You've found my brainstorm page! Feel free to comment on any of the ideas here.

What if all inline citations on Wikipedia were easily accessible from a searchable database, with automatic duplication sorting in order to determine the best-formatted version of a citation? It would make the task of referencing much less daunting to new users, as well as making it more convenient to do so in general.

Used in conjunction with the reference database feature described above, what if RS:N threads would be automatically linked to sources listed in the database? The reliability of a source could quickly be ascertained without the need for repeat discussion or awkwardly searching the RS:N externally.

With changes in technology and in corporate and academic structures, we will witness a whole new generation of link rot. WebCite provides a service to counter such effects, by archiving online sources on their access date. Wikimedia should offer a similar service, to be used in Wikipedia on a mass-scale for web citations performed within sites belonging to compliant organizations. This will ensure that sources linked to will always accurately represent the content being sourced.

A common problem I encounter is using a book source more than once for different pages; right now the best way is using {{Harvnb}} to point to a page, then to a bibliographic source. I propose that page (or any other attribute for that matter) be appended outside the second ref, i.e. <ref name=somexistingref append="page=25" />. For more thoughts on the matter, see this thread.

A simple way to automatically split list items into n columns, without having to manually count items.