[1] Note: I don't think that the usual argument against allowing link-only answers (which is that the links may change or disappear) really applies to requesting a reference to academic literature, which should be constant and findable in the future.

For example, say I just started writing a paper about a process that is remotely related to Langmuir circulations in the upper ocean. I may have some basic knowledge about them, but I want to read further and I want to read from the beginning. Google and Wikipedia will only get me so far, and crawling down the reference-chain is time consuming and tedious, especially once you reach the papers that are several decades old. Having a place to request a list of most relevant peer-reviewed articles on the subject would be of great use. Of course, this assumes that there would be a user on the website that already has more thorough knowledge on the process. Thus I can see this feature working well only later down the road when the user base gets more saturated with experts.

If you have a few people really committed to maintaining the answers to such questions regularly, it could work. But it's kind of a thankless chore with little obvious return.

Maybe it's okay in some cases where the reference is known to be authoritative and stable. But in those cases — is Stack Exchange really even needed?