reasoning is clearly better then Chatgtp. But Chatgpt is clearly better then Grok at generating photo art. Even if Chatgtp lost the h pswn.

ChatGPT uses better English. I find it a bit annoying that Grok 3 is "folksy" when discussing serious subjects. This seems to be an attempt to dumb it down for the masses.

Elroch

Mar 5, 2025

0

#15385

Optimissed wrote:

Dubrovnik-1950 wrote:

O is just nuts!

If game theory did not apply with chess. Then how does minimax strategy allow computers to play chess, or with the A-B search. To approximate the perfect strategy. Chess is not a island to itself.

And chess is absolute, and a zero sum game. With all chess position having a perfect play outcome of either a win, draw, or loss.

Chess is just too complex to know the perfect strategy, until you hit the portion of the game of chess we have solved, called Tablebases.

You aren't making sense. There are some things, like the excruciatingly simple minimax theory, that are basic to all games but also to competition in all its forms.

The perfect strategy for chess is obviously that there is no perfect strategy for chess.

While your sentence makes no sense, it is a theorem that should be familiar to everyone in this discussion that strategies for chess that achieve the game theoretic optimal value exist. See proposition 2.1 in this chapter of a book on game theory. Honestly, it's painless.

Note that this relies on finiteness. Basic chess has infinite games which never reach a conclusion, and the easy way to avoid that issue is to include a rule which ensures all games are finite (as is the case in chess as it is played on chess.com). The chapter does deal with the more difficult case of infinite games as well.

You can't win a daft argument by quoting a few totally obvious things and pretending they mean you won. Birds often have beaks! I won!

To deal with the equally absurd narcissistic claim that chess is not a game of perfect information, here is a formal definition:

A game of perfect information is a type of game where all players have complete knowledge of the game's state at every point in the decision-making process. Formally, it can be defined using the framework of an extensive-form game. Here's a mathematical definition:

A game of perfect information is an extensive-form game represented by a tuple ( G = (N, A, H, Z, P, u, I) ), where:

( N ) is a finite set of players.
( A ) is a set of possible actions available to the players.
( H ) is a set of non-terminal histories (sequences of actions taken from the start of the game), where each history ( h \in H ) represents a possible state of the game.
( Z ) is a set of terminal histories (complete sequences of actions that end the game).
( P: H -> N ) is a player function that assigns to each non-terminal history ( h \in H ) the player ( P(h) ) who moves at that point.
( u: Z -> R^n ) is a utility (or payoff) function, where ( u(z) = (u_1(z), u_2(z), ..., u_n(z)) ) assigns a payoff to each player ( i in N ) for each terminal history ( z in Z ).
( I ) is a set of information sets, which partitions the histories ( H ) into equivalence classes based on what players know when making a move.

For the game to be one of perfect information, the information sets must satisfy the following condition:

For every non-terminal history ( h in H ), the information set ( I(h) ) containing ( h ) is a singleton, i.e., ( I(h) = {h} ). This means that at every decision point, the player whose turn it is to move knows the exact sequence of actions (the full history) that led to the current state of the game.

In other words, there are no hidden moves or simultaneous decisions—each player is fully aware of all prior actions taken by all players when making their choice. Classic examples include chess and tic-tac-toe, where the entire board (and thus the history of moves) is visible to both players at all times.

Contrast this with games of imperfect information (e.g., poker), where players do not know certain aspects of the game state, such as the cards held by their opponents, leading to non-singleton information sets.

playerafar

Mar 5, 2025

0

#15386

@Optimissed has had 'perfect information' in the context explained to him a zillion times.

playerafar

Mar 5, 2025

0

#15387

Elroch wrote:

Dubrovnik-1950 wrote:

Grok's logic and reasoning is clearly better then Chatgtp. But Chatgpt is clearly better then Grok at generating photo art. Even if Chatgtp lost the h pswn.

ChatGPT uses better English. I find it a bit annoying that Grok 3 is "folksy" when discussing serious subjects. This seems to be an attempt to dumb it down for the masses.

Grok is quite colloquial at times.
Its sharper than chatgpt.
But we can expect the AI's to improve significantly year by year if not faster than that.
Can one extrapolate ahead to what AI might look like ten years from now?
Perhaps it will promote robots in households.
'Boil some water and make some coffee.'
'You didn't vaccum under the armchair'
'Pull out those weeds around the garage'.

playerafar

Mar 5, 2025

0

#15388

@ Dubrovnik - interesting that you could get chatgpt to produce that nice graph.
whereas Grok's attempt looks quite forlorn.
I guess you have to ask chatgpt in a certain way perhaps.

