ISEGRIM is wonderful. For the last 30 years they have been describing Greek coins using Vacano's style and putting them into a database which now has 60,000 coins. It's great for identifying coins. It is also great if you think you have an unpublished coin -- ten seconds and ISEGRIM tells you it was published in some obscure journal in 1934. I imagine it provides a great start for die studies as well.
The downside is that it can only be queried in an unusual format. There is no form-based query, and no URL-based query. ISEGRIM also demands a password, which was important ten years ago when computers were slow and 60,000 was a lot of records to search. (I am a huge fan of Vacano's work and would like to publicize and improve ISEGRIM but I cannot contact Vacano and don't want to start writing front-ends and stuff without permission.)
There is a lot to be for your suggestion of structured deep search for Provincials. The problem is, of course, data entry. It would be hard to teach Vacano's format to folks who just want to look at their collections and maybe sell a few coins.
Better, I think, would be something like LibraryThing that was integrated with ISEGRIM and WildWinds. For example, instead of typing in descriptions for each coins we could just type in a Von Aulock or RPC or SNG Copenhagen reference, and Tantalus would query ISEGRIM for the coin, filling in all the other fields.
Obviously this is a lot of work. Although LibraryThing is free and can offer this for books, LibraryThing has the pre-existing Z39.50 search format, so a lot of the work was already done for them.