Curatorial-Voice Learning
When choosing what to learn from, who produced it matters more than what it is about — pick the individual educator or author whose exposition you resonate with, over the resource that merely matches a pre-existing interest. Grant Sanderson frames this as a pre-LLM insight: advise a college student to weight the quality of the teacher over the topic of the course ("your pre-existing interests are kind of arbitrary right now"), and advise a reader that "who the author is maybe matters more than if it's a prior interest" — so if you liked one book, read what else that author wrote rather than another book on the same subject. The unit of trust is a person's voice, not a subject tag.
For the vault this is the selection half of a pair: it says how to choose the artifact, while Motivated Exposition says what makes the chosen artifact good and LLM as Resource Router says how to use a model to find that person. It is also the human-taste analogue of firehose's personal-dedup rule — relevance is scored against who resonates with this operator, not a global notion of the "best" resource.
Claims
- In choosing what to learn from, the individual educator/author is a stronger selector than topic-match to your prior interest — follow the author, not the subject. (principle — as asserted by Sanderson) — durable as a heuristic: expositional quality and voice travel with a person across topics, whereas a topic tag says nothing about whether the treatment will land.
- Pre-existing interests are a weak basis for selection because, early on, they are "kind of arbitrary"; resonance with the teacher is the better signal. (observation — the source's advice to a college student) — a claim about how interests form, offered as career/study advice.
- If a book landed for you, read the same author's other work before reading another book on the same subject. best practice — context: self-directed learning where you have already found one voice that works; the heuristic is weakest when you are deliberately seeking contrast or a second opinion, where topic-spanning breadth beats author-loyalty.
Related
- Motivated Exposition — the property that makes a chosen author worth following: they craft motivation and a through-line, not just correct sentences.
- LLM as Resource Router — the mechanism for finding the right person: ask the model "who should I read?" and go to that human artifact.
- Cognitive Scaffolding — a resonant voice is part of the scaffold a learner leans on; removing the educator you resonate with degrades the learning system, not just the content.
- Abstract-to-Concrete Grounding — the method half of learning to this page's selection half: once you've chosen whose voice to learn from, a concept only sticks when you build a concrete instance of it in your own work.
- Distillate: The Trick to Using LLMs to Learn — Grant Sanderson (3Blue1Brown) × Dwarkesh Patel
- Distillate: You're the Problem, Not Claude — Six Fixes to 10x Output — introduces Abstract-to-Concrete Grounding (bridge abstract to concrete / proof-based learning), the learning-method complement to this page.