Chess will never be solved, here's why

I do believe that chess we last a long time without being solved, if it ever does one day get solved.

#894
There are two differences between strongly solving and weakly solving.
1) Consider the following position

Weakly solving means all positions with a white pawn on e2, a white pawn on d2, a black pawn on e7, a black pawn on d7, a white light square bishop, two black knights, a white queen and 7 white pawns, a black queen and 7 black pawns are no longer relevant and do not need to be evaluated. This reduction in size correponds to the square root like in checkers.

2) If we can prove that 1 e4 e5 draws, then it is not necessary to establish if 1 e4 c5, 1 e4 e6, 1 e4 c6... also draw or not. That gives another reduction of time by a factor 10.

#896
That is the point: a strong solution needs to look at all 10^37 positions.
A weak solution can bypass positions that cannot be reached or positions that do not need evaluation.
White opens 1 e4. Now all positions with a white pawn on e2 are no longer relevant.
Black defends 1...e5. Now all positions with a black pawn on e7 are no longer relevant.
If 1...e5 draws, then all positions with 1...c5, 1...e6, 1...c6, ... are no longer relevant.

#899
"Then why did you mention that position?"
I tried to explain the mysterious square root reduction be means of an example. In the course of these first moves large numbers of positions have been rendered irrelevant by each pawn move and each capture. The strong solution needs those positions, the weak solution can do without.

"any full solution of chess is virtually infinite"
No, even a strong solution of chess i.e. a 32-men table base is finite: 10^37 positions.
The weak solution requires far less positions: about 10^17 positions. That is feasible.

"The answer is a positional evaluation algorithm"
No, a simple evaluation function just to guide the calculation is superior. You cannot decide the evaluation of a position by some static algorithm, you need to calculate the possibilities.

#902
That is the point: the strong solution is not (yet) practical, but the weak solution is.
Yes, the square root is a conjecture, borrowed from the checkers' proof.
It is not "every line looked at by the engine reduces the number of lines that haven't been looked at by the engine" it is rather each pawn move and each capture make huge numbers of positions not necessary to look at in the future. It is not that obvious: many people here still do not get it.

If a nation state can crack quantum encryption (which was thought to be 100% uncrackable)... Anything is possible. Those who say it can't be solved do not understand that it will be solved one day. All it takes is money to produce the computing power to solve it.

#907
3 cloud engines during 5 years plus human assistants, that is millions, not billions.
Whose money? Google, IBM, Lomonosov University could do it.
Schaeffer did it for checkers. He worked on it for years.

It would be great puzzle of mate in 120 showing the opening starting position then

Yeah that'll be right.

Probably only 27 legal positions in chess all told.

Three microseconds on a cloud cuckoo engine with a cleaning lady. Can't  fault your logic.

Occam's Razor.  Apply it.

Nobody has arrived at that (faulty) understanding except you.  There's no "we" on that front that I can see, but there have been numerous posters pointing out the clear difference between the two.

If only the world were actually the way you envision it ...

But you don't have enough time left for any major epiphanies to take hold, nor to walk any kind of path that would actually bring you to maturity...and that's (truly) regrettable.  Even moreso because you are not unintelligent and probably understand this at some level.

Which world is that?

You mean the world where you have mental powers that you have to hide from the public at large lest you be persecuted, have an IQ that fluctuates between 160 and 190 depending on your circadian rhythms, and routinely change people's lives for the better at first meeting or outright save their lives in restaurants on a regular basis?  The world without pesky Thermodynamics, where the Big Bang is impossible?  The world where you and your progeny are more brilliant and insightful than the entire worldwide scientific community and can speak authoritatively on areas that you've never even worked in?  The world where philosophy drives scientific understanding and data is irrelevant?  The world where incredibly hardy 70-year-olds that walk the hills and moors don't need to be vaccinated?  The world where everyone is jealous of your intellect and there is a complex web of accounts driving a deep rooted conspiracy to unseat you on a chess forum?

That world?

P.S. I do learn, and I have a pretty good memory as it turns out .

