Idea for Finding Backup Copies: Embedded GUID

I posted the idea below on the forums at Steve's Digicams, and got a great suggestion! Now just need to see software developers implement it!

________

Like almost all semi-serious photographers, my first step, after downloading photos, is to make long-term backup copies. New when I say "long-term", that's just what I mean. These aren't backups to revert just in case I mess up in Photoshop when editing them over the next few days. These are backups to keep forever, to cover any unforseen eventuality. The most likely one being, 10 years from now, I look at a photo and think "Wow, could have been really good, too bad I sucked so badly at PhotoShop back then".

But I digress. Making those backups is not quite my first step. Becaues, for the backups to be useful, I have to be able to FIND the original for the edited version in hand. So, the first step needs to be naming them. THEN I make copies in my "Originals" folder.

Most of the time that works well enough. But I would really prefer to be able to make backups immediately, before I name them. One, because naming is slightly inconvenient for those situations you are in a hurry--you know, you want to edit and print a copy of your child's birthday party to present to each of the attendees on their way out, etc. Two, because sometimes, as I am editing or reviewing already-named photos, I decide to change the name. Three, in extreme circumstances, with the right cropping, an original could yield two different edited photos.

So, the bottom line is, I would like to be able to trace the "parentage" of a photo in a more reliable way than by filename. I have a theoretical idea for doing so, and wondered if anyone else here had thought about this problem, or heard of ideas for addressing it?

Before I get into the details, I want to make explicit--I am not looking for any form of versioning, nor am I looking for any form of digital watermarking. I just want to know that edited photo A' derived from original A.

So anyway, my thought is--wouldn't it be great if a GUID (basically, a unique id) were embedded in the image file? Ideally, as part of a new, backward-compatible version of the JPEG standard (and others). Then, for any given photo, I could search for its parent (actually, it could be anything--parent, sibling, grandparent; the point is, they would have the same GUID, and in my scenario, that would be the parent).

It isn't hard to imagine how software utilities could enhance this capability--batch searches, etc.