When merging comments during metadata download calibre picks the longest. The goodreads plugin is probably setting the comments to a string with spaces or similar. It should instead set it to None or the empty string, or just not set it at all when no comments are present.
See the merge() function in sources/identify.py
|