Chess will never be solved, here's why

A very good point.

Hi J !
No matter how good the algorithm ...
there's always a particular thing at the base of the whole project that limits it severely.
The number of operations per second that whatever computer can perform.
It can only work so fast.
Not 'nodes per second' but 'ops per second'.
The most basic binary ops.
Even with a million trillion ops per second (a quintillion) solving chess still too daunting.
You'd think that 10^18 would make a big dent in the 10^44 number ...
but those are just 'ops'.
You need many ops just to define a position on the board - let alone analyze it - let alone 'solve' it.
And there's no 'constant' on any of those either ...
the more pieces - the harder. Peaking at around 27 pieces on the board.
(not 32 - remarkable ey? but the reason has been discussed here)
Point: That impressive 10 ^ 18 number of 'ops per second' itself gets cut way down and in turn therefore can't make nearly as big a dent in the 10^44.
Which is just far too formidable a number for today's computers.
---------------------------------
Analogy: the closest star to earth's star the sun - is Alpha Centauri which is over 4 light years away.
That means if your spaceship could do 600 million miles an hour the journey would take over 4 years (hey but what about slowing down towards the end?)
But spaceships can't even do a 1000th of that speed.
You're looking at way over 4000 years to make that trip.
Would there ever be the money allocated for that - even if that amount of money existed to pay for it?
Point: Like with spaceships - the chess project is fundamentally limited by the ops speed of its processors.
Hardware. No matter how good the software.

Bolded parts incorrect.

'enough processing power' ...
hardware.
Good enough hardware.
And good enough software.
But the 'good enough' hardware just doesn't exist.
----------------
But it might eventually.
There's something in computing called 'Moore's Law'.

Moore's law is not really a law, but rather an empirical fact stated in 1965 (transistors on chip doubling every year), then corrected in 1975 (to doubling every 2 years). However, it has done well since then. (It is accurate to less than a year in 2024! Perhaps chip manufacturers design their objectives based on it, so a self-fulfilling prophesy? Such accuracy is impossible to explain otherwise. It is truly remarkable. (NVIDIA broke Apple's record for the last point on the graph).

Moore's law says that there will be more transistors on a chip in the year 2480 than there are atoms in the observable universe.

(Assuming 10^80 atoms in the universe, 208 billion transistors on a chip in 2024, doubling every 2 years).

It all comes down to if White having the first move is a significant enough advantage.

Xd?

Hah.
Idea: if the density of transistors per unit area or volume of chip correlates directly to processor speed and Moore's law continues to hold ...
(Yes that's right its not technically a 'Law')
then in 60 years there would be a multiplicative increase in that density by a factor of 2^30.
Which is over 1 billion.
In other words the 'problem' would be knocked down by over 10^9 in 60 years in other words to a billionth of what it was ...
but with these gigantic numbers like 10^44 that isn't getting the project 'out of the woods'.
A current computer capable of one quintillion ops per second could in theory increase from that 10^18 ops per second to 10^27 ops per second ...
leaving it with a number of 'seconds' of 10^17 to attack the Job of 10^44 positions.
There's about 31 million seconds in a year. That's appr. 3 x 10^7.
That still leaves you with 10^10 years.
Uh oh!
But then you have to knock that down by the number of positions that would likely be 'solved' each second with a computer speed of 10^25 ops per second.
Unfortunately that isn't a constant nor an average and therefore isn't known but would grow with the number of pieces on the board.
----------------------------
Are the tablebase projects using Quintillion ops/seconds computers?
If they are - they're struggling to get from 7 pieces to 8 pieces.
That tells you something about how tough the project is.
Whatever hardware is being used.

And would 'Moore's Law' continue to hold?
Not if it runs into fundamental barriers.

@13439

"It all comes down to if White having the first move is a significant enough advantage."
++ It is not. +1 tempo = +0.33 pawn = not enough to win. You can queen a pawn, but you cannot queen a tempo. Moreover, each further move dilutes the tempo advantage.

so, wheres the deductive proof of this?

oh wait, you dont have any proof. thats why youve ignored the rebuttals to your delusions from 5 separate people.

@13446

"There is no such thing as 0.33."
++ There is: we know from gambits that 3 tempi = 1 pawn. Thus 1 tempo = 0.33 pawn.

"Some 0.33s go to 0" Yes, normally. "some go to a win for white" ++ If black makes a mistake. "some go to a win for black." ++ If white makes a mistake.

"whether a large enough number of them go to 0.00 such that chess is a draw with best play."
++ Yes it is. E.g. 112 draws out of 112 games

"That is (obviously) an open question" ++ The question is resolved, a draw it is.

"There is no such thing as 0.33."
++ There is: we know from gambits that 3 tempi = 1 pawn. Thus 1 tempo = 0.33 pawn"

Where in the rules of the game are there point values?

"we know from gambits that 3 tempi = 1 pawn"

wheres the deductive proof?

""whether a large enough number of them go to 0.00 such that chess is a draw with best play."
++ Yes it is. E.g. 112 draws out of 112 games"

wheres the proof?

you cant make statements of certainty without proof. even a 10 year old knows that.

tygxc all we are asking is a deductive proof from the axioms of chess of any of your statements..

@13450

"this rule of thumb is common knowledge"
++ It is derived from centuries of gambit play and thousands of games.
It gives the exchange rate between time and material.

"not an exact measurement of a single tempi in the starting position"
++ It should be exact: tempi come in natural numbers.
1 pawn > 2 tempi, 1 pawn < 4 tempi, 1 pawn = 3 tempi.
Even if it were not exact, it is clear that 1 tempo < 1 pawn.
+1 pawn is enough to win: queen the extra pawn.
+1 tempo is not enough to win: you cannot queen a tempo.

Moreover, each further move dilutes the advantage of +1 tempo:
0-0 -> 1-0 -> 1-1 -> 2-1 -> 2-2 -> 3-2 -> 3-3...
That is also why the initial +0.33 gradually evaporates to 0.00 if neither side makes a mistake.

How does this thread have over 13,000 posts?

basically it's one delusional guy (@tygxc) repeating his fantasy over and over again while a group of rational people repeatedly debunk him. the main core of his fantasy is that he doesnt understand what a mathematical proof is, and so makes only approximations as "proof" when a game solution is by definition a mathematical proof. theres also an occasional troll.

@13454

"Yes, the rule of thumb gives an approximate exchange rate for practical play."
++ Also theoretically.

"In any case, appeal to authority is a logical fallacy."
++ It is no appeal to authority: it has been tried thousands of times.
Players gave up a pawn for tempi. Players grabbed pawns and lost tempi.
Then the theoreticians summed up the findings.

Most gambits have been found unsound. Some gambits like the Catalan or the Two Knights Defense, the Marshall Gambit have been found sound. Some pawn grabbing losing tempi has been found sound, e.g. the Najdorf Poisoned Pawn.

There is no human bias: AlphaZero with only the Laws of Chess as input reached the same conclusions after millions of autoplay games.