Chess will never be solved, here's why

^^ THAT was sufficient, by itself, for me to win the argument and to be seen to win it.

I won this argument in many different ways. Others, including @Elroch, @haiaku and @stancco won their arguments against you. At the very least, you should have made sure you were seen to reconsider. You didn't even go through the motions and you pretend you haven't been trolling all this time. It was either dementia or trolling, that's for sure and I have just established that it wasn't dementia. I'd been extremely reticent about outing you, when it was possible that dementia was the cause, which was why I criticised haiaku. But I take back all that criticism of haiaku. He was right and a day or two faster than I was to be completely sure about you.

I think chess WILL eventually be solved, but not until hundreds of years, engines are simply not strong enough yet.

Presuming you are not being deliberately deceptive, you misunderstand the relevance of the proof tree to the complexity of the proof.  The purpose of the proof tree is to bridge the gap between the opening position and a tablebase. In the case of the solution of checkers, the relevant tablebase was for all positions with 10 pieces or fewer. This tablebase contains about 3.9 trillion positions. As you can see, this is an enormously bigger number than those positions in the tree searched for the proof, never mind the pruned tree (total about 3 million positions) produced with the benefit of perfect hindsight (i.e. knowing the solution).

For chess you would need a much larger tablebase to reduce the proof tree to an analogous degree.  Let's look at the figures for checkers:

5e20 legal positions

1.5e7 proof tree (for 19 forced openings - checkers introduced these because the standard opening position got "boring".)

3.9e12 position tablebase

You can see that the complexity was reduced only by to a power of 0.6 (not as good as square root), just to calculate the tablebase used - tablebases can be generated fast while making good decisions in general positions is extremely slow. 

For basic chess, (10^44)^0.6 = 2.5e26, which is 250 trillion trillion trillion trillion, a complexity way beyond the world's computers put together. Similar log scaling would give a proof tree of 1.8 e15. This might appear not to be out of range. The problem is that it doesn't even exist without first calculating that far out of reach tablebase mentioned above.

For most facts, see Checkers is solved

Thanks again.

@4069

"For basic chess, (10^44)^0.6 = 2.5e26"
++ Chess has more ways to create nonsense positions than Checkers.
This is where the 10^44 legal positions stem from:
https://github.com/tromp/ChessPositionRanking

The 3 diagrams there represent randomly sampled legal positions.
Position 1 has 7 white rooks, so at least 5 underpromoted and 2 black dark square bishops, so at least one underpromoted. The only reason to underpromote to a rook or bishop is to avoid stalemate, i.e. avoid a draw. It is not sensible for both sides to avoid a draw, so at least one side must have made a mistake by underpromoting. So this position is not sensible.
Position 2: 4 white rooks, 2 black dark square bishops, not sensible.
Position 3: 4 white rooks, 3 black rooks, not sensible.

That is why 10^37 is a better figure
https://arxiv.org/pdf/2112.09386.pdf 

This does not include positions with 3 or 4 queens, which do occur.
So multiply 10^37 by 10 to include such positions: 10^38.

Now from a sample of 1000 positions included in the 10^37 observe that the vast majority of these by similar but more complicated arguments can only result from a game with errors too.
Tromp conjectured only 1 in 10^6 can result from a real reasonable game. That leaves 10^32 positions.

Now observe, that each pawn move and each capture render huge numbers of positions unreachable during the solving process. Let us assume we use a dedicated engine to look at 1 e4 only and investigates if 1...e5 draws for black. That engine never has to look at any position with a pawn on e2 or a pawn on e7.

The fact that we only have a 7-men endgame table base and that Checkers used a 10-men endgame table base makes no big difference. From table 3 of the above reference is clear that most chess positions are around 26 men.

Now apply the exponent 0.6 to 10^32 to reflect these numbers of positions unreachable during the solving process.

Now observe, that many legal, sensible positions reachable during the solving process are not relevant. 1 e4 e5 2 Ba6 is not relevant, does not try to win, is a sure loss, checkmate in 82. None of these positions leading from 1 e4 e5 2 Ba6 to checkmate is relevant and the engine does not need to burn time on it.

Also observe, that many positions are clear draws.
The final position of
https://www.iccf.com/game?id=1164259
is a clear draw. If the engine continues to calculate, it will eventually reach a 3-fold repetition draw, but none of the resulting positions is relevant.

Also note that weakly solving only calls for a strategy: 1 strategy. If it is proven that 1 e4 e5 is a draw, then it is not relevant if 1 e4 c5 draws as well or not, so the resulting Dragon, Najdorf, Sveshnikov etc. positions are not relevant.

That leaves 10^17 legal, sensible, reachable during solving, and relevant positions.

Weakly solving Losing Chess even only took 10^9 positions not 10^44 or 10^26.

I don't think the preceding post even remotely answers the criticism made by Elroch, which supports what I tried but failed to state clearly.

Of course, <<<<Also note that weakly solving only calls for a strategy: 1 strategy. If it is proven that 1 e4 e5 is a draw, then it is not relevant if 1 e4 c5 draws as well or not, so the resulting Dragon, Najdorf, Sveshnikov etc. positions are not relevant.>>>>

has completely lost me. It would very much depend on which side was trying to achieve a draw. If white is playing for a draw, it still may not influence white's opening move, since white may find a draw easier to obtain after 1. d4, even if 1. d4 were to win theoretically for black. If black wants a win. then black would play the Sicilian. It's such comments which made me suspect dementia.

