Writing as a Thinking Tool

Writing is presented as a tool for doing thought, not merely recording it. The author traces this idea from childhood mind maps, which helped make sense of reading, to Manuel Blum’s analogy between finite automata and Turing machines: paper and pencil give the mind a form of external memory that greatly expands what it can work through. Writing slows thinking while keeping attention on a problem, allowing understanding and unexpected connections to emerge. Reviewing research papers often reveals gaps in comprehension, while writing about links between Paxos and Nakamoto consensus led to a research-paper idea. Written notes also serve as a personal memory, making earlier understanding easier to reconstruct.

The author argues that language models do not make this process unnecessary: they can reproduce established knowledge, but may miss a person’s unformed intuitions and original connections. Because AI can generate polished prose without genuine thought, writing for oneself becomes more valuable. Practical advice is to draft freely and without self-criticism, then revise aggressively across multiple drafts; use plain, concise language; write for intrinsic reasons rather than an audience; and schedule writing instead of waiting for inspiration. Regular practice, the author finds, cultivates a habit of noticing ideas, brings them to the surface, and clears mental space for new ones.