Opening trees

BY: BOBBY ANG

(An abridged version of Chess Piece, BusinessWorld, 6 May 2011)

When Chessbase 4 was first introduced to the chess world in 1992, it completely revolutionized how we study chess. If you wanted to research on somebody’s games you can simply click on a few buttons and get his games on the computer screen. Then you can instantly replay the moves and even have a "chess engine" in the background analyzing them, pointing out improvements and/or mistakes.

I remember one of our international masters, now a GM, who dropped by my place one day to look over a few games and play over the Internet Chess Club. During the merienda break we got to talking about what he considered his best game. He showed it to me over Chessbase 4 -- it was an exciting game played in one of the Far East Bank Opens and it ended in a blaze of tactics. Problem was, as the IM proudly showed off the "winning" tactical blow, the chess engine immediately pointed out that his opponent missed a mate in three.

In those early days we carefully collected, sorted and cleaned up all the games we could get. And then we shared it with our friends and they in turn will share their collections.

Later on the opposite problem arose -- too many games! How can anyone absorb the huge volume of data that has accumulated and is being added on daily? You play over a game, see an interesting position, and then ask for the program to search for similar positions, but then a thousand games pop up on your screen. How are you to digest this information?

The answer is the opening tree. Let’s say you have a collection of 1,000,000 grandmaster games. Instead of storing the games as 1 Million individual collection of moves, it stores the positions. So you have the starting position, then you see that white has eight moves (in order of popularity): 1.d4, 1.e4, 1.Nf3, 1.c4, 1.g3, 1.b3, 1.f4 and 1.Nc3. Then you pick 1.e4 and see that in response to that Black has played 1...c5, 1...e5, 1...e6, 1...c6, etc etc Each successive move "branches" to the next. Pretty soon you are deep into opening theory, and at each position you are shown what moves have been played by the GMs and IMs, and statistics as to their success rate. This is a really cool feature!

A whole new dimension of openings research opened up -- why content ourselves with human games? Thousands of computer vs. computer chess games were played and stored to serve as a guide for human players. The silicon monster was after all immune to the one- or two-move errors that commonly afflict carbon-based games. And they don’t tire -- the computers can be set to play continuously over several weeks exploring all the vicissitudes of a variation until it is almost played out. Many new ideas were born from such research -- for example, have you ever heard of the Fritz Attack in the Sicilian?

Hamdouchi, Hichem (2600) -- Zhigalko, Andrey (2591) [B86]
TCh-ESP CECLUB Honor KO Sestao ESP (7), 28.08.2010
REQUIRES JAVA TO VIEW THE GAME

After several thousand computer vs computer games played these are all merged together into one big opening tree and now they have started to sell these trees separately. You have the Fritz Power Trees and specialized Rybka and HiArcs trees.

I have found the Rybka 3 Tree to be the best (yes, I know there is a version 4 but I still like the original Tree better). You go over the moves of the Tree to see which lines Rybka3 liked with the alternatives ranked by means of the usual annotation symbols ! ? !? and ?!.

In the same way that Computer-Generated Imagery, the so-called CGI, is revolutionizing the way we do movies, the Computer Opening Trees has become the best way to study the openings. Better than reading paper-based books.

There was a time when most players did not want to get into analysis of long, forced variations because of the danger of mistakes. As Larsen said, "long variation wrong variation." Nowadays, with the advent of computers, these long variations has become the battlefield.

Reader comments/suggestions are solicited. E-mail address is bangcpa@gmail.com