Wednesday, March 02, 2016

How fast is fast?

When you do Troyis on a daily bases for, say, two hours per day, then you will encounter a plateau in about two weeks. It is next to impossible to break that barrier. I have played it for about a month, in the past.

Troyis learns you to move a knight around in a confined space. And although I plateaued after about two weeks, my knight handling had improved so much, that I scored like a 2600 grandmaster in this test.

It might well be possible to get better at Troyis, but there is no reason for it. If I take my time and analyse the game, I can come up with a few strategies to move more efficient. But that is in a sense a rather artificial way to make progress. You don't break the boundaries, but you shift them. Which in it self is perfectly fine, of course.

From that point of view, it doesn't make sense to push your limits when you reach a plateau. 98% of your energy will be spilled. An educated guess would be that you reach a plateau after 20-30 hours of training in any chess tactics task. In other words, it doesn't make much sense to improve until you plateau, and from there try to become, say, twice as fast as you are, by rabid exercising. Before you know it, you feel banned to the salt mines.

It makes much more sense, to identify the underlying subtasks, and train them in isolation. Until you reach a plateau for that very subtask.

Identifying the subtasks is no sinecure. If you compare for instance FAES easy (Find All Escape Squares) with FAES-hard, it feels like two complete different exercises, albeit the idea behind them is the same. That can only be when FAES-h has a few extra underlying subtasks, which FAES-e has not.

So for me, plateauing is where the exercise ends, not where it begins. And when the exercise ends, the search for underlying subtasks begins.


5 comments:

  1. Tempos said: "So for me, plateauing is where the exercise ends, not where it begins. And when the exercise ends, the search for underlying subtasks begins."

    Yes that the idea of this method, but after improving in the subtasks ( or related tasks? ) we get back the the original exercise and break our former plateau ( ceiling ). That does work with some board vision exercises and we have to find a method to make it work in tactics. I think i see a progress in m1-e thanks to FAES-h.

    In general it is beneficial to have a clear signal: in our case of King hunt pins are a little bit like dirt. I should be better to start to learn the clear signal and then to improve the dirt tolerance

    I think its better to start with pseudolegal moves if there are pins which cant be "erased".. just to keep it simple.. and then after improving return to legal moves.

    The reduction of mobility is an important aspect in chess, but that should be an extra chapter in the book of improvment ( i think )

    ReplyDelete
  2. Tempo said "From that point of view, it doesn't make sense to push your limits when you reach a plateau. 98% of your energy will be spilled. An educated guess would be that you reach a plateau after 20-30 hours of training in any chess tactics task. In other words, it doesn't make much sense to improve until you plateau, and from there try to become, say, twice as fast as you are, by rabid exercising. Before you know it, you feel banned to the salt mines."

    At the start of Board vision or tactic exercises we have a initial phase of "quick" improvment. Thats the phase where you get used to : The GUI, the type of positions chosen by the puzzle designer, the task itself.. . As long as you still have to think about what you have to do at this exercise you waste time. After this initial phase i experienced that either the improvement stops, or there is still some, but maybe small improvement until we reach a plateau again. As more easy und uniform the puzzles as more likely there was an improvement after the initial phase.
    You are right that after 20 hours with one of these exercises usually we reached the final plateau.. but this plateau can be broken by other exercises.
    I did reach my high values in attacker and defender at chessgym because i made the related exercises at fritz and some other exercises like
    chess-minefields, chess-fork-trainer, Maurice Ashleys Skill Builder, Chess for zebras at Lukas chess, asf.
    So i guess that a perfect training at boardvision is something like a training of say 3 exercises parallel for say 20? hours and then a switch of the exercises. And at the end to restart at the beginning.

    ReplyDelete
  3. My stats at chessgym:

    1) Defenders (Played >= 20) --> 4th place --> 84.38 (2,317) 6h 22m 19s
    2) Attackers (Played >= 20) --> 5th place --> 50.11 (4,900) 14h 45m 02s

    It looks like I am too weak at these excercises. I can barely score 50-55 at Attackers and about 70-75 at defenders.

    I am not joking, but I need to practice LEGAL moves, attackers and defenders. Unless I score 80-85 at BOTH of these without a minimum effort - will not be able to break other subtasks.

    Tempo: would you be able to dedicate a few articles (posts) to the problem of attacked and defended squares? I am especially interested how to list them ALL and how NOT to miss any of them. Any ideas? I am more and more frustrated. It reminds me MDLM practice with the simple goal of surviving (not make a success).

    ReplyDelete
  4. @Tomasz
    You did already improve and you where already at a high level
    There are natural limits to these exercises, for example you cant get quicker than the signals in your Brain.

    So what do you want to improve now: M1-e or m1-h?

    I try to improve m1-e and i will use FAES!!, FAC!,all board vision exercises of Fritz, find all moves at lukas chess and a spaced repetition set at CT, while continuing m1-e ( new skills need to be embedded ). An other promising skill might be to improve in position memory.

    ReplyDelete
  5. I want to improve at M1h as I feel I have already improved at M1e to 85-90% of my abilities (potential).

    I need better methods of looking at the board as I even miss some checks at "FAC"! I have to work out how to be sure if I found all checks, all attacked pieces and all defended ones. After I improve these tasks - I would be able to move on with others (at present "unimprovable").

    ReplyDelete