In the game of chess, perfect or accurate play by both sides leads to a draw. For one player to win, the other must blunder, at least once. Often, winning a game comes down to making fewer mistakes than your opponent.

So, to win, you must provoke your opponent to make a mistake through some sort of tactical play involving a trap or else a sacrifice. Alternatively, to win, you must not only be patient but also hope for your opponent to make a mistake. This is the solution to the game of chess. Thus, chess is solved.

Let me say more.

Letting a chess engine such as Stockfish play against itself, I have discovered that many chess openings and variations can be drawn. It stands to reason, if an engine playing against itself from an opening's start position will draw, that opening is ok to play.

I had thought openings such as the King's Indian Defence and the French Defence would be refuted in this day and age of AI-inspired pawn storms and strategies that trap pieces. It turns out, strong engine play leads to draws.

The Queen's Indian, which apparently cost Magnus Carlsen a game recently, is also drawable.

To my surprise, so too is the very open and structurally unusual Sicilian Najdorf.

Even stuff like the King's Gambit Accepted, the Evan's Gambit and the Danish Gambit, despite being gambits, can be drawn.

Even the Two Knights Fried Liver can be drawn.

Also, the seemingly dubious Dutch Defence and Bird's Opening resulted in draws.

And of course, the likes of Giuoco Piano, Ruy Lopez, and Queen's Gambit Declined would all be draws too. But I did not test those.

The two openings so far that I have found to not result in a draw are:

1. The Grob's Attack (1. g4)

2. The Nakhmanson Gambit, a variation of either the Italian or Scotch Game (1.e4 e5 2.Nf3 Nc6 3.d4 exd4 4.Bc4 Nf6 5.0-0 Nxe4 6.Nc3 dxc3).

I am sure there are others. But it will take a while to figure it all out.

To carry out these experiments, there is no need to use the strongest ELO engine. The only requirement is that the engine you use plays the same way all the time, playing the best move it finds from its evaluation whatever that may be.

If the engine plays both sides as best it can, then only an inferior starting position for one side will result in something other than a draw: only an unsound opening or variation would not be drawable.

Thus, an unsound opening or variation can be regarded as a refuted line, something that can no longer be played unless you would like to play at a disadvantage.

An engine playing the Grob's against itself is interesting because you can see it turning a small advantage into an ever larger one and eventually a win.

So, I say, chess is solved. This discovery of mine is the solution to chess.

What do you think? Discuss.

I think your research and the time and effort u have put into it will not go unrewarded, plz continue the research on 1.b4 the sokolsky opening and 1.a3 the Anderssen opening 1.c3 the sargossa opening and 1.e3 the vantcry opening I am most interested in the findings of your research in offbeat openings 

That's not exactly what solving a game means, but you got spirit.

     Your premise that a chess game can only be won if someone makes a mistake is faulty in the sense that you START with the assumption that chess is a draw with perfect play. Perfect play has not yet been perfected. No human can play as well as a top engine, and the fact that all engines continue to be improved and updated indicates that none of them are perfect today.

      With no way to measure perfect play you cannot determine what the result of a perfectly-played game might be. Starting with the proposition that the end result must be a draw (without definitive proof) solves nothing.

     Until all possible lines can be calculated through all variations to indisputable results the question remains unanswered. Don't hold your breath waiting for unassailable proof.

Ok I just tried Sokolsky (1.b4) and Anderseen (1.a3). Both are draws.

With the Grob (1.g4) I think it has something to do with king safety and white being forced to address that issue, or something like that.

The funny thing is that a weaker chess engine playing against itself gets the same result. It is a completely different set of moves but the same result. So if Stockfish draws an opening, then the weaker engine say Rybka also draws but with different moves. And if Stockfish wins, then Rybka also wins but with different moves. Fascinating discovery, which led me to say what I did about engine strength.

I suppose, if an engine plays consistently at a certain ELO for both sides, even if that play is not "perfect", then only the start position (ie, the opening or variation) matters, because the playing strength of the two sides is no longer a factor in determining the outcome.

Anyway here is Stockfish's Grob match.

You see, slowly but surely, one of the two engines expands its lead until the win.

Now, in regard to chess being solved:

Keep in mind, the games of chess and tic-tac-toe are essentially the same. These two are both two-player, perfect-information, zero-sum games.

Tic-tac-toe has been solved, computationally and mathematically. For tic-tac-toe, a computer can search through the tree of every possible move and find the best one, for absolute perfect play. Furthermore, it is said, mathematicians have also figured out entirely the solution to tic-tac-toe.

