Thursday, April 24, 2008

More about feedback

In the past I had already written a lot about feedback here. I think that everything that is said there still stands today.

If you read the comments on my previous post, then you can see that we found that feedback is the nec plus ultra method to improve your motorskills. Let's see if I can elaborate on this.

Knowledge is short lived in chess. It either fades away or it is transformed into procedural knowledge. To give an example: when you read that it is a good idea to occupy an open file with a rook, you will either do that or forget to do that. Once it is a habit, you will do it every time, even when it isn't necessary. If it doesn't become a habit you don't will do it, but when you teach children you will tell them still that they have to occupy an open line. If knowledge isn't transformed into skill you are left with the impotent ghost of knowledge. This illustrates i.m.n.s.h.o. a transfer problem as mentioned by Phaedrus.

The transformation of knowledge is helped by narratives. Why? Because it helps to make the knowledge conscious. Consciousness is paramount. The following factors play a role:
  • acquisition of knowledge
  • quality of the knowledge
  • transformation of knowledge into procedural memory. The forming of habits. With the aid of narratives.
  • assessment of the knowledge. Because procedural knowledge is only semi-intelligent, you need to know it's value. Otherwise you cannot determine when the knowledge must be overruled. This assessment is based on the feedback from experience. Again, conscious feedback will adjust procedural memory while automatic feedback makes that this process of adjustment grinds to a hold.
  • duration of validity of the knowledge. This only plays a role with opening knowledge.
From both Troyis and singing stems my estimation that it costs 22 hour or 2500 repetitions to establish the fully automatic application of knowledge. The transformation from knowledge into an effortless habit.

Let's take a closer look at Troyis.
While playing Troyis the feedback was rather low level and hardly conscious. It was simply based on trial and error. Yet the result was sufficient to score well at this test. The question is, can the amount of time I used for Troyis (about 22 hours) be diminished with better and more conscious feedback?
Possibly so. When I had taken some time to look better at the game of Troyis I could probably develop an efficient strategy to move through the maze. Trial and error simply doesn't lead to a sufficient optimized strategy.
It has to be said that the lazy way is attractive. After all, playing Troyis is no penalty. But for chess the lazy way is too time-consuming. There are too many skills to be mastered.

A factor that has to be taken into account is the difference in speed of the declaritive part of the mind, which is low, and the procedural part, which is fast.

Last friday at the club I played against an opponent who is rated 500 points below me. He has made a great progress the past years so maybe he is somewhat underrated. It is an elder guy, he doesn't like to think about the game nor to talk about it. He just plays a lot. He did see all tactics quite well. At a certain moment I made a severe positional error. I played a freeing move which in itself was a good idea, but I forgot to prepare the move. By just playing simple and logical moves his position became better and tactics came into play all by itself. At a certain moment he could win the exchange and I would be virtually lost. But he didn't see it and I could trade off a few pieces and the queens. I was a pawn behind with R+R+N against R+R+N. He had clearly no idea how to play this and I won the game in a simple way, despite being a pawn down.

This game illustrated me the following:
Feedback by means of just trial and error can make you a tactically competent player. Since the feedback is simple, you win a piece or not. But at the very moment he had to make a positional decision, he was lost. I just attacked a weak pawn and he couldn't resist to defend it. Because material is what tactics is about. If he had given up the pawn and had activated his rooks, he could have drawed, possibly. But this is something he cannot find by trial and error. And since he doesn't like to think, talk or read about chess he has cut off all other feedback.


  1. I agree with what you've said. I've won many games due to "Rooks to open files!" (Playing carefully so I can control them, or at the very least contest them and trade off the rooks) and when I mention the rule after the game it rarely comes as a surprise to opponents. It's like, "Oh, yeah, why didn't I think of that." and like you say that's the whole problem--it's instinctive for me and not something I have to try to remember (although, of course, I still have to calculate a sequence of moves that allows me to control or contest the files).

