The 3B newsletter
Short posts for developers working with coding agents: software design, computer science fundamentals and practical AI.
Free. Every post has two parts: what to do, and why it works.
The few steps that work today, in plain words.
The reason behind them, so you can judge what an agent gives you.
The idea behind it
The 80/20 approach: a focused foundation, expert guidance and a way of reasoning. Enough to start learning on your own.
100 things competing for your attention.
Vilfredo Pareto noticed that a small share of his garden's pea pods produced most of his peas. Then he found a similar skew in land ownership.
Learning often works this way too. Many fields have a handful of ideas that carry the weight and a long tail that mostly makes you feel busy.
A good guide says which is which, and leaves the rest out.
"80/20" is a metaphor, not a law. It names one principle: a small foundation carries disproportionate weight.
Ten years of teaching keep confirming one sequence.
Consistent with research: Sweller's cognitive load theory (structure first), Roediger and Karpicke's retrieval-practice studies (using knowledge beats rereading it) and Vygotsky's zone of proximal development, the guiding principle behind step four. Wood, Bruner and Ross named that support "scaffolding".
In practice, learners who treat struggle as part of the process tend to keep going.
Choosing the "20%" is a judgment call, and it varies by domain.
Reply to any post or leave a comment under it. Every question gets read, and the best ones become the next post.
Short questions are easiest to answer. Say what you already tried.
For teachers
Any subject with a foundation worth teaching well: music, history, craft, code.
Tell us your subject and one thing people get wrong about it.
Short posts for developers working with coding agents, straight to your inbox.