Extreme thinking, by Michael A. Nielsen | Hacker News Reader

Extreme thinking, by Michael A. Nielsen

michaelnielsen.org

73 pointszkz10 comments

This is a great essay. The core tensions that he outlines for developing a sense of purpose really resonated for me:

   "I believe that the key to developing a strong sense of purpose and meaning is
    to balance three activities.
    o The first is development of a common understanding with a large group of people
      with whom one is later able to feel a common sense of community.  
    o The second is development of abilities which are not common to your community, 
      and which eventually give you the ability to make a unique contribution
      to your community.
    o The third is making a creative contribution to your community, 
      to something larger than yourself."

The second (abilities unique in your community) is hard, though one way is to learn them from another community where they are common, i.e. cross-disciplinary.

At the end of the first and second principles, I was reminded of thoughts by two other writers... who were mentioned in the immediately next passage (Kay and Covey, respectively). It seems like a clear nod to the audience.

But I'll mention my first thought anyway: abstract ideas that are difficult to reason about can be made easier to work with by finding a better representation. Hopefully one exists. Alan Kay compared multiplication with Roman numerals (in Roman times, only geniuses could do it) with today's positional numerals (in modern times, children can do it). We haven't got any smarter, he said, we've just changed our representation system. Inventing better representation systems is something important that we do as programmers. This was Kay's explanation of "point of view is worth 80 IQ points" that Nielsen mentioned.

Perhaps, in the context of teaching, this itself was an example of the communication technique where you set up a trigger for the audience - a striking image to evoke their own memory, or a puzzle to provoke their own reasoning - so that they take ownership of the message, and it makes more connections to their own network of pre-existing ideas. The "nod", or implicit punchline, is confirming for those who caught the reference; is completing for those who half caught it (filling a gap is another way for increasing memorableness); and just plain informing for those new to the topic. Or am I reading too much into this physicist's writing, and he has quantumed me out?

I myself have thought quite a bit about the better representation issue.

I once read somewhere that all human concepts are fundamentally associative in nature. If you see the halftone image of an apple, you don't have to work through pixel by pixel to figure out the concept, "apple". The concept apple with its related features naturally occurs to you.

Taking this further to learning, it would seem that we all go through a stage where our knowledge is of a deliberate, serial nature and once a certain degree of habituation is reached, the knowledge gets coded enough to be associative in nature. Consider solving problems in Mathematics - initially, we all have to consciously do a state-space search to figure out the right method to solve the problem. To expert problem solvers, however, the method to solve the problem becomes immediately apparent just from the patterns in the problem. Ditto for chess, etc.

Lets call knowledge that's serial in nature second-order knowledge and the knowledge that's associative, first-order knowledge. I think there's immense promise in investigating whether the transformation from second-order to first-order can be made systematic. I.e., consider learning as a new subject in itself. I am pretty sure that research has been done on this though.

It seems that a more appropriate title might have been "Effective Learning". It's an excellent essay.

How is the card game and the bar game the same? If a man is obviously old, he can drink alcohol or other kind of drink. In the case of cards, there is only an answer.

In the card game, the rule is that if a card has a vowel on one side, it must have an even number on the other side. Replace "the card has a vowel on one side" with "a person is drinking alcohol" and "it must have an even number on the other side" with "he or she must be over eighteen". Since a card with a vowel must have an even number on it, if a card has an odd number, one must check whether the other side has a vowel to know whether the card follows the rule. Similarly, if a person is under eighteen, one must check whether they are drinking alcohol to know whether they are following the rules.

If we refer to either "a card has a vowel" or "a person is drinking alcohol" as P and either "a card has an even number" or "a person is over eighteen" as Q, we can describe the four cards/people in both situations as:

1. P 2. ¬P 3. Q 4. ¬Q

The rule can be described as P implies Q. P implies Q is equivalent to "Q or ¬P". Therefore, in cases 2 and 3, P implies Q is known to be true without needing to know the value of the other proposition. In the remaining cases, you don't have enough information to know whether P implies Q. In case 1, P implies Q if and only if the value of Q is true when one checks it. In case 4, P implies Q is true if and only if the value of P is false when one checks it.

Sorry, but I still think that checking what a very old person is drinking is stupid, no matter how much notation you add to it.

Yes, that's the point. The bouncer doesn't need to check the old man, and the card player doesn't need to check the third card.

This is obvious when given the bar context, but less obvious when the same set of rules is taken out of any human context.

Because he is "obviously old" he does not need to be checked by the bouncer to ensure he satisfies the rule "people under 18 cannot drink".

Reply on news.ycombinator.com