As I was working problems this morning, my mind kept returning to a game I played last night at FICS. My opponent forked me three times as I blundered my way into one untendable position after another. This made me realize I need to work on avoiding tactical blunders as much as I need to learn seizing tactical opportunities that present themselves. While this is two sides of the same coin, to some extent, as I work tactical problems here and elsewhere the problem sets seem to be more about finding unique tactical opportunities than merely making the best move in routine situations. Hidden within the thousands of moves we've each made, I suspect we each have a consistent set of situations, like avoiding a fork, where we each need improvement whatever our rating.
Browsing through the post on the pruning algorithim for the problem generator, I wondered whether or not it might be possible to customize problem sets from a collection of ones own games thus selecting actual situations where you personally blundered for review. The benefits of such problem sets seem self-evident to me.
I used a similar approach teaching high school algebra in reviewing for standardized tests. Using published sample questions from previous tests, I created problem packets by type. Each student worked the packets self-paced until he or she demonstrated personal proficiency by problem type. Then rather than continuing endless review they were able to move on. The approach worked so well my kids scored so much better than other classes we were suspected of cheating. We didn't. The technique worked, IMO, because students were able to focus their review on problem types they consistently missed.
While chess problems don't fit as neatly and simply into such categories, I wonder whether focusing on situations from our own games where we blunder may not yield similar accelerated results in chess training. Such situations are not necessarily very elegant tactical problems, but they are our own unique blunders. Whatever our individual rating, I suspect we each may have recurring situational themes, not necessarily tactical, where we have stumbled. Those are situations I'd like to focus on and learn to prevent.
I also used an approach similar to this re-learning backgammon. Using an analysis engine, GNUbg, I began collecting problem sets of situations where I blundered for manual review, i.e. generating my own problem set. Using that approach, I was able to reach an 1800 Elo rating in less than six months (in backgammon, a 2000 rating is rare). It was hard focused work limited to situations where I had previously blundered. But learning to avoid my own unique blunders, I was able to rapidly learn the game at a pace that surprised many expert players and teachers.
- Is it possible to do something like this with the problem generator?
- Is anyone already doing something like this?
- Does anyone else think there may be a benefit to such an approach in chess?
