The Problem With Opening Prep
For a brief period in the mid 1990s, I was determined to play the Old Indian Defense. I’d acquired a copy of David Bronstein’s classic, Zurich 1953, and had memorized a certain legendary game; now I just needed to play the Old Indian enough that someone would eventually repeat the first thirty moves of said game so that I too could rain hellfire down upon an unsuspecting opponent:
A childish dream, I admit, but one that captures a big problem with the memorization of deep opening variations: it’s a kind of safety blanket that often doesn’t provide much safety. The longer the memorized line, the less likely that exact position is to show up on the board (and the less likely you are to remember it correctly), which paradoxically means that opening preparation that focuses on high degrees of specific memorization is a much less valuable investment of time than almost any other kind of chess preparation and learning1.
The second problem with memorization of openings is that it is a fundamentally passive activity, the chess equivalent of memorizing dates and empires to prepare for an AP history exam. Rattling off a bunch of memorized moves at the board can feel good (there’s the safety blanket analogy again), but it creates a habit of playing chess without actually thinking about chess that can lead to trouble.
Here’s an example from one of my own games:
Some context about how we got here: I woke up on the final day of the tournament to find that I had the white pieces against Grandmaster Melik Khachiyan. As a late in life arrival to the king-pawn openings, I realized that I knew nothing about the Schliemann Defense to the Ruy Lopez. I did, however, know about this game that Melik had lost to Robert Hess four years prior and decided to spring an elaborate opening “trap.”
My idea was to repeat the Hess-Khachiyan game up to move 15 as a way of threatening to force a draw right out of the opening. If Melik deviated early with something dubious, I would get an opening advantage against a GM without having had to find a single good move. If he didn’t blink, well, it would be an easy draw2. When we got to the position in the diagram above, I passed on Hess’s winning attempt (16 Nd2) and instead played 16 gxf3, forcing black to make a draw: 16… Qh4 17 h3! (take on g4 and white gets mated on g2) 17… Qxh3 18 fxg4 Qxg4+ 19 Kh2, and draw agreed. Sam Shankland, sitting at the board next to me, followed me out into the hallway and made a convincing chicken dance; Melik expressed regret that he picked a sharp line that allowed me to dodge him; I gained a couple rating points and learned nothing.
The irony is that the position after 16 gxf3 isn’t drawn at all: after 16… Qh4 17 h3! black can play 17… Nh2! 18 Kxh2 Rf8!
The difference between this variation and the one played in the game is that white’s pawn is still on f3, preventing successful defense (such as Ra3) along the third rank. Despite being up two pieces, white is getting mated or losing a tremendous amount of material thanks to the threat of Rf6-g6.
If I had been thinking about playing chess (rather than about the result of the game) at any point, I might have been suspicious of playing a variation that leaves my king hanging out to dry. Given that I thought I was forcing a draw, I could have invested 30 or 40 minutes on move 16 to double-check my pre-game analysis. Instead I blitzed everything out, a confident and careless believer in my own prep over my own brain3.
A Possible Solution
Instead of memorizing a bunch of moves, I think it makes more sense to look at the opening from 30,000 feet to get a sense of what kinds of variations are possible. After making a list of variations, sort them into the following three buckets:
The most dangerous variations, lines where a single mistake is often enough to lose the game, even against a weaker opponent.
Frequently played variations that aren’t highly tactical, where understanding of typical positional ideas and common transpositions is sufficient to achieve good results.
Unusual variations, odds and ends that you don’t expect to see on a regular basis but are still important to know something about.
The distribution of variations in each bucket is going to vary by opening4, but the general idea is to put a limit on how much actually needs to be memorized so that we can spend more time looking at illustrative games that show the variety of middlegame plans available.
Here’s an example of how this could work (minus the exemplar games, which I don’t have time to find in a variation that I’m not planning on playing in any upcoming events), using the Winawer French—an opening I’ve played the black side of about 70 times in serious games—1 e4 e6 2 d4 d5 3 Nc3 Bb4.
Bucket #1: The Dangerous Lines
There’s one variation of the Winawer that is both popular and difficult enough to demand a lot of memorization: the poison pawn complex that occurs after 1 e4 e6 2 d4 d5 3 Nc3 Bb4 4 e5 c5 5 a3 Bxc3+ 6 bxc3 Ne7 7 Qg4 Qc7 8 Qxg7 Rg8 9 Qxh7 cxd4 10 Ne2 Nbc6 11 f4:
From this position the old main line was 11… Bd7 12 Qd3 dxc3, but black was struggling here, particularly against 13 Qxc3. The new main line runs 11… dxc3 12 Qd3 d4 13 Nxd4 Nxd4 14 Qxd4 Bd7, as played almost to perfection in Leko-Caruana, 2012. It’s little details like this (as well as the fact that black needs to know what to do against white’s alternative lines 8 Bd3 and 10 Kd1) that justify memorizing the key variations quite deeply, often to move 15 or 20.
I think it’s wise to have a simple alternative to the most dangerous lines in your repertoire, something you can turn to if you’re struggling in the main lines (or just haven’t reviewed them lately). My way of avoiding the poison pawn variations has been 1 e4 e6 2 d4 d5 3 Nc3 Bb4 4 e5 c5 5 a3 Bxc3+ 6 bxc3 Qa5, which has the benefit of being much easier to learn than alternatives like 1 e4 e6 2 d4 d5 3 Nc3 Bb4 4 e5 c5 5 a3 Bxc3+ 6 bxc3 Ne7 7 Qg4 00; the drawback is that it also avoids black’s good responses to the position lines of bucket two.
