You and Your Local Gradient

expanded comment
brief note on my approach / on being a budget nat friedman

  1. Most people don't get jobs through the online application portal
  2. Trying to mastermind career plans is silly.
  3. Instead being reactive to local opportunities is a much more useful skill than trying to aim for global end points.
  4. You can always be doing things and producing value. Life is a video game but the map and scoring system are hard to understand. There is of course value to deep thinking and planning in solitude but that's probably not your comparative advantage. Network effects are easier to implement and produce results faster.
  5. Two reasons: outside of medicine or law, most people don't get into good positions by writing a series of steps on paper and following each step to get it directly. That rarely works. Your plans often aren't that good. Instead, you can exploit local opportunities more, and produce value for people you know, as well as trying to meet people that they know. In the cases people do make successful direct plans, there's usually some phase of consulting people who've done the same before.
  6. Another empirical observation: our knowledge is limited so better exploration algorithms tend to be more valuable than knowing anything in particular. Akin to a magic key, their value is the fruit of the paths they will open to you.
  7. Maximizing your local gradient is easy. Doing favors for people can be as simple as giving them a contact, you probably know a lot of people you could help this way almost instantly. If you know what they're trying to do, all the better. So try and figure out their goals.
  8. Letting people in your local environment know what you're trying to do and seeing what they can do for you is often a better approach than cold emailing for advice, though that is still useful.
  9. Most people think of their goals and come up with some ideal net worth figure, but that solves surprisingly few problems. You can't buy friends or fun hobbies or happiness. Those require discipline and being a good person. Happiness is a skill, just like tennis or knitting. You can't purchase tennis ability, though money can help buy good classes.
  10. You don't know what good outcomes are, and being as rapid at steepening your local gradients as possible is a better approach to search than trying to derive the optima of your brain's utility function yourself (if such a thing even exists).
  11. I've thought about my ideal life too and it's just this. Knowing as many people as I can and working on interesting problems and going on quests with them. I want to be like Tyler Cowen or Nat Friedman. You don't need much capital to do half of what they do! You could literally do that right now, on smaller scales, and all the time.
  12. View this in information theoretic terms. A stupid Fermi estimate suggests that if everyone knows ~150 unique people, then log base 150 of 8 billion is 5. You are fewer skips away from everyone in earth than we are generations away from when Planck's original formulation of quantum theory.
  13. Maximize your value to your first order relationships, frequently helping them with what they want. Let them know what you want and by the virtue of their reciprocity and your value being something they can offer to other people, you will steepen your gradient.
  14. If it's surprising to you that you are 5 people away from everyone, then it should update you towards insanely good outcomes being the product of a few good decisions. Harder to frame in math terms, since decisions have varying chance of success and aren't an undirected graph like social networks are, but how many successful actions or good outcomes are you from a great outcome? It may be very few, though the path is unclear.
  15. Let's think in terms of percent and constrain the situation. If a system you implement has a 10% success rate at a "good outcome on a road to a great outcome" and you only need e.g. 5 good outcomes to get a great one, then do this math later, doofus. Let's imagine your system allows 2 input per day. 100 inputs will be 50 days. Roughly two months!

Simplified:

  1. Know the goals of others around you.
  2. Help them however you can. Be a force multiplier on their agency.
  3. Let them know your goals.
  4. See if they can help you in any way.
  5. Their willingness to help & your value to their friends will enable chain reactions to occur, thereby letting you blitz through the action space to good outcomes.