Voices
Scott Tolinski's complaint about GPT's mini-headings becomes a 67K-view referendum on model formatting tics
The Syntax.fm host posted on August 18 asking for "a good one liner to my agents from never doing this again" about the small bold headings GPT models sprinkle through every response, drawing 405 likes and 67,288 views. The signal is that formatting habits are now a model-selection criterion for working developers, not a cosmetic preference, and that people are trying to fix it at the agent-instruction layer rather than switching models. It is a single post with no data behind it, but it names a friction point that shows up in every agent-authored artifact.
Source
↳ Follow the thread