In tic-tac-toe, as in chess, for one player to win, the other must blunder, that is to say, the other player must make a sub-optimal move.

Granted, in chess it is possible to blunder a little or a lot. A lead by a margin of one pawn is often not enough to secure a win, or avoid the draw. So, sometimes players can get away with small mistakes.

But even so, in theory, the overall way to win in chess is the same as it is in tic-tac-toe. It is not so much about finding a win, as it is avoiding a loss.

Unless the two engines aren't strong enough to exploit the disadvantageous opening.

Which also means the strongest engines aren't good enough since they might have the same problem without you knowing.

Also you don't give a solution, you give a general approach to playing. You also state the assumption that chess is a draw as a fact.

Fair point. Still, that just means as engines get stronger, openings and variations that cannot now be refuted will be refuted.

Also, Stockfish is nowadays near 4000 ELO. Sites such as chess.com use Stockfish to assess "accuracy", in other words, perfect play.

About draws and chess, I have covered that with the example of tic-tac-toe. The science is called "game theory". In theory, with these sorts of games, perfect play by both sides leads to a draw.

chems :>) 

Do not question the bible prophecy

I think that forum posts or articles with such titles can be safely ignored for quite a while.

Yes, I'm aware of the branch of mathematics called game theory.

CCRL has stockfish at 3500.

Chess.com accuracy (CAPS) is a marketing gimmick.

It's very reasonable to assume chess is a draw, and experimenting with engine vs engine games is fine, but you can't say it proves anything

Burn the heretic!

This isn't actually a part of game theory. There's nothing about perfect information games that requires them to be drawn with perfect play. In fact, it's pretty trivial to design such a game that is won for one side or the other with perfect play, and the winning side must make a mistake in order to allow for a draw. In such a game, the winner wouldn't be the player who didn't make a mistake, since perfect play would lead to a win for, without loss of generality, White.

Tic Tac Toe happens to be a draw, and this has been proven. But without proving it, you couldn't say Tic Tac Toe must be a draw, unless there's some new proof I'm unaware of.

"Chess is probably a draw" is a reasonable opinion that lots of strong players have had for centuries, but it will probably always be an (informed) opinion, barring some crazy scifi improvement in how computers work, above and beyond their normal rate of improvement.

To say you have "solved" chess, you should present the tree of perfect play by both sides that results in a draw. We can do this for Tic Tac Toe. No computer is up to doing this anytime in the foreseeable future.

ummm, what was that middle part? i think i fell asleep.

Just because a particular opening is OK for a computer doesn't mean it's OK for humans.  You might have to play the opening perfectly for 20 moves (and know 100 variations of it) to avoid a loss. The computer can do that easily but humans are likely to make a critical mistake at some point in those 20 moves.

It's a similar thing with endgames. The computer may see that a rook sacrifice leads to mate in 10, but if you the human were to sacrifice your rook in that same position you'd lose because you don't know how to calculate the mate in 10.

I don't think you understand what "solving" chess means.

Chess is the exactly same as tic-tac-toe in game theory.

If the game is two player, zero-sum (when one player wins, the other loses), and perfect information, it always works out the same way. So, if there is a solution for tic-tac-toe, then chess is also solved.

Also, logically, if the game is fair, perfect play on both sides must lead to a draw.

If the game is not fair - not fair as in each player does not have the same opportunity to win - then it is a different sort of game.

An unfair game is like chess with one player playing with a piece or a pawn missing. By the way, I have also tested Stockfish with one side playing without a piece or a pawn. So far, the disadvantaged side always loses.

To disprove what I have said, one must show a game in which a reasonably strong engine such as Stockfish playing against itself, playing an opening accepted by GMs and chess theory, will win rather than draw. ie, Provide an example where perfect or optimal play by white and black results in a win. Impossible, I think.

     Your argument depends upon the perfection of the strong engines. We all realize that stronger engines appear all the time. How do you prove that one of these engines will not find a line that will win against it's own best counterplay? Just because a certain player is unable to defeat themselves does not mean that the line they are playing cannot be beaten. 

Stockfish playing against itself from equal positions is of course going to result in a large percentage of draws. Because it's opponent is literally itself. Equal strength players = high percentage of draws.

I, too, will draw against myself the majority of the time - because no matter how well I play, I can't outthink myself ...

So your solution to chess is to read, but not understand, game theory or what "perfect information' means, then test your hypothesis by using imperfect engine evaluations as data, and summarily declare openings as forced draws by messing around with the engine for 15 minutes.

Got it.