The “Feynman Technique” for learning
Teach it to a 12-year old who has just enough attention span, verbal and subject matter knowledge for basic concepts. ELI12 - Explain like I’m 12.
You must be able to explain - in simple, non-jargon - all aspects of the subject without any gaps.
When you find a gap, go back to the material and study it until you can fill in the gap in simple language. Review this ELI12 summary and make sure it is all simple, jargon-free and makes sense.
Try reading it out loud and make sure it sounds clear.
Optionally: try to teach it to someone else who doesn’t already understand the subject.
If you learn a new term, try to explain a concept relating to the term without using the term itself. If you cannot do so, you didn’t gain an understanding of what the term represents, you merely learned a word.
Q: doesn’t this constrain understanding to our observable surroundings? what about advanced math or other subjects that don’t have projections/models in 4D reality? (see Eric Weinstein’s explanations of his theory of everything paper towel tube)
Feynman’s tips for thinking
1.
The difference between real understanding and mimicry of others is the ability to answer simple, penetrating, “naive”, non-trick questions, directly.
2.
Bayesian updating. As observations of tests progress, you should update your ideas - along a spectrum - relative to the results of the test. What you’re aiming for is a continually-updating conclusion about the odds of a thing being true. It is anti-certainty.
3.
As investigation continues and inquiry techniques improve, the effects being investigated should grow stronger or remain the same - not weaken considerably. Each significant observed weakening should cause you to throw out previous conclusions about the strength of the phenomenon. 1
4.
Ask not “could this be possible?” but “what is probable, what is actually happening?”
There are far more physically-possible scenarios one could imagine than what is actually happening in the physical world. Because of this lopsidedness, almost anything one can dream up is not happening. If you go looking for something that could be, you are very likely to not find it.
5.
You cannot judge the probability of something after it happened. This applies especially to experiment.
If you run some tests and find a peculiarity and then from that you select the peculiar case to carry on with in your conclusion about the results of the tests, you are cherry-picking.
A new, clean test must be devised - you can’t use the data of the one that gave the clue.
6.
Proper statistical sampling that strives to reduce all filtering and selection effects is what is needed to draw strong conclusions. “The plural of anecdote is not ‘data’.”
As far as sampling goes, to get to 98% or 99% correct you need 10,000 samples, it turns out.
7.
Most errors in reasoning come from lack of information and lack of knowledge of what tools to use and how to use them.
Example from astrology: Once you know how the universe actually works, you can’t believe astrology without very convincing experiments.
Ref: Farnam Street blog