I schedule planning time where the level of abstraction is proportional to the logarithm of the recurrence period, and it seems effective at pruning cached goals and sanity-checking my meta-goals. (However, it’s difficult to test because of the time scales involved and the fact that I can’t fork myself.)
Recently, I noticed that my general skills aren’t improving as fast as I’d like, so I decided to take advantage of compound interest[1] and created a parallel structure for working, learning, and meta-learning.
the level of abstraction is proportional to the logarithm of the recurrence period
This brings to my mind the idea of a complete n-ary tree (with n being the base of your logarithm), with the highest abstraction level at the root—if you spend equal time on each node, then you’ll portion time across levels as you described.
I found this amusing—I’m not sure I know of any generally meaningful meta-thinking levels beyond say, 2.
Yes, my approach is similar.
I schedule planning time where the level of abstraction is proportional to the logarithm of the recurrence period, and it seems effective at pruning cached goals and sanity-checking my meta-goals. (However, it’s difficult to test because of the time scales involved and the fact that I can’t fork myself.)
Recently, I noticed that my general skills aren’t improving as fast as I’d like, so I decided to take advantage of compound interest[1] and created a parallel structure for working, learning, and meta-learning.
Richard Hamming, “You and Your Research”
EDIT: Fixed link misparse.
This brings to my mind the idea of a complete n-ary tree (with n being the base of your logarithm), with the highest abstraction level at the root—if you spend equal time on each node, then you’ll portion time across levels as you described.
I found this amusing—I’m not sure I know of any generally meaningful meta-thinking levels beyond say, 2.