Yeah, I basically agree with you here—I’m very happy to read LLM-written content, if I know that it has substantive thought put into it and is efficiently communicating useful ideas. Unfortunately right now one of my easiest detectors for identifying which things might have substantial thought put into them is “does this set off my LLM writing heuristics”, because most LLM-written content in 2025 has very low useful-info density, so I find the heuristic of “discard LLM prose and read human-written but lazily worded prose” very useful.
Yeah, I basically agree with you here—I’m very happy to read LLM-written content, if I know that it has substantive thought put into it and is efficiently communicating useful ideas. Unfortunately right now one of my easiest detectors for identifying which things might have substantial thought put into them is “does this set off my LLM writing heuristics”, because most LLM-written content in 2025 has very low useful-info density, so I find the heuristic of “discard LLM prose and read human-written but lazily worded prose” very useful.