But a few observations:
I knew that they would be offering fairly direct requesting for items held in the Summit consortium. This works pretty well and, interestingly, even works for me as someone at Lewis & Clark. Even if I come across an item that is held at the UW libraries, I am offered the ability to request it on Summit.
One thing I don't really care for is the display of holding libraries closest to me that is shown when I've selected a book...this just seems irrelevant and confusing to me when I'm a member of the Summit network.
The relevance ranking seems to be based a great deal on how many libraries hold an item. This is definitely a good direction to go and could be likened to Google PageRank. But it doesn't always work well. When I searched for "Bend Oregon", for example, the top hit is an EPA publication: "Pressure and vacuum sewer demonstration project Bend, Oregon". I also got lots of references to government documents with the title: "Amending the Bend Pine Nursery Land Conveyance Act..."--it's held by like 189 libraries. (I think this offers some hint of the redundancy involved in acquiring and cataloging government documents across libraries when these types of documents are often rarely used and available openly on the web).
The book that "should" come to the top is probably "Bend, in Central Oregon", which is the main book ABOUT the city Bend, OR...WorldCat Local needs to work on that "aboutness" thing...not sure how. Perhaps they need to be doing more creative things with subject headings or somehow move govt. publications lower in the ranks. Or bring in circulation stats...those would quickly lower govt. docs in the rank.
Interestingly, WorldCat.org seems to have more sensible results when you search for "Bend Oregon"...perhaps they are using more of a public library relevance ranking system that doesn't report all those depository library holdings
Relevance ranking is not an easy thing to do, of course, and is much more than a popularity contest. It's funny how library holdings are a strange take on "popularity."
Another complaint: the facets for authors don't always work very well. Seems like corporate authors without much meaning ("United States") often float to the top. Also, it's hard to browse around the publication date facet.
I don't see many openings for mashups and remixability...no RSS feeds or apis advertised. I have hopes that OCLC will open things up.
The other question that's hard to answer is the degree of local configurability. Generally speaking, it's a pretty busy display when looking at a particular title, but it would be nice to think it could be configured locally and streamlined.
I applaud the inclusion of articles and hope this expands. Bringing on board large aggregations of articles could be great. Better resolver integration would be nice as more articles appear.
Overall, this is a pretty good first shot at an OPAC product by the behemoth OCLC.