The Chess Improver: Making up fake games to improve understanding of themes
For The Chess Improver, I wrote ""Making up fake games to improve understanding of themes"".
(The original Chess Improver site is no longer active; the rescued article follows below.)
When I was playing chess as a child, one of my favorite things to do, especially before my father found out about the existence of chess clubs and we started attending a local one when I was ten years old, was to make up fake games in order to enjoy seeing interesting play using some idea. In particular, it is useful to see how one can win using an idea, against, of course, less than perfect defense. Back in the day, one could use books or magazines to find collections of games to sudy, and but there was no way to search efficiently for exactly what one was looking for. Today computer databases and sophisticated search programs and even query languages enable doing many kinds of research that were not possible 30 years ago.
Doing it yourself, with computer aid
But even today, I think it is fun and instructive to make up one’s own fake games. And there are better ways to make more realistic fake games, thanks to computers. In the past, if you were not already a master (or even if you were), there was no guarantee about the quality of your invented game. Maybe you created a fantasy game featuring an attack, but weren’t sure whether it was sound at all.
There are many ways today to create fantasy games that can serve different purposes. One can still make up a fantasy game without any computer aid, and then check it, for example. Or you can rely on light aid, not checking every move. Or, if you are true computer geek, you can automatically generate games by programming an engine to choose 2nd or 3rd best moves (according to its evaluation function), or some weighted random distribution. There are all sorts of things one could do. As a non-professional chess player, I don’t do any of the really fancy things that one could do; I’m just pointing out that one could do them.
For instruction
Recently I came across a beautiful game by Steinitz in Bruce Pandolfini’s Chess Cafe column, in which he crushed the French Defense using his own variation against it (Steinitz-Sellman, Baltimore 1885). It turns out that I have been helping a student understand the French Defense and its thematic ideas, so I assigned this game for study. But I also immediately felt that I should have him understand that of course Black went wrong in various places, and playing the French Defense does not simply mean lying down and losing with an inactive light-squared Bishop. So I wanted to also give him a game in which it was Black who played thematically and well. But I wanted full control over the nature of the game, so I decided that instead of looking for some game that fit my parameters, I would invent one, with a bit of computer aid. This is the first time I’ve invented a fantasy game since my childhood! Maybe I should do this more often.
In the following game, I generated it simply by choosing “plausible” but poor moves by White, letting a chess engine generate reasonable-looking continuations (I did not always take the first choice, especially in light of its difficulty in fully understanding the closed positions in the French), and then myself interpreting the game in human-oriented positional and tactical terms. In some decision points I deliberately allowed the game to take particular directions.
Themes the game allowed me to illustrate:
The importance of the dark diagonal from a7 to g1 that White relinquished early in the opening. The danger of passive Bishop and Knight deployment by White in light of the opening configuration (with Queen on e2 and Bishop unable to get to e3) and the desire to castle King side (another illustrative game would show the dangers of castling Queen side given Black’s a6/b5/b4 plans). Black’s timely attack on White’s center using f6. Use of the half-open f file, including a thematic exchange sacrifice on f3 to break through. Black’s successful deployment of the “French” light Bishop to d7 and then on both wings to b5 and also back to e8 where it was ready to go to the King side.
Not bad for a mostly-generated game!
The main caveat I offer is that the computer is not useful in the opening unless you already know the long-term themes, because, for example, it does always realize how poor the “French” light Bishop is until too late; in many lines it wanted to play b5 and Bb7, which just goes along with White’s plan that Steinitz came up with in the 1800s! This could be the topic of another article.