Sunday, October 15, 2006

Adjusting plans

After a major insight it is of course necessary to change plans.
As Blue Devil said in his last comment I have a problem that should be easy to correct.
A doctor of the dutch AMC (Amsterdam Medical Center), a meritorious chessplayer himself, found endorfines in the neural system and (nor)adrenalin in the paraventricular hypothalamus of chessplayers who died during a game. He said the time trouble-addicted died in fact by an overdose. His assitant added: "you sniff, you smoke, you booze, you play chess, it's all the same."
Asked about the same he said: "forms of addiction, and afraid of kick off symptoms. It's all very nice but in the end it kills you."
I don't believe I'm addicted to time trouble so kick off symptoms shouldn't be the problem.
But I have to admit I'm addicted to heavy thinking.

There are some difficulties though.
The first is the problem of awareness. If I'm simply NOT AWARE of the flow of time, it is not easy to intervene.
Last game I was very alert of my use of time. I was indeed aware that I was using too much time and EVEN THEN it was impossible to intervene.
An easy plan did not "pop up" immediately but had to be created with effort.
So that indicates the direction of the solution. I decided to pick up the Personal Chesstrainer again. PCT has a strategical module with 720 positional problems. Starting out with simple one-movers. "occupy the file", "centralize the knight" and the like.
Let's see if this helps to solve the problem of time consuming in quiet positions.

For completeness I have to say I stopped my endgame study with PCT. "I stopped" sounds rather active and deliberate. It was more "it vanished due to lack of energy". Studying endgames is a typical intellectual decision, not supported by emotion. Lack of emotion means lack of energy. This is just not the right moment to study endgames. If I solve my middlegame problems I will reach an endgame much more often. That will help endgame study. For now I focus at simple middlegame positional moves.

Within the next week I will reach the 70,000 mark at CTS. That means that I have repeated the problem window of 10,000 consecutive problems 7 times at average. I have not decided yet what to do after that. I have looked at problems that are rated 1900. Those still baffle me. I'm experimenting with doing the same problem window again, but now focussing on an accuracy of 95% in stead of 80%.


  1. dear tempo, i love this post. its wonderfull! thank you. since i wrote a long private note to likeForests the other night, at (im still gently nudging him for his private email, as i already have yours, blueDevils, samuraiChess) and he and i have many, many good friendly conversations, i am reproducing it here, since it is appropos the whole seriousness or steap slope part of 'percentage success at CTS' and might add to the dialogue.

    i am very most sincerely delighted to see you 'doing percentage success'. it is not the 'end all' or 'alpha' and 'omega' but it sure does approximate 'real chess'.


    without further ado, i copy it here with only a few tiny edits:

    hello like....
    sorry to be so out of touch with you (and samurai and takchess as well, but most notably you), but i work 14 of 16 days once a month, and to start 10 of 11 days.

    it sounds horrid (it is), but i still only work five days per week, but when my term starts, i am off sat/sun, then off the next week only sun then the end of the week fri...

    then it flips over: im off wed/thr, then mon/tue, then sat/sun, what dktransform had when HE wrote all those posts and did CTS...

    now my good friend chessWaste is 1485.01 elo, 93.4806 tonight, or six more correct, and will be 93.502, and so post '94'. i am at 22 to takchess 21st.

    it appears tempo created slowTempo, but he is 1353 at 95.0%. as i know tempo well, im sure he is trying some crazy thing like 400 correct in a row, as if you check his activity, he takes some thirty or forty secs on rare occassion, so must be trying for 100%. [this is so outstanding!]

    ... the reason i write tonight:

    i was 95.0 % for my session last night. if i am 94% i feel very bad... i pushed and pushed and couldnt get it.

    you know what? if you miss ONE single problem, you have to get the next 24 right just to establish the point you were at when you got the error!

    i was at 6/112=118 and was shooting for 5/120=125 [when i MADE ONE SINGLE ERROR]. this effort was in order to move the session up to 96.0%--my new requirement [or criterion of success]. last night, i hit 95 many times, and tried to push, but kept getting errors, so had to do a LOT of problems to get to 95.00 to finish: 8/152=160 @ 95.%.

    so tonight, i had to do the next 0/32=32 = 100% to finish 6/144=150 @ 96%. id do six more but cannot risk being overstimulated and stay up... to bed early today please!

