It's a technical subject.
Chess Will Never Be Solved. Why?
Sort:
May 6, 2022
0
#163
MARattigan wrote:
It's a technical subject.
Correct. And 'somebody' disapproves of that.
But he shows up anyway.
May 6, 2022
0
#164
I guess the pingpong will go on for a while.
But my post below (which I deleted and improved several times)
is a reply to @MARattigan about the 50 move rule stuff.
Its obvious who personalizes by far the most (more than everyone else combined) and its not me nor Mar nor tygxc.
Got to log off soon but will get this reply to Martin done first.
////////////////////////////////////////
Regarding the 50 move rule - it can be considered an adjunct to chess -
like the issue of what is 'mating material' when somebody's flag falls in a sudden death game.
If the side whose flag didn't fall has an apawn or hpawn in a dead drawn position - then that side wins. For example.
In a situation where the 50 moves is counting - the exact number of moves left dictates strategy.
Especially for the defending side.
The rule could determine whether the position is a win or a draw anyway though. With or without strategy.
But that should be Skipped by the initial computer bank and Skipped as initial goal.
It should be skipped as Secondary goal too.
Should be low on the priority list. Lowest? Not necessarily.
///////////////////////////////
In a correct way to go at the hypothetical project -
I say - mating material and 50 move rule and repetitions of position are skipped. (a possibility of a repetition of position - if there's already been a 2-fold repeat - could again determine strategy. But could also determine whether its a draw or win because a defending side could perhaps force a final repetition)
But that should also be skipped in the initial stages too.
(nobody else in these forums has confirmed yet that he/she understands why.)
In other words - such a project would be done in Stages.
Multiple computers or banks of computers assigned.
So the first computer bank's job is to generate positions - not games.
Arrangements of pieces which do provide for either side to move.
(Otherwise they'd be the so-called 'diagrams'. Not good enough.)
Does that mean there's already a legality issue?
In some positions - it would be impossible for either side to move.
For example white's g-knight at f3. Every other piece at its home square.
Would it be possible for it to be white's move? I don't know.
Is there a sequence of moves that would bring that about ?
Should these issues of 'could it have got there legally?' be left out in the initial stages of the project ?
That's harder.
It depends on how positions to be solved are to be generated.
Somehow - simplicity has to be maintained - but without the ridiculous premises being pushed over and over again by @tygxc .
That's not personal. Simply addresses his posts.
Approach - define the initial goal for the initial computer bank.
Assign other relevant tasks to backup computer banks.
Suggestion for that initial goal:
A computer bank proceeds with position generation.
It then classifies every position it generates into categories and then refers each category to the appropriate secondary computer bank.
Of which there might be several.
And like the issue of whose move it is - castling and en passant should be provided for but not in a way that confounds and cripples the initial goals.
Other computer banks get those positions -
further down the Goals priority list.
Many ways to set up those Categories.
Part of solving - classifications of positions.
Links up Hugely with actual chessplaying.
But both the position generation and the classifications would be designed to fit with and greatly enhance the efficiency and effectiveness of the different computer algorithms assigned to each category.
That would be Top Priority.
Without such stages - the whole project degenerates.
May 6, 2022
0
#165
MARattigan wrote:
Optimissed wrote:
...
To be fair, the 50 move rule and 3-fold repetition make no difference to a real solution. I presume it's a real solution we're interested in. That is hardly even worth discussing since the logical points are so clearly obvious. We're talking about a solution for chess and different competition rules do not alter it. That's just a fact.
Absolutely not a fact.
Whoever first introduced the fifty move rule (Ruy López?) probably thought that. I would guess he'd looked at the KBNK endgame, decided it needed about 33 moves and added 50% for good luck and believed that would cover all situations.
No doubt he was regarded as a pretty good player, but couldn't conceive that if you add just three more pieces to the board you get the tablebase mate in 549 that requires at least 130 moves without a pawn move or capture (tablebases don't currently tell you exactly how many).
In fact he changed it into two different games with two different solutions. We know that the solutions are different for many situations with a small number of men. Solutions with a larger number of men will also already be changed if only because they have to take account of possible positions with a smaller number of men they can reach - but who can seriously doubt that even larger phases will occur with another three men added, for example.
The triple repetition rule will not change any solution.
It will change the necessary approaches to arriving at a solution. It will for example mean that positions with the same diagram and side to move that are arrived at by different sequences of moves don't necessarily have the same theoretical outcome, so @tygxc's habit of equating diagrams with nodes in his search space is (vastly) incorrect.
