LEARNING A NEW OPENING

PART 2

by Steve Lopez

Last week, we discovered that the most important ingredient in learning a new chess opening was simple-language explanation of the ideas found in that opening. So once we've consulted a source for those ideas, what's next?

The next step is to construct a framework or skeleton upon which to hang these ideas. For the sake of our example, I've decided to investigate the Scotch Opening. There are several reason for this choice:

1) I want a surprise weapon in my arsenal, not for everyday use but to spring occasionally on opponents who are familiar with my regular repertoire;

2) I once lost a tournament game on the White side of the Scotch to a twelve year old girl and I'd like to avoid a repeat performance;

3) As a fan of the late, great Mikhail Tal, any opening named after an alcoholic beverage is OK with me (just for grins, ask a bartender to fix you a "Bogoindian" and watch his face, but please remember to tip well).

Digging into my chess library, I came across a suggested main line for the Scotch and a large number of sublines and variations. I confined my research to just a few opening books with text explanations of the openings, to cut down on the amount of material.

Once you've selected an opening and found some source material, your next step is to create a "work" database to store the games you'll be constructing or playing. You can use [CTRL-X] in ChessBase to do this, or in Fritz4 you can click on the "Database" icon and then click on the "New" button on the left of your screen. Make sure you know what directory you're placing your work database in, so you can find it later. I named my database "training.cbf" (making it an old-style ChessBase database), used the "Training" icon for its picture in CB6, and shortened its name to simply "Training" in the database window. I then clicked once on the icon and clicked the "fire" icon at the bottom of the screen, causing my "Training" icon to be highlighted in red (which makes it my working database).

The reason I chose to make it an old-style .cbf file instead of using the new data format is that I want to be able to use the data I generate with Fritz as well as with ChessBase, without having to convert individual games back and forth. Later we'll convert the data over to CB6 format to be able to use the new tools in ChessBase 6.

I then clicked on the "chessboard" icon at the top of the left-hand row of icons, opening a new game window. I entered the "main line" that I found in a book and then saved the game ("Game" menu -- "Save game", or else use the "piggy bank" icon at the bottom of the screen). I typed "Analysis" for the name of the White player and omitted the year entry when prompted for the game header information.

Next I proceeded to enter various branch variations. (The easiest way to do this is to use the [T] key. If a variation starts with Black's 12th move, for example, click on the 12th Black move in the main line and hit [T]. This will take back the move and allow you to enter an alternate 12th move. Enter the rest of the line normally). I typed in any text commentary (taken from books) that I thought would help me study the opening ([CTRL-A] opens the text annotation box). I also included any diacritics (?, !, etc.) plus any evaluation symbols given in the books (accessible from three icons at the bottom of the game window).

The trick here is knowing when to stop adding variations. The waggish comment would be to tell you that if your head explodes while studying the game tree you've gone too far. The truth is that a short main line (10 - 15 moves) with about a half-dozen branch variations is plenty for right now.

Going back to my Scotch tree, here's what I came up with after consulting two or three general books on the openings (the commentary has been removed to avoid violating copyrights):

1.e4 e5 2.Nf3 Nc6 3.d4 exd4 4.Nxd4 Nf6

[4...Bc5 5.Be3 Qf6 6.c3 Nge7 7.Qd2 (7.Nc2 Bxe3 8.Nxe3 0-0 9.Be2 d6 10.0-0 Be6 11.Nd2 d5) 7...d5 8.Nb5 Bxe3 9.Qxe3 0-0 10.Nd2 dxe4 11.Nxe4 Qe5 12.0-0-0]

5.Nc3

[5.Nxc6 bxc6 6.e5 Qe7 7.Qe2 Nd5 8.Nd2]

5...Bb4 6.Nxc6 bxc6 7.Bd3

[7.Bd2 0-0 8.Bd3 d5 9.f3 dxe4 10.Nxe4 Nxe4 11.fxe4 Bc5]

7...d5 8.exd5 cxd5

[8...Qe7+ 9.Qe2]

