TL;DR
Get the little things that make your day delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
A November 2026 blog post argues that writing is not just a way to record ideas but a tool for developing and testing them. Its author says writing can expose gaps in understanding and help form connections, while warning that AI-generated text may look like evidence of thought without reflecting a person’s own reasoning.
A computer scientist writing on the blog Murat’s Musings argues that writing helps people think by moving ideas into an external form they can examine, revise and connect. In a report published in November 2026, the author also warns that AI-generated prose can resemble evidence of clear thinking even when the person using it has not developed the ideas themselves.
The author describes first discovering writing as a way to understand material in middle school. Long passages were difficult to absorb, so he began taking notes and drawing mind maps. He says arranging information into branches helped him work out how ideas related to one another. In retrospect, he writes, the notes served less as storage than as a means of processing what he read.
He later connected that experience to advice from computer scientist Manuel Blum. As quoted in the post, Blum compares a finite automaton, which has limited computational ability, with a Turing machine, which can use paper and pencil as external memory. The blogger uses the analogy to explain how writing can let people work through more complex problems than they could keep in mind at once. It is an explanatory analogy, not a claim that human thought literally changes computational class.
The post gives two examples from the author’s work. He says drafting a review of a research paper often reveals that his initial understanding was incomplete; writing helps him place the paper in context and identify connections. He also recounts beginning a 2018 post about similarities between Paxos and Nakamoto consensus without starting from a settled thesis. As he wrote, the connection became clearer, and the post later became the seed for a research paper, he says.
How Drafting Develops Ideas
The argument has practical implications for students, researchers and professionals who use writing to understand difficult material. A draft can make a vague idea visible: once it is on the page, a writer can test whether the explanation holds together, identify missing steps and see where further reading is needed. In that sense, writing is not only a final presentation of thought but also part of the work that produces it.
The author’s account is personal rather than a controlled study, and the post does not quantify how much writing improves comprehension. Its examples nevertheless illustrate a distinction relevant to readers: a polished explanation may communicate a conclusion without showing how the writer reached it. Generating text and developing an idea are not necessarily the same activity.
That distinction matters as people increasingly use language models to draft and summarize. The post does not argue that AI tools have no value; it questions whether relying on them before doing one’s own exploratory writing can displace the process of forming an original view. The author says the model may know established literature and standard interpretations, but cannot be assumed to know a user’s unfinished intuition or make a new connection on that person’s behalf.
From Mind Maps to AI Writing
The post traces the author’s view through several stages: school note-taking, graduate study, research writing and the use of a personal blog as a record of ideas. He says he returns to earlier posts as an external memory, using them to rebuild understanding. His explanation is that writing an idea out helped him develop it enough for rereading to restore the relevant reasoning. That is his description of his own experience, not a general finding established by the post.
The author also refers to Amazon’s six-page narrative memo as an example of writing traditionally associated with careful thinking. He invokes Jeff Bezos’s statement that a writer cannot produce a clear, six-page narrative memo without clear thinking. The post argues that AI changes how readers should interpret such documents: a long, coherent memo can now be generated without demonstrating that its user has done equivalent reasoning.
For its practical advice, the post recommends beginning with an unrestricted draft. The author’s first rule is to “mess up”: produce notes in bulk, leave errors in place and avoid polishing while ideas are still forming. He suggests treating a working document as a lab notebook, where tangents and unfinished thoughts can be recorded before deciding what belongs in a finished piece.
“Without writing, you are reduced to a finite automaton. With writing you have the extraordinary power of a Turing machine.”
— Manuel Blum, as quoted in the blog post
Limits of the AI Argument
The post offers personal experience and examples, not experimental evidence comparing writing with AI-assisted work or measuring effects on learning. It does not establish how often AI use displaces independent reasoning, which tasks are most affected, or whether particular ways of using a model can support rather than replace thinking.
Its claims about what language models can do are also the author’s assessment. He says he does not trust an LLM to make a creative leap or form a new connection, but the post does not define a test for those abilities or compare model performance across examples. The broader effects of AI on writing and research remain unsettled.
Put the First Draft on Paper
The post closes with a direct recommendation: write before polishing. Its suggested next step for readers is to start a rough document while reading or doing research, record questions and tangents, and use the resulting draft as material to examine. The author advises delaying correction and presentation until there is something to react to.
For readers using AI tools, the post’s argument points toward a practical distinction rather than a ban: first try to state the problem, intuition or question in your own words, then decide whether a model can help with a later task. The author does not offer a formal workflow or test for preserving independent thought. Whether such habits measurably improve reasoning is not addressed in the report.
Key Questions
What is the blog post’s main argument?
It argues that writing can help produce thought, not simply record ideas already formed. Drafting gives a person material to inspect, revise and connect.
Why does the author compare writing to external memory?
The author says writing moves ideas out of the mind and onto a page, where they can be rearranged and revisited. He describes using old blog posts to reconstruct his understanding of topics.
Does the post say AI cannot help with writing?
No. The author questions whether a language model can be trusted to make an original connection or develop a user’s half-formed intuition. The post does not present a systematic evaluation of AI writing tools.
What does the author recommend for a first draft?
He advises writing freely and in bulk before editing: record incomplete thoughts, tangents and rough phrasing, then use that draft as something to examine and improve.
Does the post prove that writing improves thinking?
No. It presents the author’s experiences and examples, including research writing and note-taking. It does not report a controlled study or quantify the effect.
Source: muratbuffalo.blogspot.com
Popular right now
Samsung 65-Inch QN70FA Neo QLED 4K Smart TV with 8-Foot WireLogic Ruby Ultra High Speed HDMI Cable (2025)
Vitamix Touch & Go Advance 32-Ounce Bar Blender, White
Ninja Luxe Café Premier 3-in-1 Espresso Machine, Drip Coffee Maker, and Cold Brew Maker
As an affiliate, we earn on qualifying purchases.
