This week, Simon Willison resurfaced a blog post whose one line forced the developer community into collective self-reflection: as writing gets bulk-outsourced to AI, the scarcest resource turns out to be "human voice."
What This Is
Lynch lists four anti-patterns in software blogging: opening with a long wind-up before getting to the point, misjudging what the reader already knows, assuming readers will plow through every prior post you've written, and over-formal posturing disguised as professionalism.
Willison personally claimed one — using outbound links as an excuse to skip explaining terms. "I do this all the time, but I've long suspected almost nobody actually clicks through."
Lynch's core point is blunt: "Junior software bloggers share a collective delusion that they have to write stiffly and over-formally to look professional. Just write the way you actually talk."
Industry View
Willison flags a trend at the end of the post: more developers are outsourcing their writing to AI, and software blog content is turning bland and interchangeable. Readers' appetites have been whetted — they're getting hungrier.
But there's a counter-current. Several senior tech editors told us privately that "human voice" itself is an overhyped nostalgia trip: technical documentation, API references, internal corporate training — none of it needs personality; it needs accuracy and efficiency. A former editor-in-chief with twelve years in tech media put it this way: "When I need to check how to call an API, I don't want the author's soul. I want the right answer."
The more telling flip side: when everyone is using AI to write, what holds a reader for 30 seconds is the content that "doesn't read as AI-written at first glance." "Human voice" is shifting from a cost line to a premium.
Impact on Regular People
For individual careers: when colleagues all use AI for weekly reports, emails, and presentations, deliberately keeping a bit of your own voice and judgment may be a low-cost way to stand out.
For enterprise IT: internal knowledge bases and customer service scripts are fine for AI batch production; but for external brand content and customer communications, over-AI-ification is quietly eroding trust.
For the consumer market: readers are being trained into "AI-flavor sensitivity." On WeChat Official Accounts, Xiaohongshu, and Zhihu, content that's obviously AI-generated at first glance is seeing declining click-through and read-through rates.