Perhaps to make it clearer, this thread doesn't specify any particular type of solution and therefore we can assume a full, relevant solution, which means ALL relevant lines.

Secondly, the supposed definitions that SOME people are trying to use, even if they make sense in some circumstances, have been written by a game theorist and as such they are not properly relevant to chess. The very fact that they call for a strategy gives that away, since the only possible strategy available in chess is to make moves that are as strong as possible. There is no other, possible strategy in chess.

My previous comments to the effect that these "definitions" are inapplicable to solving chess therefore hold absolutely.

You say "comment critically" on this post, of all things?

It consists of Svesnikov's empty boast. How else is one to interpret it?

There's nothing there to factually criticise, except an unsupported and unsupportable claim.

This still makes absolutely no sense. Even if we assume the only problem with solving chess is money, and as soon as the money is available, it will take 5 years, that still makes the 5 year claim a steaming pile of road apples. 

So to answer your question, it takes about 9 months to build a house. Of course it could take a lot longer or a lot less, but that's about average. And you start building when the permits are secured, which is a lot more important than the money. The money part can change, drastically, with the economy and price fluctuations. So you don't have to have  the money to start, but you do have to have the money to finish. 

So you are STILL saying the 5 year period for solving chess starts when there is the money. Well, when is that? It could very well be no amount of money will matter, but even if it did, what if there isn't enough money for 300 years? I don't see the point in saying chess can be solved in 5 years if nobody has any idea when the 5 year period starts. 

Money? Don't make me laugh.

It takes an angel only, one who doesn't side satan's side, of course. That one lies.

You are right. If 1.e4 e5 is a draw it says nothing about whether either side has a winning strategy. For example, it does not prove that 1. e4 c5 doesn't win for black. Likewise, it does not prove that 1. c4 does not win for white.

If you are confident that chess is a draw and focus only on proving that this is so, and you have proven that 1. e4 e5 is a draw, you still need to prove that white can draw against 1. e4 c5, and that black can draw against 1. c4 (just as with the more open-minded viewpoint).

I'm afraid I couldn't see any choice, other than incompetence or some very upsetting type of mental freezing, caused by illness of some kind, where the salient points are clung onto, despite memory loss and increasing lack of understanding. Or, perhaps, trolling, which would be very sad, considering the effort that's gone into this and similar threads. Or a strange combination of these.

Just why? I don't believe the incompetence explanation. An odd mix of the other two, perhaps. Why did we push this so hard? I suppose it's to uphold personal freedoms in a situation where honesty and best efforts were disregarded or worse.

@4078

"it does not prove that 1. e4 c5 doesn't win for black."
++ If 1 e4 c5 is a black win, then 1 c4 is likely to be a white win.
Moreover if 1 e4 c5 is a black win, then by strategy stealing 1 c3 e5 2 c4 is a white win.

"it does not prove that 1. c4 does not win for white."
++ 'From the outset two moves, 1.e4 or 1.d4, open up lines for the Queen and a Bishop. Therefore, theoretically one of these two moves must be the best, as no other first move accomplishes so much.' - Capablanca
This was corroborated independently by AlphaZero, with no other input but the Laws of Chess
Figure 31 https://arxiv.org/pdf/2111.09259.pdf 

To be sure 4 moves need investigation: 1 e4, 1 d4, 1 c4, 1 Nf3.

The possibilities that chess is a white win or chess is a black win go against expert opinions and empirical evidence from ICCF, TCEC, AlphaZero, human GM games.

There is also a deductive argument. To win a game of chess a direct assault on a king is only possible if the opponent neglects its defence. To win a game of chess involves queening a pawn. 'Any material advantage is enough to win' - Capablanca. In the initial position material is equal, but white is a tempo up. 1 pawn = 3 tempi. You cannot queen a tempo.

Suppose that a tempo were enough to win a pawn. Then it would be advantageous to play a gambit where white sacrifices a pawn to win 2 tempi, then convert the 2 tempi to 2 pawns, and then queen the extra pawn.

@4076

"So you are STILL saying the 5 year period for solving chess starts when there is the money.
Well, when is that?"
++ I do not know. Chess can be solved in 5 years. If chess will be solved depends on somebody spending 3 millions to hire 3 grandmasters and rent 3 cloud engines for 5 years.

In 1961 President John F. Kennedy persuaded congress to put up money to set men on the Moon. In 1969 Armstrong walked on the Moon.
It took 8 years to put men on the Moon, but it depended on the money.

The discussion is over, at least from my point of view. I hope that the others do not engage you on this topic; but of course, it's their choice.

Conclusion: chess can be weakly solved in 5 years,
but when or if depends on money to hire the grandmasters and rent the computers.

The problem is the lack of people willing to spend $3 million to ask the question "is chess a draw?" and to get the answer 5 years later "we are now slightly more confident that it is but, in all frankness, didn't have a snowball's chance in hell of proving it. Thanks for the cash."

I feel there are slightly more attractive investments.

There is no such conclusion.