ELECTRONIC T-NOTES


CHESSBASE USA'S WEEKLY ON-LINE NEWSLETTER


FOR THE WEEK OF JUNE 25, 2000


THE SHOOTOUT

by Steve Lopez

Anyone who's been playing chess for any length of time has encountered the frustration of the "unclear position". You're reading a chess book, moving right along, enjoying the analysis, and you suddenly get stopped short by the annotator telling you that a position is "unclear". This usually means one of three things: that the annotator didn't feel like analyzing the position, the annotator didn't have a freaking clue as to what was going on in the position, or that the position really is unclear.

In my personal lexicon, an unclear position is one in which neither side has a firm plan that needs to be followed. The position is fairly quiet and both players may have run out of ideas. Or another way of looking at it is that both players have so many ideas that it's hard to choose just one plan as the definitive one. Or perhaps both players have made such a mangled car wreck of their respective positions that it's very hard to sort out what needs to be done first. But typically it's a quiet position in which no single obvious plan is screaming out to be followed.

In any case, it's pretty frustrating to run across the "unclear" evaluation, particularly for the average player who lacks the knowledge and skill necessary to sort the position out. That's where a playing program like Fritz can provide some guidance. There are a couple of ways to deal with unclear positions using Fritz. One is to use the Deep Position Analysis (formerly called "Correspondence Analysis") feature to generate a tree of possibilities from the unclear position. This was covered in the ETN issue for September 13, 1998.

Another means of exploring an unclear position is to let Fritz play against itself using the position as a starting point. It's a sort of "practical" approach to positional examination, as opposed to the "theoretical" approach of the Deep Position Analysis. You hand the position to Fritz and have it play out the game to find out what happens.

In Fritz6, this feature is called "Shootout" mode. You can have Fritz play a game (or multiple games) against itself at different ply depths to see what happens in the middlegame and endgame. (For an explanation of ply depths, see ETN for January 2, 2000).

The first step is to set up the position in question. For opening positions, this is probably best accomplished by hitting ALT-F2 to start Infinite Analysis and then mousing in the moves. For middlegame or endgame positions, go to the File menu, select "New", and then "Position setup" from the submenu. Click "Clear board", put the pieces on the board using the buttons located to the right of the chessboard, and click "OK" when you're done. Be sure to specify which side is to move in the position.

Once you have the position set up, you're ready to set up your shootout parameters. Go to the Tools menu, select "Analysis" and then "Shootout" from the submenu. This brings up the Shootout dialogue box.

The first thing to do here is to select an engine. Click the "New" button to bring up the Engines dialogue, pick an engine from the list, and click "OK". You can have more than one engine participate in the shootout; when you get back to the Shootout dialogue, click "New" again and select another engine. Note that two engines don't play against each other in Shootout mode, only against themselves. For example, if you choose Fritz6, Crafty, and Comet, Fritz will finish all of its games against itself before Crafty starts playing itself. Once Crafty has completed its games, Comet will load up and play its games. (And, by the way, just as "New" allows you to add engines to the list, "Delete" lets you remove engines from the Shootout list. It does not delete the engines from your hard drive).

Next set your move limit. The default is "999", which means that the games will be played to completion (keep in mind that the longest chess game in history was 200+ moves, so it's doubtful that you'll ever get anywhere close to 999). However, if you're looking at an opening position and want to know the middlegame plans without worrying about the endgame, you can set this to a lower value if you wish.

The "depth" setting requires a bit of explanation. If you want Fritz6 (or any engine) to play just one game against itself, set both values to the same number. For example, if you set both values to "11", Fritz will play one game against itself using an eleven ply search depth. But if you want Fritz to play more than one game, each at a different ply setting, you'll need to enter a range of numbers. For example, if you enter "9" in the left-hand box and "11" in the right-hand box, Fritz will play three games, one using a nine ply search depth, one at ten plies, and one at eleven plies.

Note that these settings apply to each engine you've selected. So if you've picked three engines and set the ply depth to nine to eleven, you'll get nine games, three played by each engine.

There's one more tweak we should look at: the check box for "Skip even plies". As we discussed previously in ETN (again in the January 2, 2000 issue), some chess programs can become tactically "blind" at even ply depths and consequently produce weird results. My personal preference (one which I strongly encourage) is to check the "Skip even plies" box. If you check this box and set the ply range values to "9" and "11", you'll get just two games instead of three -- one using a nine ply search depth and one using an eleven ply search depth. The ten ply depth will be skipped (10 being an even number the last time I checked).

Another caveat which is largely hardware dependant: don't set a ply value that's too high for your machine. Fifteen plies seems to be about the current maximum, even on the fastest PCs, and when I say "maximum" I mean that you'd better not need your computer for a day or more. I've seen Fritz5.32 (which is a very fast analyzer) take twenty minutes or more per move when searching to a depth of fifteen plies on a Pentium II 266 with 32 MB RAM. As always, your mileage may vary. Personally, anything over thirteen plies scares me off.

Once you've set your values, you click "OK" and the program is off and running. A status box appears on the screen, which you can safely close if you want to view the games in progress. Of course, the obvious question to ask here regards what happens to the games once they're finished. How do you save them? Answer: you don't have to -- they're saved automatically into your Engmatch.cbh database. This is a database created by the program when you installed it. The exact folder location depends on what you specified when you installed Fritz. If you used the default value given by the Installation Wizard, the location is:

If you set a different location and have forgotten what folder you specified, just click the Start button on your Windows taskbar, select "Find", then "Files or folders", and type "EngMatch.cbh" (without the quotes, of course) in the box provided.

Of course, you'll probably never need to go looking for the folder. Go to the Database window in Fritz and look in the pulldown window in the upper right of the screen -- you should see the EngMatch database listed. Click on it to select (and open) it and you'll see your new Shootout games at the end of the game list. You can play through these games and have a look at the way the program handles the resulting games starting from your "unclear" position. You can have multiple engines play the games using multiple ply depths and compare the results. This will also doubtless show you a variety of ideas and ways to approach the position (especially if you use multiple engines. From a quiet position, engines with very divergent styles of play, such as Fritz and Hiarcs, will often tackle the position in completely different ways and provide you with a choice of plans to consider).

This information can be pretty useful, espcially when you're looking at one of the favorites from your opening repertoire. You tend to play the same opening again and again, but you sometimes find yourself in a particular variation in which you have trouble devising a plan. Reviewing the ways that different chess engines handle such a position can give you a lot of good ideas on how to find a plan and handle that position when it comes up in your own games.

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.