The tablebase generation algorithms bypass the problem of repetition by associating each position with the corresponding diagram and side to move together with an integer representing the distance to some objective that eventually terminates in mate. Each diagram and side to move is assigned only one distance to the objective so repetitions cannot occur.
@MARattigan
"so @tygxc's habit of equating diagrams with nodes in his search space is (vastly) incorrect."
That's right. It is incorrect but tygxc will keep repeating it anyway.
Also does tygxc really believe that 'Galileo' stuff he's pushing ? 
But the idea of approaching the project from the point of view of varying game rules ...
well there's a different approach available.
Which is to approach in Stages.
As to how the tablebases approach that aspect - I think it was talked about a bit - but it was already noted that the tablebases fail to take castling into account. Even just that.
In other words - having to factor in previous moves so greatly complicates that it isn't even practical even for just 7-piece positions !
Regarding the whole project of 'solving' chess using computers -
it is already started. Years ago.
It has already been under way and continues to be.
Namely - the tablebases. Multiple instances of them.
Regarding positions with just three pieces on board -
there's not much issue with castling in most of them.
But there's still a few.
Like with K+R versus lone King.
If the K and its rook are at their home squares - does castling matter there?
No.
Because both checkmate and stalemate positions could not exist in such homesquare positions.
But it is a Win for the side with the rook.
Further moves don't need to be considered.
But there's still a pitfall there.
And even moreso with K + R versus K + R.
Position generation.
If the 'castling possible' position can be dismissed as a win - does that mean that all positions that could arise from it are wins ?
No.
Because stalemate positions would be possible that way.
Moral: positions not to be generated by moves.
Got to be generated by placing.
Moves are still obviously relevant. But going forward. Not backward.
But for analyzing the position concerned. Not for 'position generation'.
That ends up crippling the project early stages.
Moral: Solve in stages. Pre-projects. Positions not games.
Games come later. Much basic chess teaching works that way too.
#203
"@tygxc apparently remains to be convinced"
++ Yes, tell me how many legal, sensible, reachable, and relevant positions there are and why. That is the time in nanoseconds it takes on 1 cloud engine of 10^9 nodes/s.
That is also the approximate minimal storage in bits (draw / no draw) to store the solution.
I believe 10^17 suffices and on 3 cloud engines that takes 5 years like Sveshnikov said.
A lot depends on what is considered relevant.
I consider e.g. 1 a4 and 1 e5 e5 2 Ba6 not relevant.
A node is a position plus its evaluation.
A position (FEN) is a diagram plus the side to move and the castling rights and en passant option.
For each diagram with white to move there are about 4 positions because of up/down symmetry and left/right symmetry after loss of castling rights.
There is a duality between games (PGN) and legal positions (FEN).
Each game (PGN) can be represented as a sequence of legal positions (FEN).
Each legal position (FEN) can be represented as a game (PGN) that leads to it from the initial position.
There are much more games >> 10^120 (PGN) than legal positions (FEN) 10^44.
Thus solving chess can only focus on positions.
Transpositions are generally handled by transposition tables.
That was also used in solving Losing Chess.
Chess with discretionary capture has more transpositions than Losing Chess with compulsory capture.
May 6, 2022
0
#167
"I believe 10^17 suffices"
It does Not.
"A node is a position plus its evaluation."
No it is Not. - for example - the opening position.
"solving chess can only focus on positions."
Positions should be the centerpiece.
Solving must focus on several things.
So 'only focus' is wrong too. 😁
Among many other errors. 😁😁
May 6, 2022
0
#168
Hi all,
Let's get back to topic and no personal attacks etc, we are adults?
Thanks everyone
#230
"No it is Not. - for example - the opening position."
++ "A node, in turn, is a chess position with its evaluation and history, i.e. castling rights, repetition of moves, move turn, etc."
https://chessify.me/blog/nps-what-are-the-nodes-per-second-in-chess-engine-analysis#:~:text=Nodes%20per%20second%20(NPS)%20is,thousand%20or%20one%20million%20respectively.
The opening position is:
rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w KQkq - 0 1
with evaluation +0.3 i.e. a draw.
"I believe 10^17 suffices. It does Not."
++ What number do you believe and why?
ChessFlair01 wrote:
I don't think chess will be thoroughly solved because 1 move like h3 on the first move can spread into millions on blunders, mistakes, or maybe even traps and advantages! I think it will not be 100% solved because there are around 80 moves you can do after e4, and combined with what your opponent does, solving chess will be simply impossible. See, chess has lasted at least 100 years, and not all tactics have been solved. Obviously, all the even weirdest moves in the opening have been analyzed, like white moving all the pawns to the third file! Weird! But even that is analyzed. But even so, has checkmate ever shown itself on the board and been analyzed? What position will that end up being? What piece will you ever checkmate with? This is all impossible to think about, and all moves go so deeply that sometimes even the most simple positions will never be solved.
