if we "solve chess," will anyone play it anymore?

The number of possible chess games exceeds the number of all the know stars in all the known galaxies....fascinating how just a board game with 16 pieces each on 64 squares can do that!

The number of possible 30 full move sequences equals the number of atoms in the Hubble Bubble. The number of 40 sequences is 1000^10 times that many. There is absolutely no way to store all that on a computer for centuries to come, or evaluate all the positions.

What if instead of playing the board, the super duper computer plays with the knowledge of how far Stockfish can see, and plays agressive, risky moves accordingly? Think it could give Stockfish queen odds and win?

There are usually 20-50 possible moves per move. Count them. 31^2 = 1000. So 1000 possible moves by Black and White. 1000^number of whole moves. Just 3 whole moves deep is a billion, which is more than Deep Blue or Stockfish look at, 200 million. They search a move or two deep, then narrow it down, then search those, then narrow again. Most moves are missed.

Give your computer White, and watch it fail to see what we can see. What else does it miss?

never.. will that happen

White must sacrifice a queen to break through the wall. That is a 9 point loss, and no immediate improvement in position or material. When the computer evaluates all possible moves, it looks 2 moves deep at most, then chooses which 6 of the initial moves lead forceably to the position with the highest point total. Points are awarded for position not just material, but does the computer assign more than 9 points to breaking the wall? If not, then simply walking the queens around inside the wall is where the points are at. It does not yet know how to look across. These principles of breaking walls and sacrificing when you are up lots of material have not yet been coded into many programs. Until then, we must wonder what else they miss, and how an all knowing player could take advantage if it knew the computer's weaknesses.

We can only memorize a tiny fraction of the whole game tree. So supposedly yes. 

I doubt Stockfish suddenly thought Qxc4 was a better move after you played it. From the choices it was forced to choose from, capturing pawns and invading seemed the best option. It likely followed those narrowed best options out a ways to see checkmate. The depth the evaluation surpassed the initial extra queen was likely too deep tissue reach while evaluating all possible moves. 2.5 moves is already too deep without narrowing the choices. So I wonder what else it misses.

I doubt Stockfish suddenly thought Qxc4 was a better move after you played it. From the choices it was forced to choose from, capturing pawns and invading seemed the best option. It likely followed those narrowed best options out a ways to see checkmate. The depth the evaluation surpassed the initial extra queen was likely too deep tissue reach while evaluating all possible moves. 2.5 moves is already too deep without narrowing the choices. So I wonder what else it misses.

Another good test would be to clear the cache and let it think about that position overnight. I doubt through outcome would change. Chess players, especially grandmasters, should forward to chess software developers positions they think the computer evaluated wrong, and state the governing principles. Programs like Stockfish would then rapidly improve in strength.

Even if the modern gambits, Hungarians, and so on become "solved," people will find new variations of openings/strategy/tactics to get ahead. Chess is ever evolving, what was technically wrong yesterday are the moves to make today and moves that are technically wrong today will be the moves to make tomorrow. 

Troitsky solved the White to win KNNKP bit of chess in the 1940s but you haven't seen many resignations by Black in the meantime when it comes up.

Personally i don't understand why so many people seem so worried about this. Even if chess one day will be "solved" it will be still challenging and worth playing for humans. Too many variations to memorize. Perfect play is something that doesn't concern us.

Relative to class C players, grandmasters have solved chess, and have written opening books. Yet class C players still enjoy playing chess and exploring non book moves and getting crushed by stronger players.

I think casuals will still play it, but it won't be taken as seriously anymore at the higher levels *assuming* that GMs can actually implement the correct series of moves. So if GMs like Nakamura or Napomniatchi can implement perfect play, then I think chess will die out as a competitive sport.

However, if chess is solved but can only be played perfectly by a computer from all positions (presumably because no human can memorize all the correct plays), then it won't really affect anything. Such an outcome would be indistinguishable from the current state of chess, which theoretically does have ideal moves, its just no one knows what they are. We already use stockfish as a pseudo-objective guide as to what moves are ideal, so the only difference is we would have a perfect chess engine to compare our mortal brains to.

It would definitely change chess strategies though, as Anand has pointed out with respect to AlphaZero.

I like what Garry Kasparov said about it.  Just because we now have cars doesn't mean people are going to stop jogging.  

John, "solved" means we know the result of perfect play.  From the opening position, that could be a draw, a win for white, or (unlikely!?) a win for black.  We actually already "solved" chess for endgames positions up to 7 pieces.

In other words, if the computer opens with 1. d4 and prints "mate in 72", chess is solved.