Wednesday, February 02, 2005

Working on a plan.

I’m going to use Don’s reply to yesterday’s blog as an opportunity to expand on what I was trying to convey. Don raises some interesting points about calculation muscle versus
pattern recognition. So I thought I try to clarify my beef with the 7-circles.

When I sit down to a real game I am seeing more tactics than I ever saw in the past, it might be safer to say that I can't turn the calculation muscle off. I'm flooded with tactics, I start looking at each little move, wondering what if this, then what if that? I haven't lost a single game due to a tactical oversight. But more in part to having to make a somewhat seemingly safe move which turns out to be positional weakness at a later stage of the game because of time pressure. But that is another problem. :)

I knew that I had glazed over the muscle part in Circle 1 and placed the focus on exercising it in Circle 2.
For circle 2 I spent almost the full amount of time, not because I couldn't find the answer, but because of double checking my answer to be on the safe side. I would find the key move fairly quickly then I would look over possibilities until I thought I saw all the "surprise" moves that could be played. Like J’adoube said don’t expect your opponent to play the “best” line.

Where my gripe with the program lies is in the amount of time allowed for calculation of the harder problems. There are more candidate moves and more branches to examine. In the introduction to his book "Excelling at Chess Calculation" Jacob Aargaard gives this position.

White has a King at a5, Bishop at a4, Knight at d7, and two pawns on a7 and b5 respectively.
Black has his King on b7, Rook on a8, Bishop on h4, and a single pawn on f3. This example is "White to play and draw".

After giving many lines of calculation Aagaard states
that he imagines the average GM between 2500-2600 ELO would spend 10-15 to make the right choice and would occasionally fail.

If the "average" GM is going to spend this much time on a 9 piece endgame how much time is enough for the average C player in a 20 piece middle game?

Of course there isn't an answer, but my point is shortening the amount of time with each consecutive pass on problems that we have faced on numerous occasions just so you can cram in more problems in the same amount of time doesn't make sense. I think the focus should be placed on accuracy while using a constant amount of time through out. (Work on getting things right first, and the speed will come to you.)

The improvement could then be easily gauged by the number of correct problems answered over the same amount of time. X amount of minutes per problem, and X amount of total time. You could allow 10 minutes per problem max and 1 hour total. (Yes I know that is just 6 problems.) Obviously we don’t have to spend 10 minutes looking at the level 10 problems, so you would just answer the problem and then move along to the next. At the end of the hour record your score, and play over any missed problems. You could work through the entire set of problems using this method. But I would prefer to limit the amount of problems to say 100 per set as opposed to 1,000. I like the idea of drilling smaller sections as opposed to volume work. This allows you to actually complete something and gain a sense of accomplishment. Volume work is like going on a long trip without a map or stopping to regain your bearings and then being surprised that you are lost.

Placing all of my eggs in one basket may be ok every once and awhile. But I seriously believe I need to apply what I have learned to real games. This feedback allows me to make adjustments based off of my game analysis. Plus it keeps things fresh.

I’m still a firm believer in using mates as the first theme of the exercises, because like I said the variations are pure with limited lines of calculation. I want to focus on being able to see one variation accurately, and then after mastering the technique of one line first start branching out to multiple lines of calculation.

Anyway I have really enjoyed the dialog that has come about from sharing our opinions about training technique. It really has gotten me to start focusing on ways of trying to improve, along with forcing me to create a program that is fun as well as balanced, tailored to suit my individual needs.
Ok off to check my mail.


Anonymous said...

De La Maza in his article says you can have 200 problems you have a different circles.Check out his article.

Don Q. said...

To the above commenter:
I realize you are trying to be helpful(which is always appreciated), but as Sancho has been working on this program for 90+ days, he has probably read the article.

To Sancho:
To echo a point you made, it is ok to "place all your eggs in one basket once in a while". The obscene focus on tactics in the de la Maza program is OK largely because it is temporary IMHO. Tactics should always be a part of the training diet, but it would be unhelpful to focus on it exclusively forever.