There are millions of possibilities that even the human invented chess might remain a mystery...
May 7, 2022
0
#171
"I believe 10^17 suffices."
It does Not.
"What number do you believe and why?"
Believe ? Like its a religion ?
The point - the positions to be solved number at over 10^35.
Including by your own postings.
Apparently you have 'taken the square root' again.
In other words you've reduced the positions to less than a millionth of a trillionth of their actual value.
To shave off billions of years ! 
People have been over this with you many times.
But that's not personal. It concerns your postings.
These next three things might not happen completely:
1) post around the personalizations or hope the moderators/staff get that cut down.
2) also circumvent 'five years' and 'taking the square root' and meaningless 'evaluations nodes' and similiar.
3) somehow resolve things like the 50 move rule and repetitions and legalities.
In the other forum - they were to an extent.
But not in a way that made very firm progress.
Note that the people doing the tablebases would be far ahead concerning the vicissiudes of these projects.
And I"m going to repeat something:
Even just castling was so tough for them that it had to be left out of consideration for the 7-piece tablebases.
they probably know well how to generate positions to be computer-analyzed to facilitate the algorithms used.
As for me - I intend to post some basic positions. Using diagrams.
But not now.
#234
"Believe ? Like it was a religion ?"
++ As long as chess is not weakly solved, we do not know for sure how many positions need to be considered. Even Schaeffer who solved checkers predicted a different number 10^9 of positions than he ended up considering 10^14 in his solution on 2007. So the number is an opinion, a belief.
My opinion is 10^17 legal, sensible, reachable, and relevant positions.
"the positions to be solved number at over 10^35"
The number of legal positions is 10^44.
The number of sensible positions is around 10^32.
The number of reachable positions is far less: each pawn move and each capture renders huge numbers of positions unreachable.
The number of relevant positions is even less: e.g. 1 a4, or 1 e4 e5 2 Ba6 are not relevant.
That is how I estimate 10^17 legal, sensible, reachable, and relevant positions.
"People have been over this with you many times." ++ People do not understand.
"resolve things like the 50 move rule" ++ The 50-moves rule plays no role in solving chess.
"people doing the tablebases would be far ahead" ++ Strongly solving chess by generating a 32-men table base is not feasible with present technology. Maybe a quantum computer can do that: from 7 men to 8 men to 9 men... to 32 men.
"castling was so tough for them that it had to be left out of consideration for the 7-piece tablebases." ++ It is not tough at all, it just is not necessary and so they left that out as it saves storage so the 7-men endgame tablebase fits on a disk.
May 7, 2022
0
#173
"My opinion is 10^17 legal, sensible, reachable, and relevant positions."
First it was your belief now its your opinion.
Ever occur to you that you want it that way ?
Again - not personal because there's nothing 'evil' about wanting something to be.
Its your opinion that using a square root of positions is okay too?
And a 'node' solves a whole position in a billionth of a second ?
Is it 'lawyer talk' ? I don't think a jury would buy it.
Nobody has bought it here so far.
If castling was easy how about they would have included it for completeness?
How do you rule that out ?
"Strongly solving chess by generating a 32-men table base is not feasible with present technology."
We already know that and you've repeated that many times.
Its irrelevant to the point - which is that the tablebase researchers have already started the project.
And they would know the difficulties better than the people discussing here.
They're ahead.
Some reason to doubt that ? They made the tablebases unthinkingly ?
I'm thinking that not even one single person at any point has been able to reason one single point with you.
Not one. Not a personal attack. Just noteworthy.
"resolve things like the 50 move rule" ++ The 50-moves rule plays no role in solving chess.
Sure it does. Just not at very basic levels.
Its part of chess and its been explained how.
'People do not understand'
That you have pronounced yourself as a kind of chess Galileo ?
People do understand you saying that.
But there's a difference though.
Galileo was right.
Games will always be analyzed down, it's just that you really can't put an exact finger on how people determine their moves. That is one of the factors that we have to understand as people. We cannot determine nobodys move, but we can most definitely determine our own.Fact......
#236
"First it was your belief now its your opinion."
++ Belief, opinion, estimation, guess: all the same it is almost know.
This is the key question about the feasibility of weakly solving chess: how many legal, sensible, reachable, and relevant positions are there? My take is 10^17. What is yours?