9.0-0 0-0 10.Bg5 c6

[10...Be6 11.Ne2 Be7 (11...h6) 12.Nf4 Qd6 13.Re1 Rae8 14.c3 (14.c4 dxc4 15.Nxe6) 14...c5 15.Qc2]

11.Qf3 Be7

[11...Bd6 12.Bxf6 Qxf6 13.Qxf6 gxf6 14.Ne2 c5 15.c4 (15.b3) ]

12.Rae1 Rb8

[12...h6 13.Bxh6 gxh6 14.Qe3 d4 15.Qxh6]

13.Nd1 Re8 14.h3 Be6

The moves in bold type will be explained later. There are actually a few more sublines in this tree than I normally like to have initially, but there was a lot of text explanation that went with them, so I added the extra lines.

Once you've duplicated this process, you'll have a small game tree with text commentary and analysis symbols. What should you do with it?

There are several techniques and uses that you should be aware of for using this tree. Try them all out and use the ones that work for you.

A major aid to comprehending an opening is to understand how the position stands at the end of a line or subline. Each of the moves in bold type in the game tree above is the end of a line (and each has an Informant evaluation symbol in the original game from my database -- again, these were edited out to avoid violating copyrights). These moves at the end of the lines are called variously "terminal positions" or "leaf nodes". It's the point where concrete moves stop and general analysis or evaluation is given.

It's important to check these the evaluations of these leaf nodes for accuracy. We want to make sure that a) the evaluations are correct, and b) we understand why the position is evaluated that way.

For example, the final position of the Scotch game tree I offered is evaluated as being "equal". That seems reasonable; White has more of his pieces developed, but his Knight stands on an awful square, and Black's d5-pawn is strong. Let's check the evaluation. If you have one of ChessBase's analysis engines, just open up the game tree in ChessBase, click on the move in the notation to jump to it, and then fire up the evaluation engine of your choice (either by clicking on the "computer chip" icon or hitting one of the [ALT-Fx] combinations). Let the engine rev up to 9 ply or so and see what evaluation it gives the position. Use more than one engine if you have them. In my own tests, both Fritz modules, the Hiarcs 6.0 module, and Doctor 2 agree that the position is equal.

In Fritz 4, the procedure is a bit different. Load the game tree, click on a leaf node, click on the "Game" icon, and then click on "Infinite Analysis". Fritz will then analyze merrily until you click on the "Stop" button. You can hit [F3] to switch engines and analyze a second time (or more) as we did with ChessBase above.

It's a good idea to take a look at the line of play generated by your analysis module. It will give you more clues as the "why's" of a position's evaluation.

Another way to use the game tree is as a tool for remembering the moves. In the Game window in ChessBase, click on the second icon from the left at the bottom of the window. This icon controls the style of notation displayed in the notation box. Select "Training". The notation disappears from the notation box, replaced by a line of text telling you whose move it is and the final evaluation of the main variation.

Try to guess the first move of the game (in the case of the Scotch, it's an easy pick). Once you've guessed the move, click on the fourth "VCR" button below the chessboard (the "play" button, designated by an arrow pointing to the right). The program will reveal the first move and play it on the board. Then you try to guess the next move, then the next, and so on. In this manner you can drill yourself on the proper move order in the opening. Another neat feature is that if text commentary is encountered for a move, it will be displayed in large letters in the notation box as well.

(By the way, speaking as a complete over-the-top chess software freak, there are other programs that contain these drill-type features as well. The best of these is found in Bookup, as Bookup's training mode allows you to actually move the pieces on the board and lets you know if you're correct. And if you're one of the six or seven people who own a copy of Super Drill, you know that program will do the same thing as Bookup's training feature plus grade you on how well you did.)

Hopefully, this drill-type training wil not only help you commit the moves to memory but will also teach you the ideas found within your chosen opening.

