CFAR Notes: Part One

This is a **extremely** rough draft of my notes from CFAR (April 4th-8th 2018)

Seeking Sensibility


Everyone has a sensible point of view
  • stay curious to learn what that is
  • don’t share your own mental models onto each other
  • what do i see when i look at another mind
    • something is going on
    • don’t focus on what’s missing

Example: Division

Two ways of thinking about this: either equal sharing or repeated subtraction
psychologically different though even if you get the same result

In the division situation zero about blame (teacher isn’t to blame for not teaching, kid isn’t for thinking of it in a different way)
  • interesting to think about: how many times are are we communicating without both os realizing we see it differently
What kinds of puzzles was the kid in the division example solving?

Example: Teaching English

Teaching “I like it" kept coming up as “I like"
This is because “Me gusta" means it pleases me not like. There is an implied it/object that like in english does not convey. Although they are equivalent phrases, they are not the same word.

It’s important to model experiences so we know the ways people will receive what we say.
Easy to say what we/they don’t know but it’s not easy to figure out what they are doing/know. And it is valuable to understand what they do know so we can figure out patterns/their thoughts.

other nice things

Don’t try to contribute or comment, just be curious!
Try to ask for the other person to “taboo it"
“You make sense"
Everyone does things that are meaningful to them. Why are we doing this? Every time we interact with someone we can choose to boggle and not judge.

Inner Sim

Note: This is not imagination and is just one of many advisors
(examples of advisors: wiki, feels, explicit verbal models, inner sim, logic)

Example: Throwing a pen between your hands

I am about to toss a pen from my left hand to my right hand. Before I physically do the act, you already know the general parabolic curve it will follow to land in my hand. This is what inner sim is.

when should I not use inner sim?

when should I use inner sim?

  • just facts
  • totally new domain
  • no data in the past
  • physical interaction
  • new but similar situation
  • social landscape
  • data in past → future


Questions to ask yourself: How surprised do I feel? How much disbelief are you suspending?
This can help you figure out expectations (hidden and implicit)
outcome → inner sim → surprise


Ask yourself why you feel surprised/why you have the beliefs you have

outcome/failure → inner sim → cause of failure

This is a powerful tool to help you create bulletproof plans.


step 0: pick a plan/project goal
step 1: outline your plan
step 2: check your surprise-o-meter
step3: use pre-hindsight
step 4: bulletproof it
step 5: iterate till you create something solid

Units of Exchange

TLDR: things are made of parts, those parts can be exchanged

things we can exchange:

  • money
  • time
  • attention
  • energy/willpower
  • affection/goodwill
  • knowledge
  • pleasure
  • rest

Seek simple comparisons, BUT don’t trust them
  • just use units of exchange to get a gut feel/approximation

Example: Microwave

You have three microwaves you could buy at three different tiers.
$90, $300, $500
An average life of a microwave in the US is 10 years. Let’s say the low end microwave requires you to add another minute to your time usually since it doesn’t heat things up as well. If you amortize that over 10 years that’s a lot of time. It’s not just about the sticker price, but also about the time, and happiness

Other nice things to consider when evaluating costs

marginal utility
willpower works however you think it will

Trigger Action Plans

  • use it for noticing


  • retrain
  • turn it off
  • “summon sapience"

TAP Failure Modes

  1. The trigger doesn’t fire
    1. solution: choose something more specific and concrete. something that already catches my attention
  2. You don’t take the action
    1. solution: choose a smaller simpler more effortless action
    2. solution: make the action an option
    3. solution: goal factoring

Don’t make yourself feel bad when you don’t do something you want to stop. This leads to blind spots and you avoid noticing. You want to be able to break autopilot and that means really noticing.

Approach your bugs as a scientist, and not a moralist. Be curious about why it’s not working and why you’re doing the default.

Goal Factoring

Purpose: check how actions align with goals

The idea is to reduce negatives and make sure actions are do-able. or replace actions to decrease negatives yet reach goals

How to Goal Factor

  1. Pick an action
  2. List out negative/positives from action
  3. Check for relevance and completeness re: negatives/positive
    1. Some positives/negatives just don’t matter so they shouldn’t really be considered
    2. Make sure you’re being honest with yourself (what are you REALLY giving up? what actually drives you to complete the action? why do you really want this?)
    3. Thought experiment: There is an alternative action that the genie has that hits all the goals and avoids negatives you have listed. Would you accept it? If not, then explore what you’re missing.
  4. Search for alternatives
    1. Can be one to many mapping, many to one, one to one replacement