This in-progress sequence chronicles my journey to create “feedbackloop-driven rationality”.

This includes inventing better quick/​concrete feedback loops for rationality training (which might nonetheless miss some big picture fuzzy value), as well as developing incrementally-better feedback loops for whatever big-picture-fuzzy value you care about.

Example of “concrete”: Thinking Physics puzzles (which test “mapping out confusions” on the timescale of a couple hours)

Example of “big picture but fuzzier”: Developing the skill of making cruxy predictions about the outcomes of your plans, and becoming calibrated on those predictions.