  2. warm greetings to you temposchlucker. i continue to read all your marvelous posts, and while it is honestly hard if not impossible for me to 'get my arms around what you are usually saying' from where i stand now, i recognize it as valuable and a great contribution to our community here.

    sometimes what we do for others is not so much in transmitting to them the specifics of what we are doing and how we do it, which is really--truly--a very personal thing, but rather the benefit is our embodying the higher steps and pointing the way to others. they, too, must travel the way alone, but get reminded of the flavor of what exists, and thereby motivated to want to get there one day...

    i have my own news, but this is not the place for it. :) even the slow can learn eventually.

    take care sir. lovely. dk

  3. I like the idea of getting only feedback about gross tactics by playing blitz. I think DK's strategy of meticulously playing higher-rated players in blitz may be a really good way to force yourself to learn tactics quickly in real games. Indeed, when I played my coach so he could evaluate my play, he was like "You are too good at opening and positional stuff given how bad you are at tactics." Ouch (he won every (fast) game we played due to simple tactics.

    His advice? Play a shitload of blitz (and he liked the Circles idea too). I've been doing that a little bit lately (in the sense that 100% of my 1-4 games a day are blitz). I'm lazy, though, not Fritzing them to see tactics both of us missed.

    But once you hit that plateau, it's probably harder to add more subtle positional evaluations to ones procedural knowledge base, since often you only see the outcome of playing crappy positional moves (i.e., losing, later tactics), but don't appreciate the strategic blunders that opened you up to such errors.

    Also, probably more complicated attack sequences and combinations are much harder to learn via the rudimentary feedback achieved in blitz.

    God I wish we had subjects to experiment on. Give one pool an hour of blitz a day, the other an hour of tactical puzzles a day, and each pool one slow game a day. Which will be better after a few months of this? Have them come together and play each other in slow games after two months to see.

  4. OTOH, what of all the people who suck at tactics but have a zillion games at ICC? Perhaps there needs to be a mindfulness, an effort after the fact to at least see the major tactical bits that were missed to cash in the tactical potential in those games.

  5. Blue,
    during the past few days I felt the whole time the friction of the point you bring up know. On the one hand you hardly can obtain knowledge without it being converted into procedural knowledge before you know it, you must almost make an effort to not convert it, on the other hand persons who know everything but apply nothing, and on the other hand persons who play blitz for breakfast but suck at tactics. So I'm sitting with 3 hands in my hair now.

    In my reasoning I proclaim that consciousness of feedback is paramount. But can that really cut down on the hours and #repetitions to master a motorskill? That feels somewhat counter-intuitive (=I'm biased)

  6. Hi, This is for Soapstone, but

    I played Wetzell (and lost) at the 2001 (or was it 2000?) U.S. Open in Framingham, MA. Anyway, I wanted to clarify something you said in your latest post (you don't accept anon. comments): "The flared, martini-like lip is supposed to represent the diminishing returns we get from learning near our capacity, with increased surface area exposed to evaporation."

    This is one reason for the tapered martini glass shape. The other reason, I believe, is because the number that is your rating has a nonlinear relationship to your ability/strength as a player. To raise your rating 50 points takes X amount of ability/knowledge (Wetzell's liquid); to raise your rating 100 points takes much more than 2*X increase in liquid. In other words, chess ratings are logarithmic. The martini shape represents this nonlinearity.

  7. Following up on the last point, we have TH (Tempo's hypothesis), that something must be conscious before going into procedural knowledge.

    But then we have:
    1. Chess scholars, who consciously know a great deal about the game except how to play (e.g., gap between high declarative and low procedural knowledge). So conscious knowledge is not sufficient for procedural. What else is needed? Experience? See number 2.

    2. Super experienced players, who have played tons of games but still are not good. This suggests simply playing a lot is not sufficient.

    So are 1 and 2, in combination, sufficient to gain procedural knowledge?

    Perhaps the problem is the quality of the experience. The person is rated low, plays low-rated players. If he spent more time going over master games, sort of like DK, perhaps that would be even better experience than playing himself.

    So let's add another group to our experiment--those who study master games during the day and then play a slow game at night.

    If only sleep weren't necessary. I am glad my blog is almost over, the urge to jump back into chess has been strong lately because of these discussions.

  8. It's important to be able to distinquish the difference between positions where following a learned priciple of chess is correct or not correct. Sometimes it's easy to blindly follow a rule such as rooks on open files when in a particular situation it may be better to allow the opponent to have that open file because you have something better elsewhere on the board.

    Too many times I've cost myself by doggedly sticking to some principle instead of looking deeper into what is happening in that given position.

    The question remains how can we understand the exceptions to rules and learn to see the difference.

  9. Blue,
    your comment is worth a new post.

    good point, I will try to take that into account in my new post.