Couple of things - there's a plugin called Extract ISBN, it looks inside the books for an ISBN and sticks in the ID field - if you use it (you can run it on multiple books) before doing the metadata downloads then generally speaking you'll get better results.
If you configure the Download sources to suit your 'tastes' you wont get as much rubbish - eg many folks don't accept tags from the likes of Google or Amazon, other's don't like B&N's summaries.
But its an inexact process especially when using bulk downloads
Having said all that, having something similar to the matching rules in the Find Duplicates plug-in (see attachment) would useful for bulk metadata downloads, even better if it could could flag books that didn't get their metadata updated
BR
ps : I suspect your request got pushed down the stack so that it fell off the first page faster than normal - its the sort of request that should have elicited at least one response - if only from me.
