I was reviewing an answer to a question asking for references of books in Real Analysis. The answer was very brief saying "Lalji prasad" was the best book. I never heard of Lalji prasad before ... so I went to search for it. It seems there are several mathematics books written by this author, which I assume must be popular in India but not elsewhere. I wanted to add a comment to the answer, asking the author to add the title of the book (I suspect is Real Analysis – K. K. Jha / Lalji Prasad) but then I got the message the answer was not eligible for comment on the basis it was either deleted or closed. I do not have enough reputation to find that answer. I can not tell whether the question was deleted because it was very short but there were other answers that just listed some books (not by title) but rather by author name. I may be wrong but it seems some maybe too "trigger happy" to delete legitimate answers.
I deleted that answer, but I do not believe that "Western bias" came into my decision.
In my view the answer itself was woefully inadequate. Without prior knowledge of this author one must resort to Google (or some other search engine) in order to discover anything further about the unnamed intended text. This may be worse than a link-only answer: at least a link would presumably take you to the relevant information without any further effort.
Recommendations are tricky to get right. Where is the line between an acceptable answer and VLQ-flaggable?
- User read the book and supports the recommendation by explaining why it fits the objectives stated in the question
- User read the book and offers generic praise of the book.
- User read the book and offers only some combination of its title/author/link.
- User searched Amazon for keywords in the question and posted some titles/links that showed up.
I would consider 1 an acceptable answer, 2 a downvote-able answer, and 3-4 as VLQ-flaggable answers, barely distinguishable from spam (and completely indistinguishable from each other). But that's just me. By the way, there was an earlier discussion along these lines.
Another aspect to consider is that it's hard to judge the adequacy of an answer when the question is insufficiently specific. Quoting Shog9 (who else):
when the question is: "Suggest me some tutorials where i can learn quick" [...] it's still a bad answer - but when the question is kinda asking for bad answers this is to be expected. Close or flag the question and move on - this is much more efficient than hanging around to babysit it by deleting every crap answer that inevitably shows up.
Many book recommendation questions are vague to the point where the answers and their upvotes mean as much as a popularity contest. Everyone feels free to post a book they like, and others upvote if they also had positive experience with the book. The result is a poll-type question from which one may be able to tell which books are better known among Math.SE users, but not which books are better for any particular use.
If I had my druthers, book recommendation questions would be required to contain a minimum set of features:
- topics covered
- intended use (course supplement, independent study, teaching a class...)
- maximum price (if it is a constraint)
and answerers would be required to demonstrate that their recommendation meets those criteria.