Nothing has really changed in this discussion.  People float ideas and they get shot down, and the reality remains:  there's a next to nothing chance of chess being solved in our lifetimes.  It's more likely that God would come down for the Rapture, or space aliens from Andromeda would arrive with their fleet in orbit and just hand us the solution than for humanity to prove it using any currently foreseeable technology advances.

Still, the discussion needs to continue...because it's a bad idea to let vague notions and half-baked ideas run unopposed.  That's how we end up with a majority (though I would not say a vast majority ...) of chess players thinking Kasparov has a 180 IQ...

Not only do I have lifelong interests in chess and computer technology, I have a background and degrees in software development and system design that actually have some bearing .  Yet still, I make less unilateral pronouncements and absolute claims than you do on the subject.

Pretensions are your bailiwick.  Trolling is often in the eye of the beholder, as some who have commiserated about your body of posts over the years have could attest...

Show, don't tell.  The day you actually manage to demonstrate any of your suppositions will be a big step in the right direction.  Try starting with telling everyone that disagrees with you how weak and strong solutions for chess are indistinguishable.  Don't just pronounce it.  Demonstrate it.

Once again, weak and strong solutions are labels that refer to the utility of the results, not to the method used to obtain them.  In fact, once could try to theorize some game where the creation of a weak solution would be harder than creating the strong solution...but once the results are in, the definitions would still hold true:  the strong solution would prove win/lose/draw from any possible position, and the weak solution would only prove win/lose/draw from a single position (the starting one, assumedly, but that is not a requirement ...).

You dismiss these as being indistinguishable based on your assertion that the only way to reach a solution must be algorithmic (in which case, you will have a strong solution to chess, because how could you understand 100% of the game and codify and correctly prioritize all it's principles and then only have a weak solution?).  That assertion is not valid, however.

The discussion for chess has always centered around a weak solution, because there's no progress or even serious speculations about actually solving chess by simply understanding the values of each factor involved in a position.  Let's be clear here...chess engines have only just recently started bootstrapping their own valuations for pieces, positional considerations, etc.  Five years go, engines were calculating based on human opening books, human derived valuations of pieces and things like whether it was justified to move a knight to the rim of the board, etc., and then using tablebases in the endgame.  The tablebases and reaching a forced mate calculation represent the only reliable portion of those calculations (especially true in terms of actually solving chess). 

Everything else is imperfect, and unreliable for a proof, which is why the idea that we can use opening books and Stockfish to just reach out and try to grasp onto a tablebase ending doesn't work.  It's like building a bridge made of concrete and steel on one end and dried mud and reeds on the other end.  We don't see the Stockfish end as mud and reeds, because all we know and can understand ourselves *is* the mud and reeds...like cavemen.

To make the strong solution you are proposing requires that the cavemen understand concrete and steel and physics and engineering.  So, is it better to discuss a solution that can be built with mud and reeds no matter how great an undertaking it is, or better to imagine that we can invent concreate and steel at a time when nobody has demonstrated anything remotely like progress towards such things?

The process you described in your post has already taken place...that's how we got Stockfish .  But that's not a solution/proof, all we have done up until AlphaZero is make a machine that plays chess better than the best humans by virtue of faster calculation, perfect recall, and zero fatigue/loss of focus.  The current leap to engines that derive their own valuations has some potential to move in the direction of uncovering the principles required for a strong solution *if* you keep a pure playing pool, but that's not really being pursued currently.  Leela and other machine learning engines are being placed against opponents that play the same human-derived valuations and principles of chess, and they are learning to beat *that* flawed body of knowledge, and that will hamper/slow down the process significantly.  

But let's say Google decided to let AlphaZero run games against itself and self train for...10 years solid.  What do we get at the end of that?  We get a chess engine that can outplay anything we have seen to a remarkable degree, but that we cannot understand, and the engines as they currently sit are not designed to "articulate" their learnings *at all*.  Nor is it a foregone conclusion that we can even translate the gazillion factors that led to a given position in terms a human player could *ever* follow accurately.  When AlphaZero came out with games that were such a surprise to the chess world, it changed things, and GMs marveled over how A0 could lock down the opponent entirely with a perfectly placed bishop, etc., and those moves were poured over and explained in hindsight...