Bucket #2: Common, but Less Dangerous
Into this bucket goes everything else that white might play on move 7:
1 e4 e6 2 d4 d5 3 Nc3 Bb4 4 e5 c5 5 a3 Bxc3+ 6 bxc3 Ne7 7 a4 Nbc6 8 Nf3 Qa5 9 Bd2 (white gives up on the dream of Ba3, instead planning to play c4, opening the center for the bishops)
1 e4 e6 2 d4 d5 3 Nc3 Bb4 4 e5 c5 5 a3 Bxc3+ 6 bxc3 Ne7 7 a4 Nbc6 8 Nf3 Qa5 9 Qd2 (similar to the line above, but white goes to great lengths to develop his bishop on a3)
1 e4 e6 2 d4 d5 3 Nc3 Bb4 4 e5 c5 5 a3 Bxc3+ 6 bxc3 Ne7 7 h4 (looking to push all the way to h6, weakening the dark squares)
1 e4 e6 2 d4 d5 3 Nc3 Bb4 4 e5 c5 5 a3 Bxc3+ 6 bxc3 Ne7 7 Nf3 Bd7 (raising the question of whether or not white will transpose back to the first two lines with 8 a4 or allow the Ba4 blockading idea)
The questions black has to answer in these variations are more conceptual than tactical, and include:
When is the right time to break in the center with f6?
In what kinds of positions does it make sense to lock up the center by playing c4?
When is it better to castle kingside and when is it better to castle queenside?
Is it ever good to capture on d4 or, when possible, a4?
What is to be done about the passive French bishop?
Reviewing and annotating the games of the great masters of the French Defense, such as Uhlmann and Korchnoi, is a good way to start answering these questions.
Bucket #3: Odds and Ends
Here we put the fourth and fifth move deviations that come up less frequently. These are the trickiest to figure out how to deal with. Sometimes memorizing a sharp refutation seems useful, but if you only face each line once every ten years or so, how much will you remember? I’ve stuck a key idea alongside each variation that I would use to help guide me if I faced it over the board:
1 e4 e6 2 d4 d5 3 Nc3 Bb4 4 Ne2 (white temporarily gives up e4 to keep his pawns straight; black should try to fight back actively in the center, but this line isn’t too dangerous)
1 e4 e6 2 d4 d5 3 Nc3 Bb4 4 exd5 (should be a simple transposition back to the exchange variation)
1 e4 e6 2 d4 d5 3 Nc3 Bb4 4 a3 Bxc3+ 5 bxc3 dxe4 6 Qg4 (rare, but full of non-intuitive stuff, such as when white plays Bb2, 000, and c4 to expand in the center; tactical alertness is required)
1 e4 e6 2 d4 d5 3 Nc3 Bb4 4 e5 c5 5 Qg4 (a sharp variation where it helps to know one important tactical idea: 5… Ne7 6 Qxg7 Rg8 7 Qxh7 cxd4 8 a3 Qa5 9 Rb1 dxc3 10 axb4 Qa2, trapping the white rook)
1 e4 e6 2 d4 d5 3 Nc3 Bb4 4 e5 c5 5 Bd2 (another positional approach, hoping to build a big center while playing Nb5-d6; black needs to respond by actively fighting against white’s center pawns)
Benefits of the Conceptual Approach
In 2000 I played in a simul against British GM Daniel King and trotted out my trusty Winawer, only to be confronted by a strange idea that I’d never seen before: 1 e4 e6 2 d4 d5 3 Nc3 Bb4 4 e5 c5 5 a3 Bxc3+ 6 bxc3 Ne7 (so far, so good) 7 Ne2 (that’s weird, I guess I’ll just play the normal Winawer moves) 7… Nbc6 8 Nf4 Qa5 9 Bd2 Bd7 10 Nh5:
On the one hand, white’s Ne2-f4-h5 idea felt a little threatening: I really didn’t want to castle kingside into a potential attack. On the other hand, it felt a little stupid, since white has prioritized this knight maneuver over development of the kingside, castling, protecting the center, etc. The problem was that the move I wanted to play, 10… 000 11 Nxg7 felt like a significant weakening of the dark squares. I came upon a solution to this problem by using poison pawn lines as inspiration: 11… h5! 12 Nxh5 (there’s no other way out) 12… cxd4 13 cxd4 Qa4:
Despite his material gains, white is suffering. He can’t defend d4 with 14 c3 because of 14… Qxd1 15 Qxd1 Rxh5; black’s pieces find strong squares in the center as they take aim at the weak white king. It’s too hard to defend this sort of position in a simul, and I won in under 30 moves.
I could never have played this way without a deep conceptual understanding of the Winawer: which ideas work in which kinds of positions, where the pieces should go, the sometimes hidden vulnerabilities of the white position. My goal is to have this develop the same kind of understanding about the openings I’ll be playing in my next tournament by prioritizing the study of games rather than pure memorization.
This is of course not as true for the very top GMs, but I’m a little suspicious of how much energy they put in to opening prep compared to other aspects of the game.
The extra rest actually came in handy, as I had fight to hold a bad endgame against Walter Browne in the final round, the last game in the tournament to finish if I’m remembering correctly.
I suppose I shouldn't be too hard on myself; the same thing famously happened to Kramnik and almost cost him his match against Peter Leko for the world championship.
The Najdorf Sicilian is going to have a bunch of lines in bucket one; the Exchange French lives almost exclusively in bucket two.