Osmosis learning: a crucial consideration for the craft

bytoonalfrink7mo10th Jul 20183 comments


I've been toying with a new approach to self-improvement that is quite different from what we've been doing at LW.

It hinges on the idea that the great majority of one's knowledge is implicit, and that most learning happens through osmosis (being near someone intelligent) rather than verbal communication.

If this is true, then the best strategy for increasing one's intelligence (in some domain) is to hang out with people that are more intelligent than you (in that domain).

So if you want to be mentally healthy, live at a Zen temple. If you want to be a good researcher, work with top researchers. If you want to be rich, hang out with rich people.

I think many people already implicitly know this. Lots of people would love to hang out with a billionaire. The naive hypothesis is that they're hoping to get a piece of their wealth, but perhaps they're hoping to get a piece of their intelligence.

This leads to competition to associate with intelligent people. Since market forces apply, ceterus paribus, the cost of associating with someone is correlated with their expected intelligence according to the average person.

So if this picture is true, then the best way to get ahead is to 1) be better at identifying intelligence, so that you can get a cheap deal for a big upgrade, 2) setting up an environment in which learning by osmosis is faster and 3) this learning happens asymmetrically.

For further thinking about this approach, I'd like to suggest three areas of inquiry:

  • Intelligence shopping: how do we identify individuals and communities that are more intelligent than us (in some relevant domain), but easy to approach?
  • A theory of osmosis: under what (social, psychological) conditions does osmosis occur with the highest bandwidth? My first guess is that it would include safety, attention, intimacy and trust. How much is this related to empathy? I don't know.
  • A theory of asymmetry: if two people exchange knowledge subsymbolically, how do we make sure that the more intelligent knowledge gets preference upon disagreement? This question seems to be the least tractable to me personally. i.e. there have been numerous times where I met someone's emotions by copying them, only later realising I should have met them with a smile instead.

And bonus:

  • For which domains can we expect our movement to be strictly better than any other movement out there? In which domains (of implicit knowledge) are we doing so bad that we would do well to be humble and apply osmosis learning?