Recently I have been fascinated by the concept of patterns in chess. At his core, a good chess player is essentially a super-efficient pattern recognition machine. He can cast his gaze upon an unknown position, identify the salient features, recall the proper theme, and apply it seamlessly to the new parameters - all in a matter of seconds. When analyzing with strong Grandmasters, one is often impressed by how easily they relate moves and positions back to stripped-down, well-studied concepts. Comments like "yes, this rook endgame has been known to be absolutely lost since Alekhine - Capablanca, 1927" or "I can't accept the Tarrasch pawn structure; Karpov already exposed the deficencies of it in his first match with Kasparov" are commonplace, short-hand ways to convey whatever principle is at issue. A Grandmaster can recall and apply an appropriate theme to a position as easily as a skilled stand-up comic can craft a joke from an everyday situation.
So, how does an amateur go about acquiring such a mental Rolodex of useful patterns? With hard work, of course 
Personally, I cannot yet claim to possess the knowledge base of a Grandmaster. But every once in awhile, something clicks...
Consider the following example, which I have used as a training exercise for many of my own students. Feel free to solve it yourself. If you want a serious challenge, I recommend not moving the pieces.
Here Black decided to play 1...Nd5. How should the game proceed?
SOLUTION BELOW
The game saw the crucial line 1...Nd5 2.Rxe8+! (stronger than 2.Nxd5, which is met by 2...Rxe4) 2...Rxe8 3.Rxe8+ Qxe8 4.Nxd5 Qe1+ (the zwischenzug upon which Black was relying; 4...cxd5 5.Qxd5 would be a depressing queen ending) 4.Kg2 Qe4+ 5.Qf3! Qxd5 (5...Qxf3+?! 6.Kxf3 is the game minus a tempo for Black; 5...cxd5 is a legitimate try, though I believe Black should lose after 6.Qxe4 dxe4 7.Kg3 g5 8.hxg5 hxg5 9.f3!) 6.Qxd5 cxd5 (diagram)
This is just a drawn pawn ending, right?
Many students can reach this position from the initial diagram. However, most EVERYONE waffles on the evaluation of the resulting endgame. Their thought process usually trails off at this moment ("...ehrm, it looks pretty symmetrical here...maybe White is a little better, but I'm not sure if it's a win"), and they stop calculating.
As an aside, I was rather impressed when an attendee at last summer's Western Invitational Camp - rated 2050 USCF and calculating from the first diagram - kept right on going: 7.Kf3 (White intends to attack the d5 pawn) f6 (Black hurries to prevent Kf4-e5) 8.h5! (diagram)
Essential. White aims to fix Black's kingside pawns and secure the f5 square for his own king.
The student continued: 8...Kf7 9.Kf4 Ke6 10.g4! (diagram)
Black's kingside is frozen.
Now a brief "reserve tempo" skirmish breaks out: 10...a6 11.a4 a5 12.b3 b6 13.f3 (zugzwang; Black's king must give ground) 13...Ke7 14.Kf5 Kf7 15.f4 (diagram)
Zugzwang again!
Roundabout here (nearly 30 ply out from the initial diagram!) the student correctly proclaimed that Black was lost. Bravo!
Black actually played 15...g6+ and soon resigned, but a more "textbook" finale would be 15...Ke7 16.Kg6 Kf8 17.g5 (a decisive breakthrough) 17...fxg5 18.fxg5 hxg5 19.Kxg5 (diagram; analysis)
Black cannot simultaneously defend both d5 and g7. Barring the possibility of the queen ending I mentioned with 4...cxd5 Qxd5, the conclusion is that Black loses by force after 1...Nd5. Here is the game in full (to push yourself even further, try to work out why Black resigned in the final position).
Cool game, no? I think it's an excellent illustration of multiple themes (transitioning from a middlegame to an endgame, zwischenzugs, king activity, reserve tempos, "fixing" an opposing pawn structure, etc.).
About a week ago, I was leafing through GM Vassilios Kotronias's book,
The Grandmaster Battle Manual. In Chapter 5 ("Facing Lower-rated Opponents), the author presents the game Salvador - Epishin, Bratto 2008, in which the players reached the following position:
Hey, this looks kinda familiar...
The game went 30.Qd4+?? Qxd4 31.cxd4 and I didn't need to look at Kotronias's notes to determine that 31...Kf6! -+ is absolutely winning for Black. The game continued 32.f3 h5 33.Kf2 Kg5 34.Kg3 h4+ (observe the importance of this pawn once again) 35.Kf2 Kf4 (zugzwang) 36.Ke2 Kg3 37.Kf1 g5 38.Kg1 f5 and White resigned. He is helpless to prevent the breakthrough ...g5-g4, leading to the eventual win of either the "d" or "g" pawn. (diagram)
Compare to Salov - Short (analysis)
It's a little tempting to end this post with something self-congratulatory like "everybody look how awesome I am - I found and correctly applied a pattern!" To be honest though, I know this small sub-type of pawn ending doesn't remotely translate to more ELO points, nor does it give me any street cred among strong players. However, discovering and applying a cool pattern like this is EXACTLY what keeps me playing, teaching, and learning chess. It's so extremely satisfying to know that we - chess players who strive to improve - have the capability to make these kind of connections if we put forth the effort.
It's a beautiful thing.
Personally, I cannot yet claim to possess the knowledge base of a Grandmaster. But every once in awhile, something clicks...