My view is that we should adopt a strict position on these kinds of question, but not an outright ban. As I have mentioned elsewhere, my view is motivated by this example from TeX SE, which I think shows what a good resource a list question can become if well-managed (look at the number of up votes/favourites!).
Comparison of that example with some other, less successful cases suggests some rules of thumb for a good list question:
- One suggestion per answer. This makes the answers easier to browse, avoids duplication and, most importantly, allows people to vote on individual suggestions. From the TeX example, this has meant that the voting mechanism can serve to promote what the community thinks are the best suggestions.
- Standardised format for each answer. For example, a title in standard formatting that states what item this answer is recommending. A standardised bullet-list of metadata (e.g. operting systems supported for software, topics covered for a book, etc.). A standardised image (e.g. screenshot or cover image).
- All answers are community wikis. Necessary so that the community can ensure that the list remains an up to date resource.
- A fairly well-defined and meaningful scope. Big list of economics books is probably not useful. List of textbooks in industrial organisation is likely to be more useful.
My feeling is that with guidelines like these we can support some very valuable list questions (book recommendations, software recommendations, seminal papers on a topic, etc.)