The value of looking ahead

The tree functions of ChessBase and Fritz are really valuable tools for a lot of reasons. You can use opening trees as replacement books in Fritz so that you can practice an opening by forcing Fritz to play it. You can look at the statistical values in the tree to see how various moves have fared in actual games. I've used the trees extensively to find effective opening traps in my correspondence games; in fact, about 20% to 25% of my correspondence games end with my winning in under twenty moves.

But blindly "following the numbers" can backfire on you, as I was recently reminded during a conversation with a chessplaying friend. We were discussing statistical game trees and he mentioned a game in which the numbers led him astray. I was reminded of one of my own correspondence games in which this very same thing happened.

Too many correspondence players use opening trees as a crutch, letting the numbers do the thinking for them instead of doing (at least some of) the thinking for themselves. This can have unexpected and unfortunate repercussions when a "bust" to a line appears in another move or two. You're thinking you have a good position; after all, the tree says that it has a 65.7% success rate for the color you're playing, so how can it be bad? Then, suddenly, that "good" position turns around and bites you in the backside and you find yourself in a deep hole.

Let's have a look at my own example from a USCF Rated correspondence game I played in 1998. I had the Black pieces. My opponent was probably expecting a King's Indian Defense:

1.d4 Nf6 2.c4

...but being the irrepressible gambiteer that I am, I threw him a curve:

2...e5

...sending us into a Budapest Gambit, in which my opponent obliged my whim and accepted the pawn.

3.dxe5 Ng4 4.Nf3 Bc5 5.e3 Nc6 6.Nc3

So far, so good, right? It's not a "standard" Budapest, but it's been played a few hundred times and the numbers show that I should be in decent shape:

I mean, I'm OK here, aren't I? White's still got a wee edge (55% according to the stats for 6.Nc3) but I've got play and none of the top three moves on this list look bad. Obviously, I'd love to see something with more than a 50% success rate, but I have nothing really to complain about. So I did the natural thing: I more or less automatically played the top move on the list:

6...Ngxe5 7.Nxe5

And, although I didn't know it at the time, right there is where my game pretty much tanked. After White's seventh move, I have only one option:

7...Nxe5

This is still looking OK, eh? I have one option, with a 46% success rate, so all is apparently well. Let's have a look at what's been played after 7...Nxe5:

I made a dangerous assumption here: I figured that my opponent was going to play the "usual" 8.Be2. But you know what they say -- he who underestimates his opponent loses. So guess what my opponent played?

8.f4

And I got that sinking feeling in my gut. What am I supposed to do with that? It's not an instant loss -- I can recover from it. But it's not what I wanted. It's very much like being a schoolkid who gets a bad wedgie from another kid in gym class: it's not fatal, but it's not what you want.

There have been two moves played here in tournament practice and, if you look at the player and performance ratings, better men than I have been unable to extricate themselves from this particular pit. I did what any reasonable person would do in this situation: I panicked, threw out the opening book, and started doing a bit of thinking for myself. I just plain hated the look of the Knight on g6 (in fact, I'd played a game not long before in which a similar Knight got caught in a Kingside pawnstorm), so I ditched that idea and retreated the pony to c6.

8...Nc6

Now what's the lesson here? Let's go back and look at the tree after 6.Nc3:

I won't hammer you with endless screen shots of tree displays, but had I looked ahead, I would have seen the possibility of digging myself into an identical hole by playing either of the Knight captures. The move 6...Ncxe5 will likely transpose straight back into the game's main line (as shown above), so it would have been much safer for me to play 6...0-0 instead.

Nothing can do the work for you here -- you have to do it yourself. Even the "Book analysis window", which shows the critical lines of play, don't foresee 8.f4 as part of the options it presents; it's been played just too few times for the automatic functions of the program to pick up. So you have to look ahead and do your own evaluation.

Realistically, given the same position again today, I'd probably still play 6...Ngxe5; castling seems too wimpy ("solid" play is not why I chose to gambit a pawn) and, to be honest, not too many of the "average"-level players I face are going to weaken their Kingsides with 8.f4 after the exchange of Knights on e5. But the next time around I'll at least look ahead and consider the possibility, which is a whole lot more than I did in this game.

As it turns out, my opponent didn't play the "proper" followup (namely 9.Bd3), though he did choose to move the Bishop:

9.Be2 0-0 10.0-0 Re8

...which, believe it or not, transposes back into known games. White has three viable options here (according to the statistical tree at least): 11.Bd3, 11.Bf3, and 11.f5; Black does well in none of these lines. But my opponent chose to give me a gift:

11.Qd2? Bxe3+ 0-1

Faced with the loss of his Queen, my opponent resigned. But the odd part is that his position is still theoretically playable; at the cost of two pawns his development is complete while I still need at least three moves to connect my Rooks. I'd still rather be Black here, but White might have played a few more moves just to see what happens. It would seem that I wasn't the only one who wasn't taking the time to look ahead.

And, yes, while this does qualify as one of the 20%-25% postal miniature victories I mentioned at the start of this article, I sweated a fair little bit along the way -- I never said that I didn't have to work (and worry) a bit to earn them.

Until next week, have fun!

---

© 2004, Steven A. Lopez. All rights reserved.