Wednesday, February 19, 2014

Al Woolum - Chess Tactics Workbook Revision 1 Complete

Based on some training data from Bright Knight, this book is the next logical step difficulty wise after Bain.  It contains about 60% checkmate problems, many trivial one-move problems and several more difficult two and three move problems.  A few of the problems aren't usually valid in terms of problems since they call for things like en passant captures and castling.  Also close to 10% of the problems have some errata listed.  All of that aside, it's a good problem book for your money and is definitely more challenging than Bain.  On to my results.

As expected, I significantly improved at problems that I had already seen, so I won't list any charts for that.  For comparisons sake, after my 6th pass I was able to complete 74.4% of the problems in under 15 seconds vs 86.3% for Bain.  And just like Bright Knight, I was able to improve on problems I hadn't seen before also.

Here's the chart showing the percentage of problems I completed in under 15 seconds for each pass.

As you can see, I was able to improve from 30-38% on the first pass of the first 3 sets to 50% on the first pass of the last 3 sets.  My improvement in subsequent sets seems to have leveled off after the 4th set.

Like in my Bain revision, I identified 182 problems that I wasn't able to complete in under 15 seconds at least twice in my last three passes.  I've broken these up into two equal sets of 91 and will work them like they are a new problem book.  I'm starting to feel like Chess Hero is either a better tactics trainer than Anki or just better at getting accurate times.  After 3 passes through my 1st supplemental set, I'm already at a 92% solve rate in under 15 seconds.  And these were problems I struggled mightily with before.

My next tactics set will be Dan Heisman's - Back to Basics: Tactics and it will be done entirely in Chess Hero, although there are a couple problems that aren't really meant for solving as a pure tactical puzzle graded by the computer.  I am scheduled to start that training set 2/26/2014.

The Woolum sets were 132 problems each, and I'm feeling like that might be too much when trying to learn tactics patterns.  The ~65 problems per set in Bain was much easier to deal with.  Going forward, I think I am going to try to keeps the sets under 80, even if it would take me twice as long to do a complete revision.  Also complicating things was converting all 792 problems in Woolum into pgn format.  My method up to this point had been to load a pgn into Chessbase and copy diagrams into Anki for flashcard use.  That proved very time consuming for Woolum since no pgn file already exists for this book.  I'm also going to prefer books with pgns out there somewhere to ones without.  I don't consider it unethical in the least to download a pgn version of the book when I've already bought a copy as well.  It's just making my use of the book more suitable for my needs.  I guess this is making my conversion to doing software based tactical training all the more likely, but I like the old-fashioned progression through increasingly difficult chess books.


  1. It is not suprising that you perform quicker at problems you have seen before. It is interesting that you perform better at problems you never seen before. Buut they are still in (/from) the same book and so they are related to the other problems in that book ( same writer ~ = same ideas ).

    The questions of all questions ;) : Are you now better in other/common tactical problems?

    The best available test system i know, is the blitz mode at chesstempo. After 1000 maybe 2000 attempts the rating will be stable there.
    Then you can test how much your improvemet in tactic-skill realy is.
    Repetition has its drawback, the "patternrecognition" might be wrong. You see the same problem with the same solution and the patter to recognise the problem might not be "realy" related to the solution at all.
    But without ANY repetition of ANY type learning is impossible too...
    I think a training with a bigger set of easy problems solved with a very high score should be a better training ( at CTS = Chess Tactics Trainer)

  2. I agree that it's possible that I'm getting better at problems that are being presented by a specific author and those same patterns in another book might be missed. But then again, are the ideas of forks or discovered checks specific to an author? Those ideas are pretty universal.

    20 years ago when I was playing chess scholastically, I was a USCF 1000 player. Which means I was dropping material left and right and not taking advantage of all the material that was offered to me. The other night I beat a USCF 1700 player in 35 moves. I've been playing chess again for 3 months and at least 90% of my time spent has been with tactics. I'm able to see threatened tactics and avoid them and take advantage of them when offered. Not all of the time, but a lot more than I used to.

    The idea of pattern recognition isn't to be 100% right in all cases and play a move without thinking. It's to realize that a possible tactic exists in the position in front of you and to verify that your intuition is correct. If you don't know about Greco's sacrifice, you might not even analyze a move like Bxh7+. Or you might analyze it and spend 10 minutes trying to figure out the idea. But if you know the pattern and the idea behind it already, you can spend a lot less time than 10 minutes figuring out if it's a move you want to play or not.

    To me, stored patterns are used to suggest a set of good moves or prevent bad moves in a given position. A chess intuition that points you in the right direction. It might not be the right move in a given position, but your set of plausible candidate moves is increased. It's only a "drawback" if it's a blitz game and you play the move without verification.

  3. I made my own experiences with tactics training and the things are often not that easy. For example: i made a special "easy mate problems" training. result was : i did get better(=quicker) at mate-problems but worse in other problems. Same would happen with any other tactical theme. To find forks you need to find 2 weaknesses ( = not overprotected pieces ) wich can be forked by one of your own pieces. To find matingpatterns yo need to look for weaknesses in the opponents kingsafty. Depending what you are looking for first ! , one or the other will be (seemingly) "better".

    And: what is better? To repeat 1 example of "fork" 100 times or 100 examples of "fork" 1 time? The repetition of 1 example is only necessary/usefull if you cant find enough different examples of the same or close related pattern.

    CT-Art is creating "artificial" examples by using the same example but with opopsite colors and/or by mirroring the examples to the queeenside

    Extreme simple Example

    8/8/8/8/8/2R3K1/8/6k1 w - -
    6K1/2r5/6k1/8/8/8/8/8 b - -
    8/8/8/8/8/1K3R2/8/1k6 w - -
    1K6/8/1k3r2/8/8/8/8/8 b - -

    By repeating only one of these (4) puzzles you dont get necessary much quicker in the other (3) versions of the exact same idea/pattern, but what is such an "improvement" good for? You may get trained to spot : if the white King is at b8 and then move the rook ( no matter where the black king is and / or if other pieces are at the board too )