Tuesday, February 12, 2008

Chess and the art of motorskill maintainance.

Phaedrus provided the following diagram which usually shows a clear correlation between rating and solving time. Michael Adams did this knighttour in 1.5 minutes while an 1800 player usually needs more than 20 minutes for it.

For this knighttour you have to move the knight before your minds eye from a1 -b1-c1 etc. h1-h2-f2-c2-a2- a3-a4 etc. all the way to a8. Visiting all the free squares in the right order. The black pawns and the squares they cover are tabu. The black pawns remain stationary.

I solved this problem in 3 minutes and 50 seconds, which equals to a 2600 player. When I tried to figure out why I could solve this problem so fast, I realized I had played a lot of Troyis. Which basically trains the skill to move a knight in a restricted area of random shape.
The motorskills I trained with Troyis provided the speed and the visualisation you need to solve this problem. And that is exactly what you get for free when you train your motorskills: speed and visualisation.

The motorskills you learn while playing Troyis are not frequently used during a chessgame, so you cannot expect that an improvement will effect your play by much. So the problem was to identify the motorskills that play an essential role in our chessgames. With the aid of Phaedrus I found 3 essential areas:

  • Target-consciousness. Scanning for potential targets.
  • What do you want for Christmas. Automatic positional considerations. How to improve your worst piece.
  • What does a piece do exactly. New: what does the piece do on its new square. Old: what did it do on the old square which it doesn't do anymore. Clearance: what square and lines are cleared by moving the piece and which pieces can make use of that.
Usually the amateur adresses these points with his conscious thinking, which is way too slow and error prone really. By transferring everything you can from your conscious thinking to your unconscious working complex motorskills you free up your thinking for more important tasks. Besides that, everything that is mastered with your motorskills is fast and you can visualise it. For free, without specialised speedtraining or visualisation exercises.

From a scientific article in one of my old posts:

In amateurs, "focal gamma bursts" were most prominently detected in the medial temporal lobe and in grandmasters researchers measured the bursts most often in their frontal and parietal cortices, parts of the brain linked with long-term memory and the ability to perform complex motor skills.

For some reason we always focussed on the long-term memory part and not on the complex motorskills.

These thoughts are the results of the complete overhaul of my ideas in the past weeks. I can imagine that I have forget to mention things. In case you find anything vague or inconsistent about this post please feel free to ask. I possibly left something out unintentionally.


  1. i am genuinely impressed.

    at the same time, i while very much inclined to total silence so as not to offend or feel better to say nothing (honestly, i dont wish to BUT), if you can do this that fast, and rightfully an accomplishment,

    why are you not better at blitz, even a 10 min game such as 2/12?

    yes, yes: i know: while most better players can play better blitz, not all stronger players (relative to class C =you) do in fact translate this.

    but this is not an easy exercise, and implies being nimble or adroit, would seem to allow you to play good, solid, rapid chess? no?

    in various measures, your efforts there are instead average and it doesn't truly make sense to me.

    hope that you dont mind.

    warmest, dk

  2. Transformation seems to be missing the point here. The speed with witch you do this knight exercise normaly has a strong correlation to elo. At least when it is done by players who never did this exercise before.

    The fact that Tempo can do this exercise in the same speed as players much stronger than him, suggests that his experience with Troyis has given him a knight handling ability that can be compared to IM's and GM's.

    Unfortunately for him, playing chess involves much more than handling the knight in a puzzle like this. But it seems to prove that you can reach a master level in chess related skills if you do the right exercises. If he were to find other skills that are as chessrelated as the knight handling practised with Troyis, he would probably boost his chessskills.

  3. DK,
    Phaedrus declared it very well. If a simple game like Troyis can improve your knighthandling to masterlevel, then it is a matter of identifying the right motorskills which do matter more and write a similar computerprogram a la Troyis to exercise them. That's why I'm toying around with Zillion.

  4. what?? 20 minutes for a 1800? that can't be right. it took me 2min 50s.

    I did play some troyis a year ago, but I don't know.. I don't think it made that big of a difference. I also trained KNB-mate until I could blitz it in 15-20 seconds, but that doesn't really have longer than 2-move knight paths...

    I really can't see how it could take any of us more than 5 minutes? I mean, we all have the basic 4-move paths down cold, don't we? after that, it's just autopilot.