    so one bad problem and you need the next 24 to get even!

    at 80% if you miss one, then you have to get 9 of the next 10 to get back to parity

    at 88% if you miss one, then you have to do 15 of the next 16 correctly to get back to parity [0/8=8 to get even then + 1/7=8 sum: 1/15=16, continues the rates at its norm].

    at 96% if you miss one, then you have to do 25 of the next 25 correctly to get back to parity [when you miss, at the rate of 2/48=50, 1/24=25, you have to make it back, so you do the next 25 right THEN 24 of the next 25, or 49/50 or 98% to continue the next trance]....

    it gets harder and harder and harder exponentially as you raise...

    as you get tired, you miss, then must do 24 correct THEN another 24 with one miss, and you cannot get back in that session [rate at 96% in this case...]

    not to be too silly, but isnt 93.5% at 1485 better than 96.2% at 1319???? [tempo, when i wrote this, you were more like 95.0% or 94.1% and 1356 or so, the lofty 96.2% at any rating is probably better than 93.5 even at 1485].

    i always told these guys, you all say you consider a long attempt in secs or a minute to be failed problem, but you try it?

    now mouse and tempo are all doing it, following you and me. yes. it was already there to take, but i grabbed it, and you were there too and it is a shift in user perception, if you ask me.


  2. DK
    Now I have (almost) done the problem window of 10,000 problems 7 times, I have to ask myself where to go. There are 2 main directions. Rating and accuracy.
    The past year I concentrated on rating. Not worrying about my accuracy, which hoovered around 80% without special effort.
    Now I ask myself, how well do I know those 10,000 problems? In other words: are these patterns committed to my long term procedural memory?

    Typically CTS presents to me problems I can solve within 10 seconds at average and I have to learn to solve them within 3 seconds to get the full rating award.
    If you compare it with learning a language, I could say: I learned 10,000 words, it costs me 3 seconds at average to retrieve a word and in 4 of the 5 cases it is the correct word. How would we value it when somebody would speak English in such a way?

    3 seconds for a word to arise is extreme slow, while 1 word wrong per 5 words is extreme much. Such person we would call a very bad speaker.

    If I reformulate my question, it would be: will I focus on new words (=higher rating), or on accuracy and speed.
    In the case of a language I would answer: I have to improve my accuracy and speed first. That would be the prove that it is committed to LTM.

    To do so, I will have to focus on the problem window I allready repeated 7 times, to become even more familiar with it. Only if the answers on these simple problems flow with an accuracy and speed that is comparable with your native language you can say you master them.

    So that is what I'm gonna try. It's an experiment, so I might change my mind when insight grows.
    I go for an accuracy of 95.0% minimum. This means I try to crank up my accuray as much above 95.0% as can be easely done and then I go for speed, untill my accuracy drops below 95.0%. Then I start a new cycle of improvement of accuracy etc..

    That's the idea.

  3. i love this. i absolutely love this discussion and idea.

    thank you.

  4. DK,

    I don't want to be overly critical, but I do prefer people to be accurate with some things. You seem to really like the term "exponential" even when it's not accurate.

    If you are going for a percentage correct of p, then each wrong answer must be accompanied by p/(1-p) right answers. This does increase very rapidly as p increases (and of course is divergent as p -> 1), but it is slower than exponential.

  5. np. thank you loomis. i knew that exponential was not the right term, so ok for an email but not good for a public note. anything you say always is appreciated, and know you to be fair.

    but you get the idea nevertheless--that any miss in the series as the series rate is higher becomes progressively harder and harder, more and more daunting to recover errors from--not just numerically, but experientially if not physically sustained.

    in my session completed minutes ago, IF i got one single problem wrong in 25, id loose ground off my goal of achieving 96%. but again, thank you.

    on a lesser note, at CT-Art 3.0, i am at problem 608 at level three to almost finish that nasty set, so eager to pick up where samuraiChess left off at the start of level four.

    CT-Art is the alter that i worship at now, then secondarily getting chessWaste to 94.5 =>firegarden: 1491.1 today at 93.54%, so a long row to hoe indeed.


  6. Tempo, this inspired me to post on time management. What do you do for that aspect of the game?