by Steve Lopez

This week we'll continue our discussion on searching for position fragments (begun in last week's ETN). I wasn't aware of this last week, but these searches are also possible in Fritz6, Junior6, Hiarcs7.32 and Nimzo7.32. Since I use Fritz for playing and analysis and ChessBase7 for database functions, I'd forgotten that these searches could be done in the playing programs. A special "thanks" to the people who reminded me.

The King's Indian Attack is a pretty popular opening among players who don't want to learn a lot of variations. It can be used effectively against a variety of opening setups by Black. This flexibility is what makes it a good candidate for a position fragment search. Once again, open the search mask, click the "Position" button, select the radio button next to "'Look for' board" and place the pieces as shown in the diagram:

This gives us the typical KIA formation, with the e-pawn advanced, the Kingside Knight developed toward the center, the Kingside Bishop fianchettoed, and short castling. I've set the range of move numbers to 5 through 7 to cut down on the number of transpositions from other openings (in other words, openings that weren't intended as KIAs but ended up with the Kingside structure shown in the position fragment).

In this case, the "Exclude" and "Or" boards aren't needed; we're just looking for a particular Kingside pawn structure for White.

Once the search is completed in Mega Database 2000, we see a list of 22783 games in which this White structure appears. By further refining the search to only show annotated games (as discussed last week), we cut that number of games down to a somewhat more manageable 465.

Searching for tactical positions can also be done, but this is a bit trickier. You can, of course, do a text search on an annotated database for words like "pin", "fork", and "skewer", but it's also possible to do a search for tactical position fragments. The reason this is trickier is because you have to look for specific piece configurations; there's no "magic button" that you can select to get all of the pins in a database.

Let's look at an example:

In this case, we're looking for cases in which a White Bishop on c4 is pinning a Black Rook on f7 to the Black King on g8 between moves 9 and 20. However, starting the search now will only bring up games in which these three pieces are on the specified squares -- they won't necessarily be pins. There might also be pieces on d5 and e6, blocking the pin.

How, then, do we designate a pin? It's simple: we just go to the "Exclude" board and tell the program that there can't be any pieces on the diagonal between the White Bishop and Black Rook.

We use the "circle" buttons for this. The last piece button in each row is a circle. This is used as a "wild card" (the manual says "joker", which is one of those neat little translation glitches you'll see from time to time. "Joker" means essentially the same thing as "wild card" but isn't exactly standard English computer vernacular). The wild card in this case stands for any piece, just as a wild card in poker can be used as a substitute for any card. Put both a White and Black wild card on d5 and e6 to show that no pieces of either color are allowed to be on those squares. So click the "'Exclude' board" radio button and drop in the wild cards:

Once that's done, you can do the search. You should come up with 629 games when the search is completed in Mega Database 2000.

Forks and skewers can be handled in a similar manner. You just need to keep in mind that you're looking for a specific fragment, not all forks or skewers in general. I've seen (and played) plenty of games in which a White pawn on b5 forks two Black minor pieces on a6 and c6. So, with curiosity getting the better of me, I decided to do a search for these games.

First we place a White pawn on b5 on the "Look for" board. But with this search we need to fudge a little bit -- Black needs to have two minor pieces on the board, on a6 and c6, but it really doesn't matter what minors are on what squares. To define this somewhat fuzzier parameter, we turn to the "Or" board and put a Black Bishop and Knight on both a6 and c6:

So the search is defined as "all games in which a White b5-pawn forks two Black minor pieces between moves 8 and 20". Upon clicking "OK" (and after a short wait), we come up with 245 games in Mega Database 2000.

It's not hard to get the hang of position fragment searches in ChessBase and Fritz. The key things to remember are the differences between the "Look for", "Exclude", and "Or" boards, as well as the effects of the "First" and "Last" numerical values. The biggest thing to keep in mind is that partial position searches are possible in the programs; you don't have to do an exact position search every time. Play around with this search feature and remember that you can combine it with other search features to limit the number of hits you get. You'll be searching like a pro in no time.

Until next week, have fun!

You can e-mail me with your comments, suggestions, and analysis for Electronic T-Notes. If you love gambits and sacrificial play, stop by my Chess Kamikaze Home Page and the Yahoo Chess Kamikazes Club.