I wanted to extract Einstein’s thinking methods into something reusable. The obvious structure was a file called “Einstein” containing his heuristics — thought experiments, aesthetic judgment, productive stubbornness. Then another file for Feynman. Another for Munger.
The obvious structure was wrong.
When I actually extracted the heuristics, they didn’t want to stay in their containers. Einstein’s thought experiments and the Talmudic tradition of testing rules at absurd extremes turned out to be the same move, discovered independently across centuries and cultures. Munger’s inversion and the Stoic premeditatio malorum — same move. Einstein’s sensory thinking and Feynman’s multiple representations and Darwin’s reasoning-by-analogy — all variations of a single cognitive operation: change how you’re looking at it before you change what you’re doing.
The person isn’t the unit of knowledge. The move is.
Once I saw this, the whole approach flipped. Einstein isn’t a container of heuristics — he’s a specimen. You examine him to learn about the class, then you discard the container. The heuristics go where they naturally belong: the thought experiment technique joins your truth-seeking toolkit, the aesthetic judgment joins your simplification practice, the productive stubbornness joins your self-regulation system.
The specimen is the drill bit, not the container.
This reframing solved a practical problem too. When you organise by person, you get duplication (where does “thought experiments” go — Einstein or Talmud?), false boundaries (Munger’s inversion and Stoic negative visualisation are the same thing filed under different people), and retrieval failures (when you’re stuck, you think “how do I see this differently?”, not “what would Einstein do?”).
But the most interesting discovery was what happened when I mined six specimens — Einstein, Feynman, Munger, Stoicism, Darwin, the Talmudic tradition — and tried to sort their heuristics into my existing system. Some heuristics mapped cleanly to one home. But the best ones didn’t. They mapped equally to two different places, belonging fully to neither.
That dual-mapping wasn’t a filing problem. It was a diagnostic. When a heuristic extracted from Einstein also fits your judgment system and your simplification system equally well, and then the same pattern shows up independently in Feynman and Munger and the Stoics, you’ve found a concept that lives in the cracks between your existing categories. The specimens are stress-testing your architecture, not just enriching it.
Six specimens, three gaps found. Representation shifting — the ability to see the same problem differently by switching mode, notation, timescale, or analogy — showed up in five out of six. Productive not-knowing — actively mapping your ignorance rather than papering over it — showed up in all six. Every single specimen, across cultures and centuries, had a version of “know what you don’t know and build systems to stay honest about it.”
The convergence is the signal. If one thinker says something, it might be idiosyncratic. If six independent specimens from different eras, cultures, and domains all converge on the same move, you’ve probably found something load-bearing about how good thinking works.
The specimens dissolve. The moves remain.