- Our great strength and our great weakness is that we can model the unseen
- Can build network of causes connected to sensory experience
- Can build network of causes disconnected from sensory experience
- Empiricism - constant ask what your beliefs predict and what they prohibit
- Maintain a list of things that would falsify your beliefs
- Don’t ask what to believe, ask what to anticipate
- Replace words with definitions to clarify points of disagreement
- Tabooing words can dissolve disagreements caused by differing interpretations of the same word
- Tabooing words can reveal disagreements caused by using the same word to paper over distinctions between two concepts
- If a philosophical term is giving you trouble, try to formulate the argument without using that term at all
- Confidence levels inside an argument - probability estimate generated by model
- Confidence levels outside an argument - probability estimate generated by model and probability of model being correct
- Doubt in validity of model should push you back towards your priors
- If your model is coming up with very low or very high probabilities, examine the assumptions behind the model
- If you’re irrational, more knowledge can hurt you
- People are way more likely to use knowledge about biases to knock down arguments they disagree with than arguments they agree with
- Biases and fallacies can be fully general counterarguments if you’re not careful about applying them properly
- Imperfect information still beats zero information
- Fermi-estimating statistics can still give you enough information to make decisions
- Just thinking about and going with the base rate can counteract many cognitive biases
- Applying made-up models can train us to notice where our intuitions are wrong