2017-11-06 RRG Notes

Gears in understanding

When looking at a model, it can be useful to ask how deterministically interconnected the variables of the model are
How do you know whether your model has this property?
- Is the model falsifiable?
- If the model were falsified, how much and how precisely can you infer other things from the falsification?
- How incoherent is it to imagine that the model is accurate but a variable could be different?
- If you knew the model were accurate, but you forgot one of the variables, could you rederive it?
Example: gears in a box
- If you have gears in an opaque box, then you can hardly make any predictions about the system, and your predictions aren't well constrained
- If you can see the gears in the box, then you can make many more predictions and learn more about the system when your predictions are falsified
Example: Arithmetic
- There's a tension, in practice, between memorizing and drilling algorithms that let you compute answers quickly and "really understanding" how those algorithms work
  - Is there really, though? Maybe you just don't spend enough time memorizing and drilling.
- This tends to devolve into debates about what it means to truly "understand" a mathematical algorithm or concept
- Valentine is proposing that we should define understanding in terms of how well the student knows the component parts of the model and how well the student is able to re-derive missing parts if they forget
Example: Valentine's Mother
- If Valentine tells you that their mother likes history, that's just a random, unconnected fact in your mind
- But for them, that fact can be derived from their mother's other interests
- A world in which everything else was the same, but their mother did not like history would be confusing for them in a way that it wouldn't be for us
- Their model of their mother has more deterministically interconnected pieces than ours does
- Getting to know someone can be interpreted as "adding gears" to our model of them
- Models can be viewed on a continuum between collections of unrelated facts and axiomatic systems where everything is a logical deduction
Example: Gyroscopes
- Most people think that gyroscopes behave in a non-intuitive fashion
- For most people, it seems coherent to imagine a world in which physics works exactly the same way, except that when you suspend one end of a gyroscope, it falls like a non-spinning object
  - Well, yes, but that's because most people's day-to-day experience of physics doesn't include relatively high angular momenta
  - When you look at a still object and a spinning object, it's not immediately obvious why spin should matter in the object's behavior
  - I really wish that rationalists would stop taking things that aren't intuitive and saying that they ought to be intuitive
- The nice thing about physics is that everything is interconnected
  - Falsifying any part of physics leads to updates about all of the other parts
  - It is incoherent to imagine all of physics as it currently is, with one or two parts working completely differently
- The interconnectedness and determinism of the model tracks a true thing about the world, which is why we want our models to be more interconnected and deterministic
Gears-ness is not the same as goodness
- Interconnectedness should not substitute for accuracy - if you have a model that's more vague but also more accurate, go with that
- Models should also inspire ways of understanding that are useful
- Most of our models don't connect all the way down to physics, so it's all right not every part of your model can be strictly derived from axioms
Insisting that models be interconnected and deterministic is a powerful tool for cutting through confusion
My thoughts
- Why do rationalists come up with new terms for things? Why can't we call "gears-ness" "determinism" or "interconnectedness"?

In Praise of Fake Frameworks

Valentine uses a lot of "fake frameworks" - models that are probably or obviously wrong in some way
Examples of fake frameworks:
- Extroverts and introverts
  - Extroversion and introversion aren't words that refer to clear traits
  - Realize the intuition is wrong, and then use it anyway
    - Is this actually possible to do?
  - Put the intuition in a mental sandbox and see if it comes with any useful predictions, even if it's wrong
- MBTI personality types
- Without the intuition that led to folk frameworks for personality no one would have done the cluster analysis that lead to the OCEAN personality traits
Ontology is a set of "basic" things that you use to build a map or a model
- "Point", "line" and "plane" in Euclidean geometry
- "Mass", "position" and "time" in Newtonian mechanics
- People get confused when they switch ontologies without noticing
- A road is a basic unit in the ontology of street map
- Talking about how roads are made of atoms is true, but also unhelpful when attempting to navigate
- In the same way, MBTI can be helpful as a high-level approximate abstraction, even though it breaks down at some level
It's a type error to say that an ontology is "correct" or "wrong"
- Ontologies are basic elements from which you build models
- It's the model that's correct or wrong, not the ontology
- The concept of ki in aikido can be useful even if it has no single correspondent in physics
It should be possible to try on ontologies without hurting your core belief system
My Thoughts
- Again with the terminology. Why shouldn't we call it "ontological flexibility" instead of "fake framework"?
- Rationality is nowhere near popular enough to get away with starting its own terminology without starting to seem like a cult

