You Are What You Read
The author argues that Amazon's strength came from a pairing of deep writing and deep reading, and that cheap AI-generated writing is eroding both. When documents are free to produce, their presence no longer signals effort, so readers can't assume a memo is worth their time.
- Amazon's writing culture only worked because reading was equally serious; the author spent 1-4 hours a day reading high-signal documents.
- Writing used to be expensive: a six-page memo took about a week, so the document itself was proof someone cared. AI removed that signal.
- Readers are increasingly skipping docs or tagging suspected AI text "AI;DR," and the author urges readers to pick a few trusted curators and read them deliberately.
The author follows up on a viral piece about the decline of Amazon's writing culture. The original article reached close to a million impressions, and the response made him realize he'd only told half the story. The missing half is reading. At Amazon, deep reading was the norm: he says he spent one to four hours a day on it, and that habit, paired with the company's habit of writing six-page memos, was the "secret sauce." Readers could assume a document was high-signal because writing one took a week of real effort. Nobody spent that long on something they didn't care about, so the existence of a doc was itself a signal.
Reader responses sharpen the concern. A former colleague, The Wiz, described doc reviews where nobody had read the document, and a current Amazonian, Noble Udoh, tags AI-generated documents "AI;DR" and notes that people are increasingly summarizing AI-generated text with AI. One reader put it bluntly: a doc used to prove someone had sat with a problem for a week, and now it proves they have a subscription. The author also points out that some LinkedIn pushback to his piece was itself obviously AI-written, which he found more depressing than the criticism. He leans on the programmer's maxim "garbage in, garbage out," arguing the reverse holds too. Quality in, quality out explains why reading at Amazon paid off. Now that writing costs nothing, its signal value has collapsed.
He borrows Cal Newport's food analogy to make the case: AI writing is fast food, cheap and easy to consume but empty, and unlike most food it carries no label. Newport's advice is to slow down, know what you're consuming and why, return to a few reliable sources, and stop snacking. The author closes by asking readers to name a curator they trust, so he can compile a shared list of sources worth reading.