Metadata for Manuscripts

There are lots of different kinds of metadata out there for library and archival materials. Some might say, too many kinds. There are different metadata schemes for every kind of material. There is Dublin Core, METS, MODS, even Marc for XML. All these metadatas are trying to solve an age-old problem with archival materials: they're too unique to have a standard applied to them. Books are easy; they all have title pages and authors and they're all wrapped up in neat packages that lend themselves to cataloging. Archival material, on the other hand, can be anything--and usually are. Paintings, bills of sale, letters, buttons, book manuscripts, musical instruments. The list goes on and on. And, especially with paper things like letters and bills, there is the problem of having so much paper on your hands that you cannot simply describe every single piece of paper as a single entity. So we group things, and then we catalog the groups. More or less. It's an inexact science.
At least, it was until the standards started getting made. Dublin Core and METS and MODS are all designed to help archivists catalog the things that are uncatalog-able. And now there's a new(ish) metadata standard: PREMIS.
PREMIS stands for something fancy, but at core, it is preservation metadata. Yes, now archivists can code not only information about the creator and the date and the content of a piece of paper, but also the physical condition, history of the physical condition, and any repairs that have been made to the paper.
Are we getting over-metadata'ed? Do we really need metadata, encoded into XML, that tells us if something is fragile and old? Can't we go to the paper itself and see what condition its in? Or this strictly for statistical purposes?
I've made myself a copy of the PREMIS manual--all 283 pages of it. So, we'll see what this is supposed to do/be. It can't be for the users of the material (which is generally who I consider metadata to be for), so it's not something that should be displayed to the user, probably. And a lot of archives don't even have the time to go through and process basic information, let alone preservation data. Maybe this is one of those things that is reserved strictly for libraries where everything is already processed and they have lots of extra time on their hands to go through and put in fuller, more complete information on each and every collection. Good on them, I say.