by Steve Lopez

We've previously looked at searches for a specific position. This time around we'll examine searching for position fragments.

A position fragment is a part of a position that is important no matter what else is happening on the board. Most of your standard middlegame motifs fit this description (Rook on the seventh, Knight outpost, etc.). In fact, this description is what makes fragment searches so important for developing players.

A comment/complaint that I frequently hear from beginning to intermediate players is that chess books often don't contain enough examples to illustrate the concepts they teach. A writer will introduce an idea, illustrate it with an example or two, and then it's off to the next idea (frequently leaving the reader wanting more).

"Why should a beginning or intermediate player own a chess database program?" is a question I often hear. A database is a wonderful reference tool, not only for finding all the games of a certain player or tournament (although these are certainly valuable functions), but also as a library of supplemental information to be used in conjunction with printed chess materials.

This is the primary use for all those unannotated games you've been collecting. You read about a positional concept in a chess book, play through the couple of examples the author provides, and then go find more examples to play through from your database.

Here's a concrete example. Let's say that you've been reading about isolated d-pawns (an amazingly rich and complex topic, so the couple of crumbs the chess author throws you are just not enough). There are a couple of tips you should know about isolated d-pawns:

1) The player with the isolani should try to control the square in front of the pawn with as many pieces as possible (making it difficult for the opponent to blockade the pawn, as well as preparing to advance the pawn).

2) The opposing player should blockade the pawn by planting a piece on the square directly in front of it, try to control that square with other pieces, and swap off as much material as possible (as the lack of support will tend to weaken the pawn).

Obviously, you'd like to see some concrete examples of these ideas, but the two examples from your book just aren't enough. So you fire up Fritz to search the database that comes with the program. The database "fritz5.cbh" located on the CD contains 303,043 games, which should be plenty for our purposes.

Go to the database screen, click the "Search" button, then click the "Position" button in the Search Mask. Here's how you would search for an isolated White pawn on d5. Making sure the radio button for "'Look for' board" is on, place a White pawn on d5.

If you were to start the search now, you would get every game in which White had a pawn on d5. But what you want are games in which the d5-pawn is isolated. How do you accomplish this?

Click the radio button next to "'Exclude' board". This is where you tell Fritz what you want to be absent from the board. You would place some White pawns on the search board as seen in the following illustration:

You'll notice that in the 'Exclude' board, we've placed White pawns on all the squares where we don't want to see a White pawn. There can not be a White pawn on any of these squares. This is particularly true of the c- and e-files (the absence of pawns here defines the fact that the d5-pawn is isolated). We've also placed White pawns on all squares of the d-file except for d5; this insures that we'll see no doubled pawn positions.

It's also a good idea to narrow the range of moves that are searched (as described in Base Basics 4 a few weeks ago); this will speed up the search by reducing the number of moves Fritz needs to search through. A good set of values here would be "10,1,30" (remember that "Last" and "Length" are inadvertantly reversed). If you set the range when you set the 'Look for' board, the values will carry over to the 'Exclude' board.

Now we're ready to search. Click "OK" and when the program takes you back to the Search mask, click "OK" there, too. Then go get a cup of coffee; even on my 133 MHz Pentium, the search takes a few minutes.

So when you come back, you still see a mammoth game list. You wanted information -- you got it! I stopped the search after 45,000 games had been examined. Fritz found over 3500 in which White had an isolated pawn on d5. Fritz also has a limitation of 4999 games in a specific search, so further refinement of your search parameters is a good idea in cases like this (as the program may not provide a complete list of matches if many games meet your search parameters).

This obviously qualifies as information overload. How do you narrow the search still further?

One way is to limit the number of years across which it searches. A club-level player frequently has trouble understanding games played by current grandmasters, but games from the early part of this century are often easily understood. So you might try limiting the search to games between 1901 and 1935. You might also try limiting the search to games of specific players (and combining this with a limited span of years should really speed things up, as Fritz searches header information first before looking at the actual moves of a game).

Another way is to be more specific in defining the position fragment. For example, you might try games in which Black has a piece blockading the passed d-pawn:

This is what the "black dot" symbol is used for. It means that any Black piece can be on that square. However, in this case it's probably not what we want, as the black dot can also mean that there's a pawn on the square.

It looks like we'll have to go with specific Black pieces on d6. As the Knight is considered the best piece for blockading a passed pawn, we'll check this out:

Also note that I've narrowed the range of moves slightly to "15,1,28". The 'Exclude' board remains unchanged (except for the range of moves). If I wanted to be technical, I could remove the d6-pawn from the 'Exclude' board, as the Black Knight makes that an impossibility anyway, but it's not necessary to do so.

After clicking "OK" and waiting a few minutes, we see our result: 852 games contain a White isolated d-pawn blockaded by a Black Knight between moves 15 and 28. This is considerably better than a couple of games or game fragments from a book! Now you can play through some games and get a better feel for the isolated d-pawn's positional ideas mentioned earlier.

How do you decide which of the 852 games to play through? If you're like a couple of grandmasters of my acquaintance, you play through them all (I've seen these guys use ChessBase's automatic replay function to view hundreds of games at a sitting, all at blinding speed). For mere mortals like us, a couple of dozen games will probably be sufficient. You might start with the annotated games and then just scroll down the list, picking out players that seem interesting to you (John Nunn seems to be a huge favorite among my crowd). Another thought is to copy all of the games into a new database, so that you'll have them whenever you want them without having to perform a new search.

There are some other neat features and uses for Fritz's position search, but this article (and Base Basics 4) should be enough to get you started. Experiment with the Search mask and see if you can discover some valuable tools to help you with your current chess study. There are even more tools available in ChessBase's Search mask, and we'll examine those in the weeks to come.

I hope this series on Base Basics has started our new users off on the right foot, and perhaps provided a refresher for the old guard. A database is a valuable tool for anyone (regardless of experience or ability) who is trying to improve his or her chess. Learning how to do so is an ongoing process and what Electronic T-Notes is all about!

Until next week, have fun!

Your questions, comments, and submissions are strongly encouraged.

You can reach me by e-mail.