  5. WW,
    yet it is true. The difference in speed is caused by if you execute this exercise with your conscious thinking which is very slow and error prone or with your automatic motorskills. A little Troyis can make all the difference. You seem to learn motorskills easy or you train usually in a way that adress motorskills. That can be the reason why you improve so rapidly in chess in comparison to the rest of us.

    Papa Polgar proved that any child can learn these motorskills easy at young age. Troyis proofs that adults can reachs the same results with specialized motorskill training.

    Besides that there is a difference in natural ability, as you and MdlM showed. The heavy thinkers usually are bad in chess since they prefer to think heavily in stead of trusting their unconscious automatic motorskills.

  6. See my first comment here. You are describing procedural memory. I think this is the right track.

  7. Oh, that's the post you linked to! :)

    As I said there, I look at playing chess like riding a bike, a huge, unwieldy, complicated, bicycle that takes a very long time to learn to ride. Or look at airplane pilots, how quickly they flick and check a zillion little switches. Chess is more complicated than flying a jumbo jet.

    Philosophers would call it the difference between "know how" and "know that." Know how, the ability to do something well. Know-that, the ability to conceptualize and explain and consciously entertain something. It is a key difference: the GMs have great know-how, but their know-that isn't always great, which makes them crappy teachers sometimes. Often the adult amateur who sucks has fantastic know-that, but awful know how. I have met quite a few people who can kick my ass at chess but I know-that more about chess than they do.

  8. Blue,
    we came pretty darn close in the past. But I couldn't transform it into a working exercise. There are two critical points:

    I did some calculating and found that for me about 22 hour is needed to master any complex motorskill. Which boils down to approx. 2500 repetitions.

    Secondly, the motorskill must play a definite role in winning a chessgame. That means it must be important and you you must need it quite frequently. The motorskills you learn with playing Troyis don't fit this bill. Manoeuvring with a knight in a limited space of random shape is used for regrouping your knight. Which never alone wins your game. It is all-important to identify the essential motorskills. In the post you see my bet.

    Unless both criteria are met, it will make no difference for your rating.

    Troyis gave me the idea where to look. The microdrills of DLM served the same purpose, training motorskills that is, but were far from ideal. He trained it every tournament, which was basically every other week.

  9. You are on the right track. I don't know about this 'zillions' stuff, but the way to get better at any procedural skill that takes actual skill is to do it with a tutor watching. E.g., swimming the best instruction is with a coach who films your stroke and helps you improve by pointing out what movements you should have made. Simply practicing is not enough, indeed may be counterproductive as you end up in local minima with a really crappy, but comfortable, stroke in your swim.

    So, just playing in tournaments isn't enough, it's also how you approach the games after the tournament. E.g., do you take the film to your coach for help, or just forget about it? (Luckily with chess we have many sources of coaching from blogs to Fritz to the best of all, a consistent mentor over an extension of time).

    I'm personally in a happy compacency stage right now. Having fun playing, not caring about rating, playing only blitz, and only when I feel like it.

  10. Hi. Check this out :-)

    The Caquetio Knight


  11. I toured a1 to h1, then h2 to a2, then a3 to h3, etc all the squares on the chessboard that neither contain nor were attacked by pawns.

    At 1am on my first try it took me 6 min, 30 sec. That supports that Troyis has improved the skill at which you and wormwood can move knights, but makes doubtful the claim that it takes an 1800 player 20 minutes to do the tour. Do you have a good source for it? Did anyone else (1600-2000) try it out?

  12. I just tried out Troyis--very nice, especially that it's free. :)

    I only made it to level 9. Given how it seems to have helped you guys, I'll probably play around with it as long as it's still entertaining.

  13. I played Troyis approximately 60 min over the past 10 days, reaching level 12 and a rank of #1690. How has this impacted my knight vision?

    Today I re-tested and completed the exercise in 5 min, 30 seconds. This seems to validate that the game has educational as well as entertainment value.

  14. LF,
    the application of the skills you learn with Troyis during a chessgame is fairly limited ofcourse. You learn to move a knight in a limited space without thinking. How often do you need that skill? But it gives you a good idea how to train complex motorskills. The real meat is to identify the motorskills that are crucial for a chessgame and then to train them. In general you need about 22 hours to learn a new motorskill or 2500 repetitions.

    Phaedrus is my source for the diagram. He is a chesstrainer himself.