The MTG Color Wheel

Magic: The Gathering (MTG) colors form an ontology of personality types and motivations
MTG has 5 colors, each with its own characteristics
- White - peace through order
  - Tries to achieve peace by imposing order
  - Coordination, cooperation and restraint are the solution to all the unhappiness in the world
  - Angels, clerics, knights
  - Asks what is the right course of action to take, where "right" is determined by their moral or cultural framework
  - Archetypal organization: church
  - Dystopia:
- Blue - perfection through knowledge
  - Seeks perfection, and attempts to achieve that perfection through knowledge
  - Figure out the truth and apply that truth to the fullest extent
  - Change the rules, rather than just applying them
  - Asks "What course of action makes the most sense?" Sense is determined by careful thought or expertise.
  - Archetypal organization: university or research lab
  - Dystopia: efficiency pursued without morals or limits
- Black - satisfaction through ruthlessness
  - Desire for power and agency
  - Ability to reshape the world around it
  - Capable of cooperation and alliance, but only consequentially
  - Amoral, not immoral
  - Asks, "What will leave me best off?"
  - Archetypal organization: hedge fund
  - Dystopia: Totalitarian dictatorship
- Red - Freedom through action
  - Seeks ability to live in the moment
  - No real organization
  - Dystopia: anarchy
  - Asks, "What do I feel like doing?"
- Green - Harmony through acceptance
  - Most of the suffering comes from trying to fix things that aren't broken
  - The color of Chesterton's Fence
  - Asks, "How are things usually done? What is the established wisdom?"
  - Archetypal organization: hippie commune
  - Dystopia: tribe with rigid and unchanging traditions
In addition to being defined by goals and methods, the colors disagree with each other in meaningful ways

White prioritizes the group over the individual
Black does the reverse
Green sees the environment as something to be cherished and preserved
Black sees it as something to be exploited
Green sees genetics and environment as determinative, blue believes in overcoming and transcending one's origins
Blue sees red as impulsive and rash, while red sees blue as repressed and ufeeling
Red and white disagree on structure and commitment

Colors also can ally interesting ways

White and blue both agree that structure is important
- A white/blue agent asks how do we know what's right and good?
Blue and black both agree on personal growth - transcend social roles and social norms
- A blue/black agent asks how best they can achieve their goals
Black and red both agree that independence is something to be fostered and defended
- A black/red agents asks how they can get what they want
- Embraces hedonism and "live-and-let-live"
Red and green agree on the importance of authenticity
- A red/green agent asks where they are now, and where should they go?
- Being present in the moment
Green and white agree that a whole is greater than the sum of its parts
- A green/white agent asks what is fair and good

Colors can combine with their opposites in interesting ways as well

Black and white combine to form tribalism
- Asks who's in their circle of concern
- Scarcity mindset
- Progress is a zero-sum game
Blue and red form creativity
- Freedom combined with investigation
- Wild artistry and mad science
- Asks what can be achieved and what might be possible
Black and green both embrace the cycle of death and rebirth
- Asks what costs must be paid to achieve the ideal
- Belief in the virtue of evolutionary struggle
Red and white are the colors of heroism
- Asks what needs to be done and what would a good person do
- Morality and adherence to laws that may be higher than the laws of the society that one is in
Blue and green are the colors of truth-seeking
- Asks what don't they understand?
- Pursue knowledge, but disagree about what should be done with that knowledge

So what do we do with the color wheel?
- Classify things and then make predictions based upon those classifications
- Colors give you a set of associations that allow you to bias your predictions in useful ways
- Remember - colors only speak to goals and means; every color can be good in some ways, and evil in others
- Colors can help you understand how and why people do things, and thus predict what they'll do next
- For example, orderly, structural interventions against someone who's primarily red is going to feel stifling to them and cause them to rebel
- Recognize that every color is a different frame by which to interpret the world, and use the colors to better see how someone else will see what you're doing
My thoughts
- This is going to be hard to keep in your head if you're not a Magic The Gathering player
- Useful shorthand, otherwise

On Types of Typologies

Meyers-Briggs Type indicator
- It's inconsistently applied and unscientific
- Yet, Scott consistently recognizes himself in the description of INTJ
We can reconcile "Meyers-Briggs is unscientific" with "Meyers-Briggs is a useful tool"
Meyers and Briggs had no scientific basis on which to declare that there were four factors of personality - just armchair observations and anecdotes
Factor analysis shows that there are actually 5 personality traits, only one of which corresponds to Meyers-Briggs
Meyers-Briggs doesn't need to give new information in order to be useful
Meyers-Briggs can be useful for drawing conclusions in the same way that nationality can be
- Isn't that just plain stereotyping
- Even if it isn't stereotyping, you can't analogize between nationality and MBTI because nations impose cultural norms in a way that MBTI does not
Five-Factor and MBTI are trying to do different things
- Five-Factor is trying to give a mathematical, objectively correct breakdown of personality useful for research purposes
- MBTI is trying to separate people into categories that are easy to think about
- Even poorly drawn categories can be useful
My Thoughts
- It seems to me that Scott is defending stereotyping on the basis of pseudoscience
- I mean, if we're going to split people up and make pre-judgements about their personality on the basis of things that don't correspond to existing scientific truths... why not judge people by their skin color? Or their astrological sign?