Link Search Menu Expand Document
Max Learning FI

Notes | Tutorial 42

Date: 2020-09-21
  1. re flux and goals 26min
  2. speedrunning 38min
  3. reach - patio 59 min
  4. overreaching
    1. similar projects to grammar stuff
      1. how was grammar diff to things I’ve done before
    2. max and Blockchain
    3. OR

holding back b/c improving

ongoing problem – maybe forever problem. not a problem that goes away

  • comparative advantage between work stuff and public FI stuff

plan for ongoing learning curve; whatever I make in next year can be done better in 5 years

re flux and goals 26min

  • wholistic: in terms of world, leverage points, etc. best path doesn’t need to be direct.

todo: write about this

speedrunning 38min

  • repetitive stuff -> automatic -> can learn and think at the same time

  • persistence - debugging 49min

reach - patio 59 min

https://twitter.com/patio11/status/1307811149128163328

overreaching

  • error rate
  • not managing / exceeding
  • persistent and not able to solve a problem
  • has to do with the process by which you encounter problems and attitude with solving them / approach to solving them

ET: error rate > error correction rate

What errors exist/possible depend on your goal.

Can still make errors like doing bad exploration (or wrong area of study, maybe), get bad overview.

more meta goal: test out exploring, gauge effectiveness – can do that even if full exploration is beyond you atm

  • can go meta to do this

Some things are probably beyond you anyway, so better to leave alone till you have more basics / foundations down.

There are some ways to engage with end goal near the beginning, can get some info

“going meta” changes the goal of the activity to a different ~level of emergence, sorta. like “build planes” can become “go to work with a guy who builds planes and see if you like it”

  • motivation for OR stuff
    • having discussions, someone makes 1 or 2 errors
      • you talk about those
        • they respond with 1 or 2 more errors
          • etc
    • never get back to original topic b/c they’re not resolving errors
      • economics stuff

84min autopilot means you can work on problems like tilt much more easily b/c you can isolate it. not as easy to work on tilt in social situations, for example, b/c there’s lots more factors.

sort of like minimal test case; fewer extraneous details. overlapping ~minimal cases help to; allows for better isolation.

  • 89 min; inverse yesno / bifurcation / goals / something

big part of OR: not enough consideration of prereqs or planning

  • what’s the plan
  • what’s it require
  • do I expect to succeed
  • can’t achieve anything to
  • some goals beyond resources (including skills, time, ppl to help you), not just money and land and whatever
  • any prereq is a resource; trying to get some knowledge
    • any requirement for making the project work is a resource
  • fully explore what’s required to make the project work, then compare to what you have available
  • doing to much can lower both success rate + abs(output)
    • sometimes you could add resources (e.g. buy another computer)
    • but being over capacity in other areas is overall harmful to output
  • ppl sometimes start projects that have prereqs that take years (e.g. physics degree)

similar projects to grammar stuff

  • learning from some kinda course work - learn from text
  • reading things and sentence analysis
  • using a dictionary to figure out what something means
  • jigsaw puzzles / shapes together
  • complex / modular from simple things
  • classification of things and general relationships
  • breaking down into subproblems
  • learning to automate classification to allow more complex constructions

from ET

  • multi day
  • talk with ppl
  • share work
  • self judge/mark
  • learn from videos
  • learn from tutor

how was grammar diff to things I’ve done before

  • grammar
  • streamed?

want as few different things as possible

if lots of different things, do projects that only do a few differences, build up to large project

if something goes wrong in small project, you can isolate and work on the problem

ET: like small games -> easy to master

If you don’t do small things you can be fooled by like the union of problems -> has a lower ceiling than the actual space of results do.

max and Blockchain

  • written like 4 micro-minimal chains

OR

Sign of OR: if you can’t break down into parts

another sign: surprise

children start out pretty okay (they don’t try to exceed capacities, don’t want to); pushed into overreaching via social pressure

  • doesn’t give you space to get ahead

better to aim for being under capacity because variance makes things unpredictable

  • spare capacity?
    • flex projects
  • only schedule resources for like 2/3 capacity for mandatory things
    • rest is flex stuff, can drop to increase capacity for mandatory things
    • examples: money/budget or scheduling during waking hours (or work hours)

for next time:

writing about goals, life plan, big picture

  • prerequisites, resources used for diff things
  • understanding of success criteria
  • ppl sometimes jump into projects before knowing the success criteria and goals
    • doesn’t work for hard things
    • can get some benefits with hard things, but less efficient

You can leave a comment anonymously. No sign up or login is required. Use a junk email if not your own; email is only for notifications—though, FYI, I will be able to see it.

Comments powered by Talkyard.