Chess Will Never Be Solved. Why? - Chess Forums - Page 18

tygxc

Jun 14, 2022

0

#341

#439
"A daunting task indeed, but consider all the things that you can do with a hand-held device today that the supercomputers of 50 years ago could not do; you can't say with certainty that 23rd-century computers will not be up to the challenge."
++ Per Sveshnikov modern computers with good assistants can do it in 5 years.
The good assistants are essential to collapse the search space from 10^44 legal positions to 10^17 legal, sensible, reachable, and relevant positions.

not_cl0ud

Jun 19, 2022

0

#342

tygxc wrote:

#439
"A daunting task indeed, but consider all the things that you can do with a hand-held device today that the supercomputers of 50 years ago could not do; you can't say with certainty that 23rd-century computers will not be up to the challenge."
++ Per Sveshnikov modern computers with good assistants can do it in 5 years.
The good assistants are essential to collapse the search space from 10^44 legal positions to 10^17 legal, sensible, reachable, and relevant positions.

You're talking about sensible (and speaking of that...), but in chess, just think of all the possible positions. Like:

Like that lol (unless someone is you-know-what, they probably won't reach that)

not_cl0ud

Jun 19, 2022

0

#343

R.I.P

not_cl0ud

Jun 19, 2022

0

#344

Whereas tic-tac-toe is solved thanks to a quite small space of possible games, chess is nowhere near solved because its space of possible games far outstrips what could be dealt with by current computing technology. As noted in another answer, endgame tablebases exhibit optimal play for all positions with limited numbers of pieces.

not_cl0ud

Jun 19, 2022

0

#345

#460

playerafar

Jun 25, 2022

0

#346

The posts of @Optimissed have disappeared from this forum and the other chess will never be solved forum.
Which suggests that he has been muted by chess.com.
When I went to the forum there were a great deal of empty pages at the end.
Mutes of a member often cause that.
As to whether its one of those one day mutes or not - I don't know.
But my post here should fix the empty pages thing.

Duck

Jun 25, 2022

0

#347

Rip @Optimissed

playerafar

Jun 25, 2022

0

#348

He'll be back probably.
He's been muted before.

Ziryab

Jun 25, 2022

0

#349

Probably temporary. He’s not able to tell anyone because mutes apply to messages.

MARattigan

Jun 26, 2022

0

#350

playerafar wrote:

Regarding Haworth's law - I was able to bring it up on the net using @MARattigan 's hyperlink.
But it only appears fully - in pdf format.
Which is not exactly user-friendly to work with.
I refuse to use Word or Excel or PowerPoint ...
so I tried some pdf editor chrome extensions -
so as to get some copy/paste going with simple plain text.
No go. Same user-unfriendly.
When I removed the last editor from chrome - they wanted to know why so I told them.

But - Haworth's law isn't really a law - and it arises from 'Moore's Law' which apparently isn't a 'law' either.

As Haworth makes clear, Haworth's law is a conjecture. It isn't connected with Moore's law; the reference to Moore's law is just an analogy. Haworth didn't publish any predictions beyond three men more than had already been verified by tablebases. The confidence intervals quoted represent confidence that the results will turn out to be in the given ranges on the assumption that the conjecture is true, not confidence that the results will in fact turn out to be in the given ranges.

That doesn't mean they don't have some validity - nor that they do.
It would be nice to get them posted here.
Various people here - well it would be super-easy for them to do so I imagine.

I already posted the PDF several times; it's quite short. Here it is again. (Original at http://centaur.reading.ac.uk/36276/3/HaworthLaw.pdf.)

The last three points on the graph are predictions rather than verified values.

The question of validity is the same as that involved in your school science practicals when demonstrating things like Hooke's law, Charles law, the relationship between the length of a pendulum and the period of small oscillations and so forth (except I never managed such convincing straight lines in mine).

Just as the examples I mentioned are successful in predicting values at points other than those actually measured, Haworth's law, which was initially based on the 3-4-5 men Nalimov tablebases has successfully predicted values for the longest forced mates for 6 and 7 men (under basic rules) close to the values subsequently verified by tablebase generation.

On the other hand each of the examples I gave is actually only valid over some limited range. If you hang 1, 2, 3, 4 and 5 gram weights on your spring you get a nice straight line, but if you hang a fridge on the end you just get two bits of spring.

