Learning through building pedagogy

Research — This is a living document. It may be incomplete, outdated, or full of half-formed thinking.
#learning #ai

Starting point

There's a long tradition of learning by building --- from Papert's constructionism to modern project-based curricula. But AI tools are changing the dynamic. When you can generate working code instantly, what happens to the learning process?

Threads to pull

  • Constructionism --- Seymour Papert's idea that learning happens best when you're constructing something shareable. How does this hold up when the construction is AI-assisted?
  • Deliberate practice --- Ericsson's framework for skill acquisition. Building from scratch forces you through the uncomfortable parts. Does AI assistance let you skip the parts that matter most?
  • Desirable difficulty --- Bjork's research on how making learning harder can make it stick better. Is the friction of manual implementation a feature, not a bug?
  • Scaffolding vs. crutches --- when does AI help act as productive scaffolding and when does it become a crutch that prevents real understanding?

The Feynman anchor

"What I cannot create, I do not understand." --- chalked on Feynman's blackboard at Caltech, found there at his death (February 15, 1988). Note: it's "create" not "build," and the phrasing inverts the common misquote. A companion statement sat on the same board: "Know how to solve every problem that has been solved."

There's some debate whether the idea originated entirely with Feynman or echoes other thinkers (Wittgenstein, among others), but it reflects his well-documented insistence on re-deriving known results from scratch rather than taking solutions at face value.

Feynman also said: "It's the way I study --- to understand something by trying to work it out or, in other words, to understand something by creating it." This is the most directly useful companion to the blackboard quote for a building-as-learning frame.

Quote library (learn by doing/building)

  • Aristotle (Nicomachean Ethics): "For the things we have to learn before we can do them, we learn by doing them, e.g. men become builders by building and lyre players by playing the lyre." --- the original "learn by doing" quote.
  • Confucius (attribution debated): "I hear and I forget. I see and I remember. I do and I understand."
  • Benjamin Franklin (also attributed to Confucius): "Tell me and I forget. Teach me and I remember. Involve me and I learn."
  • Galileo Galilei: "You cannot teach a man anything. You can only help him discover it within himself."
  • Pablo Picasso: "I am always doing that which I cannot do, in order that I may learn how to do it."
  • Seymour Papert: "We learn best of all when we use what we learn to make something we really want." Also: "Learning is most effective when part of an activity the learner experiences as constructing a meaningful product."

For a tech/builder audience, the strongest anchors are: the Feynman blackboard quote, his "creating it" study quote, and Papert's constructionism framing.

Papert and constructionism

Papert (MIT, creator of the Logo programming language) turned Feynman's blackboard quote into an educational theory. His philosophy of "constructionism" --- sometimes summarized as "learning-by-making" --- argues that understanding deepens when learners build artefacts that are personally meaningful and shareable. This is the most directly relevant theoretical grounding for the "build it from scratch" approach in tech writing.

Things I'm reading

  • Papert, S. --- Mindstorms: Children, Computers, and Powerful Ideas
  • Ericsson, K.A. --- Peak: Secrets from the New Science of Expertise
  • Bjork, R.A. --- research papers on desirable difficulties
  • Andy Matuschak's notes on learning

Early observations

Building an LLM from scratch (the blog series) is my own test case. Implementing the basic building blocks by hand, rather than calling a library, forced me to understand shapes and broadcasting at a level I wouldn't have reached otherwise. But there are diminishing returns --- hand-rolling everything isn't always the point.

The interesting question is where the boundary sits between productive struggle and pointless suffering.

← Back to research