Elroch

Mar 5, 2025

0

#15389

To be frank, though I always have formal precision in mind, isn't the simple definition of a game of perfect information enough for most people?

A game of perfect information is a type of game where all players have complete knowledge of the game's state at every point in the decision-making process.

All that is implicit is that the "game state" determines the moves available, the states reached by those moves, and the value if the state is terminal (think "mate" etc.).

Can @Optimissed understand what a game state is?

Thee_Ghostess_Lola

Mar 5, 2025

0

#15390

hes just arguing (& spot on) a def. game theory is too broad. this is justa study about chess. iows chess is a simple (improper) subset a GT. theres a union but theres also a uniqueness...imho.

...and thx El for clearing it up .

ardutgamersus

Mar 5, 2025

0

#15391

no way ts shi still goin on lk bro js put the fries in the bag 😭

TCandyouknowitstrue

Mar 5, 2025

0

#15392

ardutgamersus wrote:

no way ts shi still goin on lk bro js put the fries in the bag 😭

Tell that to the nonsolvers who argue based on philosophy and human comprehension instead of objective mathematical facts

DiogenesDue

Mar 5, 2025

0

#15393

ardutgamersus wrote:

no way ts shi still goin on lk bro js put the fries in the bag 😭

Still going through your anime phase, I see. Things take time, apparently.

playerafar

Mar 5, 2025

0

#15394

Elroch wrote:

To be frank, though I always have formal precision in mind, isn't the simple definition of a game of perfect information enough for most people?

A game of perfect information is a type of game where all players have complete knowledge of the game's state at every point in the decision-making process.

All that is implicit is that the "game state" determines the moves available, the states reached by those moves, and the value if the state is terminal (think "mate" etc.).

Can @Optimissed understand what a game state is?

Its not about that. Its about the use of jargon in the forum being exploited by a particular person whose name begins with a round vowel to conduct a particular and familiar type of obnoxious operation.

playerafar

Mar 5, 2025

0

#15395

@ Dubrovnik
Your points are reasonable.
I composed a post just now addressing the 'worst offenders' in the terminology.
But came up with a 'shortcut' just now. Like this.
--------------------------------
Alternative words for the forum topic/opening post:
'Is a chess position a 'node' when talking about computers solving all of chess?
If that's true - then the opening position with 32 pieces on their original squares is a 'node'.
Right?
So then - what is the 'nodes per second' speed of supercomputers solving that one position?'
Answer to forum question:
The speed is much less than one triillionth of a trillionth of a node per second.
Reaction: 'Gosh that's pretty slow. Takes those Dev guys trilllions of years just to do one node??
A hamster could do better than that.'
---------------------
Possibility: 'Nodes per second' might be the worst offender of all the jargon terms.
But there's other 'candidates for the title'.

AnishAAgrawalChess

Mar 6, 2025

0

#15396

https://www.chess.com/c/zXGKa8Fg

AnishAAgrawalChess

Mar 6, 2025

0

#15397

Reset Checkmate

Thee_Ghostess_Lola

Mar 6, 2025

0

#15398

Where do you guys go to school.

the only person i ever told where was Varelse. and i told him to NEVER EVER tell a/o.

Elroch

Mar 6, 2025

0

#15399

Thee_Ghostess_Lola wrote:

Where do you guys go to school.

the only person i ever told where was Varelse. and i told him to NEVER EVER tell a/o.

It was Hogwarts, right?

playerafar

Mar 6, 2025

0

#15400

Regarding Dubrovnik's nice presentation of game theory (from Grok)
a certain passage stands out: I added a b c.
'Zermelo’s theorem, published in 1913, proves this:
in any two-player, finite, zero-sum game with perfect information, one of the following must be true:
a) Player 1 (White) has a strategy to win no matter what Player 2 (Black) does.
b) Player 2 (Black) has a strategy to win no matter what Player 1 does.
c) Both players have strategies to force at least a draw.'
----------
(Used single quote marks because its not a direct quote.)
-----------
Its neat in a way. Because each of the three situations seem to exclude each other.
But for much of a game they all appear to be possible. Not one but all three.
(yes - consequences of the game being unsolved)
It looks incomplete though.
It seems there ought to be at least a fourth category.
Only 'seems' though.
From a player's point of view games go through a metamorphosis with more than three possibilities.
Like 'unclear' - 'advantage to' - 'equal' - 'at least a draw' - dead draw - book draw - forced win - book win and so on.
Also players tend to distinguish 'tactical' from 'positional'.
Point: most of the unsolved nature of the game is concerned with positional - not tactical.