Study the old masters

In the old days there were no opening books no opening theory. Then some individuals developed the first steps towards what would become modern chess theory. An individual, when he first starts playing chess knows no theory. Perhaps someone tells him that 1.P-K4 is a good first move. Then the individual assimilates more and more knowledge as well as personal experience. Perhaps this individual would realize that it is desirable to have a centralized knight or that a bishop pair can be most helpful if combined with a passed pawn and an open position. We see in the old masters the same issues that come up in our own games because our experience is not developed enough to the stage where games(say later than 1970) are relevant or comprehensible to us.

So I say, first study the old masters. Nimzovich, Rubinstein, Capablanca etc. Join in their struggles and you will surely become a better player.

EDIT: I was going through the games of the old masters(from the excellent book by Chernev) when I realized that they were struggling with the same issues as I was. This is in stark contrast to studying a game between Kasparov and Karpov in some obscure Grunfeld line and why the assessment after 30 moves is +=

I would say go back even farther than that, and study Morphy, Steinitz, Andersen, and players from that era.

I was looking at Rubinstein's famous games against Duras and Rotlevi in Chernev's "12 great players" book this weekend.

He played like Stockfish almost the whole time - and then did some totally sound very deep Queen sacrifices with lots of branches and subvariations.

If Rubinstein played at that level more consistently he could have been better than Capablanca.

Rubinstein is considered a very strong player of that era. It's unfortunate that he went insane and never played beyond a point.

And yeah I second the thought about studying Morphy, Steinitz and Andersen. All great players.

The first thing the old masters should teach you is that there isn't one way to be great at chess.  All of the past masters yield key concepts which can help to improve your play.

The "romantic" players like Anderssen demonstrate just how powerful tactical play can be.  Steinitz and Tarrasch show how a structured, positional approach can often close off those tactics until it's too late in the game to spring them.  Nimzowitsch teaches that for every rule of thumb there are critical exceptions.  Reti and Marshall teach us how being unafraid of losing can yield the most amazing of triumphs.  Lasker teaches that you can be great at the game pretty much indefinitely, no matter how much the game evolves during the span.  Alekhine taught us that the seemingly contradictory concepts of classical and hypermodern approaches can be brought together to make something even stronger.

Of course, the most immortal of players (among whom I'd count Morphy, Lasker, Capablanca, Alekhine, Petrosian, Fischer, and Kasparov) represent a level of understanding which can continue to teach any player more and more as they evolve and improve.  Virtually no one is prepared to learn absolutely everything these types of players can teach us.

"... there are major advantages to studying older games rather than those of today. The ideas expressed in a Rubinstein or Capablanca game are generally easier to understand. They are usually carried out to their logical end, often in a memorable way, ... In today's chess, the defense is much better. That may sound good. But it means that the defender's counterplay will muddy the waters and dilute the instructional value of the game. For this reason the games of Rubinstein, Capablanca, Morphy, Siegbert Tarrasch, Harry Pillsbury and Paul Keres are strongly recommended - as well as those of more recent players who have a somewhat classical style, like Fischer, Karpov, Viswanathan Anand and Michael Adams. ..." - GM Andrew Soltis (2010)

So true! Its easier to understand the games from past masters and each of past master has something to teach us; I use a lot ideas from Nimzovitch in my game like, create a weakness in your opponent position, double your opponent pawns and fix them so your opponent can't undo them. A lesson I learn from Morphy is how to open lines to attack my opponent, from Capablanca sacrifice pawns to keep your piece active in the endgame, Capablanca had simple rule, make sure all your pieces are working together, one bad piece can ruin a your position. And many lesson from these great past masters, Rubinstein was really incredible and had great inside in this game, he created many gems in attacks, positional and endgame. Pillsbury did a lot for the Queen pawn opening, he made it into a weapon, before him and Steinitz it was only 1.e4 games. Paulsen was a great opening innovator, he and his brother did a lot for the Sicilian defense, Tarrasch and Nimzovitch contribute so much in the French defense, we so indebted to these great past masters.  

Today GMs are just copycats, the way the Ruy Lopez and Italian were played by Steinitz, Tarrasch, Capablanca and Teichmann:  Teichmann vs. Rubinstein Carlsen 1911 1.e4 e5 2.Nf3 Nc6 3.Bb5 a6 4.Ba4 Nf6 5.0-0 Be7 6. Re1 b5 7.Bb3 d6 8.c3 0-0 9.d3! Na5 10.Bc2 c5 11.Nbd2 Nc6 ( black: Schlechter 11...Qc7 12.Nf1 Nc6) and both games Teinchmann crush these two great masters., Steinitz vs. Chigorin 1892 1.e4 e5 2.Nf3 Nc6 3.Bb5 Nf6 4.d3! d6 5.c3 g6 6.Nbd2 Bg7, if you haven't seem this game, it is a must in how to attack the fianchetto position. There are many concepts that amateurs never learned because they don't give the time to play over these great past masters games.

What is particularly worth studying about ANY of the masters' games is what kind of middlegame they tried to get to. The purpose of the opening is to reach a playable middlegame but, for most of us, we study the opening but not the kinds of middlegames they usually lead to!

In many cases, it's worth studying amateur games if you want to see how to exploit weaknesses that are seldom found in masters' games.  Dan Heisman's The World's Most Instructive Amateur Game Book is excellent because it illustrates the difference between not thinking clearly about positions and using chess principles properly.  For example, "13...e5! 14 dxe5 46:00 Slowly played, but perhaps White could have decided more quickly if he had known a general principle that comes in handy in this situation: if your opponent breaks with a center (d- or e-) pawn and, if you capture with a pawn and he can't recapture with a pawn, then capturing is usually correct."

Note that opening, middlegame, and endgame books were available from the mid-1800's on and some books that are still very important today, like Emmanuel Lasker's Common Sense in Chess (1895) and Nimzowitsch's My System (1925) are over or close to 100 years old.

Today, the Internet has made chess information much easier to get than previously, but even in the 1990's when I coached a high school chess team, certain chapters from My System and from Keres and Kotov's excellent The Art of the Middlegame (first published in English in 1964 but older) were required study material.