We already have detailed guidelines for identification requests (and an older meta post for this type of questions. From my point of view, the issue is not in the guidelines but in new users not being aware of them. When new identification requests are reviewed, we point (new) users to the guidelines in the tag wiki. This usually helps.
Closing a question too quickly can turn new users away from the site. Among the identification requests that have been closed, the reasons have been related to site scope, duplication of existing questions and, at list for this question about a short story involving a painting an apparent lack of detail.
The requirement that the question is about a book that exists is rather hard to check. How would a question asker prove this without actually identifying the work (i.e. naming author and title) they are looking for, thereby making their question pointless? For the potential answerers, how would anyone be able to prove that a book does not exist? Even in the rare case where a question turns out about a fictitious work, providing a reliable source for this would actually constitute a valid answer. (Think of a question that turns out to be about one of the fictitious works that Stanisław Lem reviewed in A Perfect Vacuum. An answering summarising the main points of the corresponding chapter in Lem's book would be a valid answer rather than a reason for closing the question).