Another technique is what I call "blitz memorization". Basically, all you do is play through the game over and over and over again until you feel that you're able to recognize the positions and patterns that arise from your chosen opening. In Expert Systems: Principles and Case Studies, editor Richard Forsyth says, "Human intelligence is approximately 99% pattern recognition and only 1% reasoning." In other words (applied to chess), it's a whole lot easier to see a position and say, "Hey! I've been here before!" than it is to sit and try to figure it out from scratch. "Blitz memorization" can possibly help you develop your pattern recognition skills (and, by the way, this tip is merely a variation on the last method I just described, namely the "drill" method, and should probably precede it in your process of learning an opening).

Once you think you have the basic variations and ideas down it's time to move on to some practice. Here's where chess-playing programs come in handy. But in moving on to this phase of the program, we crack open a whole new can of worms.

The first debate revolves around the question of how strong a training partner you want or need. One school of thought leans toward the idea of using the strongest chessplaying program possible, set at its highest difficulty level, to aid you in your training. The theory here is that the program will play only very strong moves and show you where any weaknesses lie in your knowledge of the opening. The downside of this approach is that you will be unable to accurately judge your progress, as you'll never win any games.

The second approach is to use weaker programs or strong programs set to weaker levels. This way you can tweak the program to give you a challenging game without being overwhelming (and give Class B-E players a game similar to what they would get under actual tournament conditions).

Personally, I favor the second school of thought. It's hard enough to learn a new opening without getting your skull kicked in by the computer. A master-level friend of mine, Pete Prochaska, once advised me to set Fritz at a level where I would win about 40% of my games. This would be a level where I would feel challenged but not threatened, while still improving my game.

I'll show you how to do this. Fire up Fritz4 and click on the "Levels" icon. First hit [F3] to choose an engine. The choice will depend on your playing strength and the speed of your computer more than anything else. If you're rated under 1300, you should probably load the Fritz 1.20 engine. Higher rated players can go ahead and load Fritz 3.10 or Fritz 4.01.

Next click on "Handicap and Fun" (or just hit [CTRL-H]). The only setting you should change is the "Playing Strength" setting. Keep in mind that Fritz uses FIDE ratings, which tend to be about 100 points lower than USCF ratings. So set Fritz's playing strength to be about 100 points higher than your USCF rating. This will approximate playing against someone 200 USCF points higher than you.

If you're like me (a complete chess software geek), you probably own more than one chessplaying program. Go ahead and use multiple programs, as different programs will tend to play in different styles. This will simulate the variety of players you'll meet over the board. Be aware, though, that this approach has its pitfalls, as the following opening illustrates. I was White, playing against "Program X" (name changed to protect the guilty), a Windows program I bought at a local toy store:

1. e4 e5 2.Nf3 Nc6 3.d4 Nxd4??

Whoops! It turns out that "Program X" doesn't have the Scotch in its opening library; consequently it played this dog and I proceeded to whup up on it.

Speaking of opening libraries, you can make a Fritz PowerBook out of the analysis tree you created earlier. Just open up your analysis tree in ChessBase, then click on "Save Fritz Book" in the "Technical" menu. Save the book into the directory C:\FRITZ4\BOOKS; when you name it, give it the extension .FBK (for example, I called mine SCOTCH.FBK). When you go to play against Fritz, click on the "Levels" icon in Fritz4 and the "Load book" command. In the window that opens, click the "Load" button and select your new homemade book from the list that appears.

There is also a debate about time controls (a moot point if you're changing the strength of Fritz, as you can't select a time control if you've changed the playing strength). As you know, the longer a computer has to think, the better it plays. So which is better -- long time controls or short?

At this juncture, I feel that it's better to play four 30-minute speed games than it is to play one game at tournament time controls (40/2). What you're trying to get here is experience and a lot of it in a hurry. So play faster games; it's the best thing for you right now.

As you play these games, save them into your training database and have Fritz analyze them afterwards. This will help you pinpoint inaccuracies and deficiencies in your play. It will also give you a good record of your improvement. You'll be able to go back and look at your early games and see how much your understanding of the opening has grown.

Next week, we'll delve into deeper opening knowledge. Until then, have fun!

I'm very interested in reading your opinion of this (or any other) issue of ElectronicT-Notes, as well as my book Battle Royale. I invite you to post comments to our ChessBase Users Group or else you can e-mail me directly.