Great question, and I look forward to the answers that this post will receive (and excuse my lengthy response).
Worse, however, is the fact that I feel like I'm always pointing out other people's work, and -- since your email has compelled me to search through some old bookmarks and enter a few, new phrases into a search engine -- that's exactly what I'm going to do at the onset of my reply.
So, with my poor prelude out of the way, I'd suggest that you first take a look at this online paper by Ben Goldman and the excellent links therein (which I just discovered minutes before):
I'd further suggest that you pay particular attention to the discussion on Matthew Kirschenbaum's book, "Mechanisms," in which Goldman mentions that "[Kirschenbaum] describes wanting to know more about the history of these files [that he found via the finding aid], and to see them in their native software environments" (rather than just downloading them from DSpace).
And then, in my bookmarks, I have a link to an EAD version of those very same "Michael Joyce" papers (though I don't know if it's a current version, as I couldn't find that same EAD file on the Harry Ransom Center website itself). Here's an example from the <dsc>, where the <c01> tag is the first series in the collection:
Afternoon, a story
So, in this case, you can see that they've opted to arrange the collection by series, and then at the item level (c02) they provide a title and a link to their secured DSpace instance (you have to be in the physical reading room to access these materials, I believe).
Now, as for how I might do something like this, I've been thinking about that for a little while, but I don't have any fully born-digital collections at my institution to work with. Nevertheless, I think that it would *completely depend* on the type of materials and how they were transferred to the archival institution. In the case of something like the above, I would have to take Matthew Kirschenbaum's advice and imagine that it would be far better for a researcher to be able to access a virtualized desktop of those materials (which is probably much easier said than done in practice).
Barring something like that, though, this is how I might advocate for the <dsc>(s) to be encoded:
1) Scrap the series (possibly. Or, if I applied series to the dsc, I would still permit a user to re-arrange the materials in the dsc by ignoring how they have already been grouped by series/subseries)
2) I would use the <dao> to contain the link and the unique ID for the item, but probably for nothing else.
3) I might also use a controlaccess list to provide an extra-level of faceted navigation in the output (but still permit a user to re-arrange the documents by ignoring those subjects if they wanted).
So, with that in mind, I might start encoding a born-digital collection pretty flatly, like so:
<unittitle>Some digital "papers"</unittitle>
<materialspec type="pronom">XML (1.0)</materialspec>
All of that said, I'm interested in what others have done when presented with those sorts of collections. I also have other ideas for a few ways to output a container list like that and to still keep it "interactive," but I've yet to really work on it to see how well it would actually work.
From: Encoded Archival Description List [mailto:[log in to unmask]] On Behalf Of Jay Burton
Sent: Monday, January 11, 2010 5:55 PM
To: [log in to unmask]
Subject: EADs for "Born Digital" Collections
I am writing to learn of others' experience in creating EADs for materials
and collections that are "born digital."
1- How do you arrange your <dsc>? (Do you impose a Box/Folder structure on
the material or do you have some other organization?)
2- How are you setting up your <dao> and/or <daogrp> links to the digital
3- If anyone would be willing to share a basic EAD schema showing their
approach to "born digital" materials.
Thank you for your help,