But those moves were a surprise, and understanding the brilliance of the end result of the bishop lockdown is not remotely good enough.  We would have to understand the countless prioritized decisions that went into building that position from scratch, and A0 cannot explain itself.  We don't understand, nor do we have any realistic path laid out for even starting to do so.

A while back you were in a discussion about various multiverse theories, etc.  The same principle applies here.  If you are talking about building a solution that solves chess like a bunch of Geometry theorems build on top of each other, then you need some hooks to get started...and absent those hooks or building blocks to get started, talking about your solution is like talking about multiverse solutions while we lack the ability to even detect/perceive anything outside our universe.  Thus, the discussions center around brute force technology building a weak solution.

The definitions for solving games are not hypothetical, they are solid and apply to all "games".  The fact that. say, a poster misuses the terms and definitions, or attempts to redefine them, does not change that.  Better to challenge the flawed interpretations.

P.S. Next challenge...prove based on my post above that I don't have the intellect to understand anything in this thread.

Let's be a bit more direct. The correct response by @Optimissed to not understanding the difference between a weak and strong solution is to recognise his understanding is not complete and to try to find out how they differ.

For this purpose, this example may suffice. Let's define a game exactly like chess except that the starting position is that diagrammed below, with white to move: A strong solution of this game entails knowing how to play accurately in every position with the same material on the board (because all such positions can be reached from this one.

A weak solution of this game is a little easier: it is to exhibit a winning strategy for white (left to reader).

Part 2:

Of course how different a weak solution and a strong solution are for chess is less clear than the toy game above. The best that one might hope is the reduction of the complexity by something like a square root, but it is likely in practice a lesser gain. However, it is entirely safe to conclude that there is _some_ saving available in a weak solution.

A rough argument for this is based on the irreversibility of pawn moves. If there is an optimal strategy where the first pawn to be moved is the e-pawn, then white never needs to be concerned about any positions where one of the other 7 pawns has been moved and the e-pawn has not.  Now if black has an optimal strategy that meets this strategy where his c-pawn is moved first, black never needs consider a position where he has moved another pawn before the c-pawn. Combining the two leads to a class of positions that never needs to be considered in either strategy.

The generalisation of this concept leads to my idea of classifying all positions into equivalence classes based on whether they can be reached from each other in a specific direction.  What happened to the classes of positions with another pawn moved before the e-pawn above can happen to vast numbers of equivalence classes of positions that are avoided in the two optimal strategies, thus simplifying the solution by reducing the effective complexity. (On reflection, it is probably easiest just to think of reducing the complexity of each strategy separately, since combining them is a trivial factor of around 2).

You still don't understand...the solutions don't need to proclaim themselves as "weak" or "strong" which you still seem to be using as fuzzy rankings of...something.  They either fit the criteria, or they don't.  

If I followed your logic, then how could anyone do any kind of science experiment with documented results?  I might as well just poke an object with sticks until it magically reveals its inner workings, *then* define the stick poking process that already occurred.

Weak and strong solutions are exactly distinguishable from each other, by their definitions.  Hindsight is not really applicable in this context...but if it were, there would still be a significant and easily documentable difference between the various methodologies that could be used to go after weak and strong solutions.  If you want to discuss methodology, then discuss methodology and stop mixing up your terms.  Stop arguing that the beakers are mislabeled or that two beakers are the same beaker and propose which liquids you're going to pour in them to get a chemical reaction.

I'm not the one who is easily confused by simple analogies or definitions (clearly).  I have no trouble at all visualizing the processes involved.  It's the scope of what's involved that is the problem.  I know how to go after solving chess, and could tackle either type of methodology (brute force or *attempting* to build a prioritized lexicon of chess principles with absolute valuations for every possible position) but if all of humanity combined does not have the resources to bring to bear to achieve either (because proving those principles will still require brute force as well), then it doesn't really matter.

Meanwhile, there's a good deal of "the emperor has no clothes" going on here.  You pretend that you have greater understanding, but can produce nothing but complaints about everyone else's understanding .