- Title. Because everything has a name, right? That's what we do, we label things.
- Description. Because I want to know a little bit about your resource, especially if your resource is one of many in a list of search results.
- URL. I need to be able to find your resource. Perhaps this should be 'location' rather than just URL because some of us still use stuff that's not on the web (really, some still do).
- Controlled vocabulary keyword(s). I need to know that when I say potato, you say potato, and when you say tomato, I say tomato.
- Copyright statement. Because very often you don't own the thing you created, your institution does, and I'll need to know who I have to ask to use your resource. And when you do own your resource, I need to know that, too.
I think you can do a lot with these data. "Like what?" Alan might ask. Well, for me, there's really no point using any metadata unless you intend to share your resources. If you've got 3 resources to share, just give them to me and be done with it. If you've got more, these basic pieces of data that describe your resource will probably help me find them when I search a database. A controlled vocabulary keyword would help me find you resource more effectively (especially if you give your resources non-descriptive titles and don't use the same words that I'd use to describe the resource). The copyright statement is just good manners as it makes it easier for me to work out who I need to ask to use your resource.
The first 3 fields in my list or 5 fit nicely with RSS and are the kinds of metadata that are so common we'd hardly even think of them as metadata. The copyright statement is something that's also pretty straightforward, or at least it's something that we understand to be important. Although many people don't understand who owns the resource. The only difficult think is keyword(s). These can be optional of course though searches are much more accurate when they're used. And using a controlled vocabulary to assign your keywords saves so much confusion. Pity there are so few agreed controlled vocabularies.
As for Alan's other question about why use OAI, I can't think of a convincing answer. Although I do know that one of the goals of using OAI is the ability to exchange metadata records rather than the object itself, which makes sense from a bandwidth point of view. I think an RSS aggregator approach would be just as effective, not least because RSS aggregators are easy to use and readily available. If you used RSS 1.0 then you could even use Dublin Core module to include the copyright and keyword fields in your feed.