It is very well established (throughout Stack Exchange) that "list questions" are "not constructive". As indicated by the votes (caveat: I just upvoted this answer) for this answer for limited list questions on the on-topic question, limited lists are at best marginally on-topic.
Given that questions asking for the single most X of Y are considered on topic where X and Y are not subjective (e.g., questions about the earliest example), are questions of the N most X of Y on-topic?
- Ten most popular (by number of sales) novels
- of all time (answer will likely require updating)
- of some particular time period (possibly of interest to someone writing about geeks in that time period--at some point this presumably becomes "too localized")
- Ten most recent novels by different authors dealing with anarchism (along the lines of The Dispossessed)
- Ten least popular (by number of sales after two years in print) novels by Nebula- or Hugo-winning authors
- Obviously, the answer would change with time as new novels and new award winners come into existence (and novels gain late sales), but would probably be reasonably stable.
I think such questions would not be "not constructive" (i.e., they have objective answers of limited size). However, this category seems likely to encourage questions that are "too localized".
Like the "earliest" type of questions, such would also tend to generate multiple different answers, but (worse) each answer might be partial, encouraging a collection of partially redundant answers. This would tend to give the questions the appearance of being "not constructive".
If this category of question is considered on-topic, it would seem desirable to have a defined number to be used for such questions (to avoid the problem of "the previous question only asked for the most N, this question asks for the most N plus M").