Writing serves as a powerful cognitive amplifier, according to a blog post by Murat Bufaloe exploring the connection between writing and thinking. The insight draws from computer scientist Manuel Blum’s observation that the only difference between a finite automaton and a Turing machine is paper and pencil—yet this single difference represents an entire complexity class of problem-solving power.
Buffaloe recounts his own experience using mind maps in the late 1980s to process dense reading material. Initially struggling to retain information, he discovered that creating notes transformed comprehension: “My pages were not for storing the information as much as they were for processing it.” This externalization of thinking allowed him to organize and make sense of complex material in a way his unaided mind could not.
The mechanism, Bufaloe explains, works similarly to “out-of-core execution”—an algorithm technique that uses external storage to handle problems larger than available memory. When reviewing academic papers, he finds that the act of writing a review exposes gaps in his understanding, forcing deeper engagement. In one instance, writing about connections between Paxos and Nakamoto consensus led to unexpected discoveries that became the foundation for a research paper.
Buffaloe also notes that writing serves as personal external memory. He frequently references his own blog posts to reconstruct understanding, observing that thoughts never committed to writing tend not to stick: “The thoughts you never wrote down do not stick well, because they were never fully thought.” Quoting Leslie Lamport: “If you think without writing, you only think you’re thinking.”
In the era of large language models, Bufaloe argues writing for thinking has become even more valuable—a “comparative advantage” for those who still engage in the practice themselves. While LLMs can generate polished text without underlying thought, they cannot generate the novel connections and intuitions that emerge from a writer’s own cognitive wrestling with ideas.
Buffaloe offers three practical rules for writing as thinking: first, produce drafts without self-criticism or perfectionism; second, aggressively edit and revise with detachment; third, prioritize clarity and simplicity over complexity. He emphasizes that this slowness is the point—the extra cognitive time allows insights and connections to surface naturally.
Key facts
- Writing externally elevates human problem-solving capability by an entire computational complexity class, similar to upgrading from finite automata to Turing machines
- When writing a paper review, Bufaloe discovered he initially understood only 60% of content until writing forced fuller comprehension
- Bufaloe’s 2018 blog post about Paxos and Nakamoto consensus connections, developed through writing, became the seed for a peer-reviewed research paper
- Writing serves as external memory; thoughts never written down do not stick because they were never fully developed
- LLMs can generate polished text without genuine thinking, making personal writing a widening comparative advantage