Thanks for the report. I've been thinking about changing the script to load comments directly from lesswrong.com instead of through ibiblio.org, and I'll move that up in my todo list. If anyone is hacking on the LW code base, my work would be made much easier (and the script would be faster) if there are APIs to download all comments after a given comment ID, and to download comments that match a set of comment IDs.

ETA: Fixed now