Similarly Haworth's law cannot continue indefinitely. If 𝑳(𝒏) denotes the length in ply of the longest forced mate under basic rules with 𝒏 men on the board, then obviously

𝑳(𝟔𝟒)⩽𝑳(𝟔𝟑)+𝟏

so the law must break down at least by the time 𝒏=𝟔𝟒. (Positions with 64 men on the board are illegal, but that - as with almost all illegal positions - is an "accident" of the starting position and not taken into account in the tablebases so far produced, ergo also not in Haworth's law.)

So I would say a longest forced mate of around 3 trillion moves with 32 men on the board would be a best guess for the depth of the longest mate with 32 pieces on the board under basic rules, but an insecure guess. It also has to be remembered that there are 105,959,504 endgame classifications with 32 men but only one that contains any legal positions.The longest mate(s) are unlikely to be from legal positions and it's extremely unlikely that the starting position is one such.

playerafar

Jun 26, 2022

0

#351

@MARattigan
Good that the two forums are now woken from those blank pages.
Thank you for the pdf's.
Is there any way you can post them more enlarged?
I'm really bad with pdf stuff ...
Probably my fault I don't know how to enlarge it.
But if you do enlarge it - then a lot of people might read it.

Regarding the damage to these forums with the 'nodes per second' foolishness ...
In my opinion its still not too late for them to be rescued from that.
So that hundreds if not thousands of people can then relate the topic to the realistic speeds of computers instead of that issue being constantly obfuscated.

MARattigan

Jun 26, 2022

0

#352

Just use the browser zoom. But it's actually converted to png, so you won't be able to copy text from it - you need to do that from the original. If I enlarge it the site will just shrink it again to fit.

I don't think it's really worth editing the pdf.

playerafar

Jun 26, 2022

0

#353

You're right Martin.
200% Chrome zoom was more than enough.
If I ever hit on a way to enlarge pdf and post it -
or convert it to plain text so that it'll present properly in forums -
then I'll enlarge it here and many more people will read ...

then the pdf won't look like old manifests about Sunken Treasure ships from 500 years ago !

Amazing that in 2022 such pdf's could still be user-unfriendly.

MARattigan

Jun 26, 2022

0

#354

I've added a link to the original again. PDF editors I've used tend to be either a pain or cost money. Maybe there's a latex unPDF.

playerafar

Jun 26, 2022

0

#355

If we got really super-beautiful forums going here about solving chess ...
and the research project guys showed up here -
they're still not going to mail us checks for $50,000.
Or even for $0.05.

That's meant as a joke about time investment here.
The nodes/square root guy isn't going to get a check either.

MoveNotToMove

Jun 26, 2022

0

#356

After 1.e4 tgere are exactly 20 moves: 16 pawn moves and 4 knight moves.

mpaetz

Jun 26, 2022

0

#357

Optimissed wrote:

On this occasion, it was due to using a perfectly innocent word meaning "blunder". I'm afraid that Chess.com has very deep-rooted and probably incurable psychological problems. Behaving the way they do, people will just get more and more contemptuous of the "rules". On the last occasion it was a similar problem with a different word. I've a feeling Chess.com is racist. But thanks for your concerns.

Chess.com has a list of "offensive" words that cause automatic mutes. Note that there is another forum going on--"muted for typing an author's name"--about someone using Philip K. Dick's name without constantly including the "Phillip K." part. I've gotten an immediate 24-hour mute for saying "nobody gives a rat's a**" but neglecting to * out the s.

Probably they've gotten enough complaints from parents over the years that the list of forbidden words is the easiest way for them to deal with it.

playerafar

Jun 26, 2022

0

#358

Its also easy to get nailed because of quoting something too.
Copy pasting. Whatever.

tygxc

Jun 27, 2022

0

#359

#472
"After 1.e4 there are exactly 20 moves: 16 pawn moves and 4 knight moves."
++ That is right, but of the 20 only 2 to 4 make sense.

tygxc

Jun 27, 2022

0

#360

#459
"chess is nowhere near solved because its space of possible games far outstrips what could be dealt with by current computing technology"

++ There are far less chess positions than chess games. The number of legal, sensible, reachable, and relevant chess positions is 10^17. That takes 5 years on existing cloud engines.