by Steve Lopez

Wolfgang Krietsch writes:

I'm really getting confused with the ECO system. Since I'm a chess software addict (can you imagine?), I tend to play around with different database & plaing program besides my main combination of Chessbase and Fritz. What confuses me is that sometimes different programs give different ECO codes for the same game...

Actually, ChessBase may give you different ECO codes for the same game, believe it or not. It all has to do with a funny concept known as transpositions...

Before we delve into the wack world of transpositions, we need to make a side detour to define exactly what consititutes a ChessBase "key".

A key in ChessBase is comparable to an index in a chess book. In a paper book, you can flip to the back and usually come across a list of the games in the book, indexed by openings. So, for example, if you're reading a book on the Sicilian Defense and you want to find all the games in which the Wing Gambit was played, you'd just look up "Wing Gambit" (or the move order) and there would be the list.

ChessBase makes this process even easier. Assuming you're looking at a pre-assembled database with a key that someone has previously created, you'd simply click on the "key" icon, click on the entry for "Wing Gambit", and you'd get a list of all of the games from the database in which this gambit was played. Then you could just click on a game in the list and play through it on your screen.

ChessBase can accomplish this because someone created a key with an entry for the Wing Gambit. How does the program "know" a game is a Wing Gambit? Because there is a board position attached to every item in a ChessBase key. In this case the board position looks like this:

ECO (the Encyclopedia of Chess Openings) is simply a classification system for opening positions that uses an alphanumeric code rather than words to classify and describe openings. For example, the Wing Gambit is ECO code B20 (along with a couple of other offbeat replies to the Sicilian). There are 500 of these classifications. ECO runs five volumes with 100 codes per volume. Here's (roughly) how the volumes break down:

A - the English opening and all openings not covered in the other volumes
B- 1.e4 openings in which Black does not reply with 1...e5 (excluding the French Defense)
C - 1.e4 e6 openings (the French Defense) plus all 1.e4 e5 openings
D - all 1.d4 d5 openings
E - all 1.d4 openings in which Black does not reply with 1...d5

Thus the Wing Gambit (B20) is part of the 20th classification within Volume B.

Let's imagine for a moment that someone handed you 500 manila folders, each with a chess diagram on the front of it. He then hands you a stack of tournament scoresheets. He tells you to play through the games and when you hit a position that's pictured on a folder, file the scoresheet in that folder. That is exactly what ChessBase is doing when it classifies games into an opening key. Each individual entry in the keyfile is a "manila folder" and ChessBase is filing away the gamescores.

It's pretty simple. Instead of looking at move orders, ChessBase looks at positions.

So it doesn't matter if the game was played between two Class B players who played 1.e4 c5 2.b4 or two unrated patzers who played (for some ridiculous reason) 1.b4 c5 2.e4; the positions are the same and will be classified together in the same key.

Two different openings that wind up at the same identical position are said to transpose into each other. In the example I just gave, the first opening is obviously a Sicilian Wing Gambit. The second opening is a Sokolsky (or "Orangutan") that took a left turn at Albuquerque and became a Sicilian Wing Gambit.

And this is where the trouble begins. It would be great if chess would cooperate and we wouldn't have to deal with all of these messy transpositions. Unfortunately, Caissa throws us plenty of curves.

Here's an example I ran into yesterday in a game I was playing. The game began this way: 1.e4 c5 2.c3

which is an Alapin Sicilian (ECO code B22). But my opponent tried to hoodwink me by playing one of the more unusual lines, so we wound up with this: 1.e4 c5 2.c3 e6 3.d4 d5 4.e5

Still an Alapin Sicilian, right? Let me show you a line from ECO's Volume C: 1.e4 e6 2.d4 d5 3.e5 c5 4.c3

Uh-oh. I thought I was playing an Alapin Sicilian, but I now have the exact position that arises from a line in the French Advance Variation (ECO code C02)! So which is it - an Alapin Sicilian or a French Advance?

Believe me, the position doesn't care how the pieces got to the squares they're on and in what order. Grandmasters love to talk about the "requirements of the position", in other words, what plans are expected to be carried out by both sides. The requirements of this position are constant, regardless of which move order got us here. Between you, me, and the wall, if I ran into this position in a postal game and needed some advice on the "positional requirements", I'd toss aside my book on the Alapin and grab one on the French, pronto!

This creates a problem. Going back to our hypothetical "manila folder" example, how would we classify this game? Is it a Sicilian or a French? What folder do we put it into?

We're now back at our original question. How would ChessBase classify the game I played yesterday?

When classifying a position (either when saving a game or when sorting games into a key) and two or more positions from the game appear as classification positions, ChessBase uses the position that occurs last in the game. Thus, in our example, ChessBase will classify this game as C02, a French Advance. Another chess program might classify it as an Alapin Sicilian (as it would play through the game and stop at the position after 2.c3).

In ChessBase for Windows 1.11 and in earlier versions, games could actually be classified both ways. The ECO utility that accompanied the program (the utility that added the ECO code to the header info) played through the game forward and stopped at the first classification position that it found. The function within ChessBase itself that sorted games into keys played through the games backward until it hit a classification position. This is why you would sometimes click on an ECO key entry and find games that had a different ECO code in the header: these were games that transposed from one opening to another.

It's a knotty problem and one that will likely never be solved to everyone's satisfaction. I'm currently involved in a postal game that ChessBase classifies as a B10 Caro-Kann but ECO calls a B13 Caro-Kann, because of a transposed move order. I've done searches for both ECO codes in ChessBase and found games classified under both codes that match my game.

The best solution is to stay alert! Look for transpositions in the games you play and remember that it's more important to learn how to handle a position than it is to memorize a ton of opening variations move-for-move. If all you do is memorize move sequences you'll almost certainly be caught with your pants down by an unexpected (and unnoticed) transposition in one of your games!

You can reach me by e-mail with your ideas and suggestions.