I was with him until this part...
> Stross, however, admitted he "can see some circumstances where being able to train a local model on my work could be useful to me."
> "I'd quite like a tool (running entirely locally on my own hardware, with no cloud service and no copyright-thieving grifters making bank on it via subscription fees) that digests a manuscript and derives a scene-by-scene timeline, that I could then query interactively and use to plan my next round of edits," he wrote.
Sigh. You were doing so well.
(And by the way, even open weights models are generally doing mass theft. The amounts of data that are required for training an LLM can't really be sourced on public-domain content alone, and certainly couldn't be trained entirely on Stross's own work while still being remotely usable.)
