## Friday, 1 June 2012

### Basic Tactics Revision Results

I decided that it was time to revise my basic tactics training, which used five of Dan Heisman’s 7-10 Basic Tactics Books:

John Bain’s Chess Tactics for Students
Al Woolum’s Chess Tactics Workbook
Jeff Coakley’s Winning Chess strategy for Kids
Dan Heisman’s Back to Basics Tactics
Bruce Pandolfini’s The Winning Way

This training is summarised in my previous articles: The Bain Experiment, The Woolum Experiment and The CHP Experiment.  In these experiments, I divided the problems into batches of equal difficulty (or as near equal difficulty as I was able to manage), and solved them repeatedly.  I found that repeatedly solving the earlier problem batches improved my performance on my first pass through the later batches.  I decided to revise these problems by solving them once, and got some very encouraging results.  Here is my old chart from the Bain Experiment, for my first pass through problem batches A+B, C+D and E+F:

Bain First Attempt

Overall, I scored 95%, at an average of 13 seconds per problem.  Here is the corresponding chart for my recent revision pass:

Bain Revision

Overall, I scored 98%, at an average of 7 seconds per problem.  My performance (9 months after my last revision) was much better than on my first attempt; but not as good as it was shortly after the previous 11 relatively closely spaced repetitions.  I learned Bain over a period of 6 months, so revision was due after about 6 months, and was a little overdue.  There is some indication that I improved slightly from one batch of problems to the next.  Here is my old chart from the Woolum Experiment, for my first pass through the problem batches A to F:

Woolum First Attempt

Overall, I scored 83%, at an average of 12 seconds per problem.  Here is the corresponding chart for my recent revision pass:

Woolum Revision

Overall, I scored 95%, in an average of 14 seconds per problem.  The most striking result here is steady the reduction in the percentage of problems that I failed to solve correctly within 30 seconds.  My revision of the earlier problem batches appears to have improved my performance on the later batches, particularly on the harder problems.  My performance (10 months after my last revision) was again better than on my first attempt; but not as good as it was shortly after my previous 8 relatively closely spaced repetitions. I learned Woolum over a period of 3 months, so revision was due after about 3 months, and well overdue.  Nonetheless, I studied other tactical problems in the meantime (which was my rationale for stopping revision at 3 months).

In the CHP Experiment, for Coakley, I exceeded my best performance on Bain from the outset, but did not subsequently improve.  On this basis, I decided to omit Coakley from my revision.  Here is my old chart for Heisman + Pandolfini:

Heisman+Pandolfini First Attempt

Overall, I scored 83%, at an average of 12 seconds per problem.  Here is the corresponding chart for my recent revision pass:

Heisman+Pandolfini Revision

Overall, I scored 95%, at an average of 12 seconds per problem.  The most striking result here is the steady improvement in the number of problems that I solved correctly in under 5 seconds.  My performance (10 months after my last revision) was again better than that on my first attempt; but not as good as it was shortly after my previous 8 relatively closely spaced repetitions.  I learned these problems over a period of 3 months, so revision was due after about 3 months, and well overdue.

It is very encouraging that a single revision pass has resulted in an improvement comparable to that resulting from several passes in my original experiments, for the harder problem sets.  Would solving other problems at the same level have had a similar effect?   I doubt it.  I expect that revising old problems is more time effective than learning new ones.

1. How did you keep track of your answers and your time? Did you use paper to write down your answers, and then just do an average time per batch (by dividing total time by # of q's)?

I'm going to try this myself but I'm planning on taking the positions out of PGNs, organizing them in a PDF, and printing. Then I was going to just write down my answers and do an average time for each batch.