I suspect a large contingent here will really hate this suggestion but here it goes:
The McPhee method sounds like a great framework for making writing prompts. That is, prompts for LLMs to write things.
>in stage one he accumulates notes; in stage two he selects them; in stage three he structures them; and in stage four he writes. By the time he is crafting sentences the structure of the piece as a whole, and of each section, even paragraph, and the logic connecting them all, is already determined, thanks to the mechanical work done in the first three stages. McPhee is on rails the whole time he writes his first draft. From there it’s all downhill and the standard thing that everybody does: revision, revision again, then refinement—a sculptor with ax, then knife, then scalpel.
I know hackernews kinda hates LLMs but I don think this idea has to be so offensive. Much of the work and value from the author is in collecting these fragments and structuring them. Purely from a communication standpoint, I have no issues whatsoever with an LLM stitching them together and choosing the vocab and grammar.
OP describes this ^ long arduous process and then notes:
> Your writing can only be as good as your taste.
That is, using an LLM to help with "collecting these fragments and structuring them" might be okay--should the writer still be able to deeply immerse--but the "stitching them together and choosing the vocab and grammar" of a faux writer LLM is likely to leave a bad taste in the mind of some readers (e.g. those HN types who "kinda hates LLMs").
That's interesting because I thought the exact opposite. How could the LLM generate the notes? Those are from the person's direct observations, investigation, etc. right? So the idea would be to farm that part out? As off-putting as LLM style writing is, letting the LLM make the content seems reckless and error prone.
I do agree that the LLM style vs. the authors style is off-putting. I have used LLMs to help me write things and I do not like how its not "my" voice. I see no reason it couldnt use my voice given some of my writing samples. In any case, I've found it very easy to revise it in my own words. This is part of the "revise" stage mentioned in the process. In addition, this step is barely necessary for some things, such as technical writing.
The McPhee method sounds like a great framework for making writing prompts. That is, prompts for LLMs to write things.
>in stage one he accumulates notes; in stage two he selects them; in stage three he structures them; and in stage four he writes. By the time he is crafting sentences the structure of the piece as a whole, and of each section, even paragraph, and the logic connecting them all, is already determined, thanks to the mechanical work done in the first three stages. McPhee is on rails the whole time he writes his first draft. From there it’s all downhill and the standard thing that everybody does: revision, revision again, then refinement—a sculptor with ax, then knife, then scalpel.
I know hackernews kinda hates LLMs but I don think this idea has to be so offensive. Much of the work and value from the author is in collecting these fragments and structuring them. Purely from a communication standpoint, I have no issues whatsoever with an LLM stitching them together and choosing the vocab and grammar.