This is a draft. Critical feedback is welcomed in the comments at the bottom of the page.

Most people suck at explaining. I think it's because it is hard to serialise your mind. To convey an idea, you have to encode a tree-like[^1] neural network as a sequence of words that will construct the same shape in the mind of another.

When building a pyramid, you don't place the capstone[^2] first. And ideas, like real-world structures, build on prior foundations. By analogy, you can convey ideas faster by leveraging prior knowledge.

One of Clojure's strengths is that its core data structures[^data-structures] are easily serialised. Consider the mathematical expression (3 × 4) + 5. We can represent the order of operations as a tree:

[^data-structures]: Lists, vectors, hash-maps and sets.

Serialise an Expression Tree to Lisp code

The Lisp expression (+ (* 3 4) 5) evaluates to 17 and is written in parenthesised prefix notation. When the source code of a computer program is deserialised, it is stored in memory as an expression tree[^3].

Ideas are harder to serialise because they are stored as connections between neurons, which are connected via multiple routes. Idea A refers to B and B refers to A. What should their order be? This makes them tricky to encode sequentially without a good syntax for referencing undefined concepts.

English makes provision for unnamed things and how they relate spatially, but I am curious if prepositions in other languages like German make explaining easier by mapping more closely to how memories are stored[^4].

The problem of serially ordering ideas suggests that a good notation for conveying ideas will have native support for declaring undefined references, like x in math, and late-binding of definitions. Or, you should explain related concepts in parallel.

How to Win at 30 Seconds

In the board game 30 Seconds you design clues that will prompt your teammates to say a term. I love this game because it trains you to serialise your thoughts and transmit them without error or distortion.

One strategy that rarely distorts for me is a top-down definition followed by bottom-up clues to trigger episodic memory:

"Famous formula 1 driver," (top-down category)
"who drove for Ferrari and had a ski accident." (bottom-up events)

Answer: Michael Schumacher.

That works well for recalling old knowledge, but what about teaching new concepts? A good protocol is the ADEPT method:

  1. Use an Analogy,
  • Diagram,
  • Example,
  • Plain English, and
  • a Technical description.

Did it work for this essay?

Explaining is a skill. It takes empathy to place yourself in the mind of another and build on their foundation. With the right strategy and practice, you should beat me at 30 Seconds.

[^1]: Assuming ideas are stored as connections between neurons, which can be modelled as recurrent neural networks and cycles are hard to serialise, hard to serialise, hard to serialise. [^2]: The tip, or final piece of a pyramid. [^3]: Computer programs are represented as abstract syntax trees, or ASTs. [^4]: Linguistic relativity.