"Its your opinion that using a square root of positions is okay too?"
++ The conversion from sensible positions to reachable positions is a key point. We know how Checkers and Losing Chess were solved with about the square root. For chess it may be more or less.
"And a 'node' solves a whole position in a billionth of a second ?"
++ No, a node does not solve anything. The whole set of 10^17 nodes solves the game.
"Nobody has bought it here so far."
++ Nobody understands so far.
"If castling was easy how about they would have included it for completeness?"
++ It is easy, it just needs more storage and then it no longer fits on the disk of a desktop.
"Strongly solving chess by generating a 32-men table base is not feasible with present technology. We already know that and you've repeated that many times."
++ You are the one who keeps mentioning table bases.
"the tablebase researchers have already started the project."
++ They are now completing the 8-men endgame table base, that is a long, long way from 32.
"I'm thinking that not even one single person at any point has been able to reason one single point with you." ++ So far not, try harder to understand. Most people do not reason at all and just spew some insults and ridicule.
"The 50-moves rule plays no role in solving chess. Sure it does."
++ No, it plays no role at all. In no GM game or ICCF WC game is the 50 moves rule ever involved with > 7 men. ICCF WC games end on average at move 39. It is just a practical rule to avoid a game and thus a whole tournament to drag on for weeks e.g. as a weak player cannot checkmate KBN vs. K.
"That you have pronounced yourself as a kind of chess Galileo ?"
++ This refutes the pseudo argument that I would be wrong because I am in the minority.
"Galileo was right." ++ I am right too, or, more exactly: Sveshnikov was right.
May 7, 2022
0
#176
Looks like you're using the square root.
I thought so.
In other words simply skipping all but one millionth of a trillionth of positions because somebody found a way to justify it in checkers.
"They are now completing the 8-men endgame table base, that is a long, long way from 32."
We already knew that. And water is wet too.
"No, a node does not solve anything"
You might have slipped up there and admitted something accurate.
Plus its chess not checkers. A 35 digit or so number of positions to be solved.
How many times have people told you that you can't accurately arbitrate something without proof and then claim you've got proof at the end because of the invalid arbitration ?
They 'don't understand' ?
Oh I think they do. If you told them pi = 8 they'd understand that too.
That you're Mistaken !
No effort needed except by those who aren't familiar with the value of pi.
May 7, 2022
0
#177
Several examples of the fifty move rule here:
But I can almost guess what's coming ...
No no no ! Its 100 move rule !
In some of them it is invoked. Master level play.
It is you who needs to 'try harder' to understand
But that isn't going to happen.
https://en.wikipedia.org/wiki/Fifty-move_rule#Examples
I'll 'coach' you for a minute though.
Because you went silent.
Make sure you assert now that its 100 moves - not 50.
Not personal attack.
Just an idea.
#240
"Several examples of the fifty move rule here:"
++ No, you are mistaken: I wrote the 50-moves rule is never invoked with > 7 men.
How many men do you count in your examples?
#239
"somebody found a way to justify it in checkers" ++ No, somebody weakly solved Checkers and to do that it turned out he only had to use a tiny fraction of the legal positions.
"chess not checkers" ++ Losing Chess is in some sense closer to chess (same pieces, almost identical rules) and in some sense chess is closer to Checkers (a draw unlike Losing Chess which is a win).
"A 35 digit or so number of positions to be solved."
++ No: 10^44 legal positions, about 10^32 sensible positions, far, far less reachable positions, far, far less relevant positions. 10^35 is way, way too much. Try at least to understand.
I have explained this many times, but apparently not enough yet.
"How many times have people told you that you can't accurately arbitrate something without proof" ++ I always supply proof or at least evidence, facts & figures. People then spew nonsense without any proof at all. An example is your "A 35 digit or so number of positions to be solved" There are only about 10^32 sensible positions, the number of reachable positions is much smaller. The number of relevant positions is much smaller. So it cannot be a 35 digit number.
Another example is I state the 50-moves rule is never invoked with > 7 men.
Then you say I am wrong and I am mistaken, while your examples count how many men?
#239
I try to explain again
This position is a draw and it is in the 7-men endgame table base, but the position is illegal.
This position is a win and it is in the 7-men endgame tablebase, but it is not sensible, so it is counted in the 10^44 but not in the 10^32.
This position is legal, sensible, reachable, relevant
This position is legal and sensible, but not reachable from the previous position
This position is legal, sensible, but not relevant
Mathematics is a lot not about equations, but essentially I agree.
Possibly quantum computing could make a difference.