To expect the submitter to do this is ideal, but not scalable. (I know I've tried to add comments in articles as I've posted them, but gave up because of the number I submit. I do read them but now comment sparingly.) There is a real problem with generating volumes of information quickly. If you want speed, there is a limit on the amount of meta information a user can reasonably generate. If you want usefulness, this requires more time on behalf of users submitting.
Newspapers have this problem and solved it by training news editors to create catchy titles that neatly summarises the entire article in a small sentence - maybe even a few words. But not everyone writing articles on the web is trained in Journalistic techniques, let alone people who read this site. There is also the other problem when re-phrasing or changing titles of articles. Mutilating them into something un-recognisable from the original.
This is where the software has to work harder. By supplying users with the tools to classify, sort and value the submissions with tags, comments, summaries (already have ranking). I'm thinking here of delicious where tagging has allowed users to sort quickly by tag, then if they want to add more data themselves (descriptions, summary). The other way is to gather meta data from the source document (tags, microformats) where possible (un-likely as it takes time to process documents to generate useful things like summaries - even longer for meta data like microformats). Now the question is, could "pg with news.yc create some tools to help users or process new submissions extracting meta-data"?