Eric Hellman has an interesting post in the current discussion on Web4lib about relative approaches to search by libraries and Amazoogle. He suggests that libraries should focus on browse, and leave search to those that specialize in it:
I think that libraries should consider returning to their historic roots that have nothing to do with "search". Forget search- a billion dollars says that Google and Amazon will do search way better than any real library on the planet, and libraries can now leverage these searching capabilities in very real ways.I think that the browse message is an important one. We really should be doing more to release the value of bibliographic data in ways which exploit the structure we have created.
What libraries CAN do with their "rich stores" of data is to facilitate browsing- which is what libraries have always done well. Please take a look at today's very best digital library software package - iTunes. It's modestly priced. It manages collections using xml and leverages large stores of remote metadata. There's a small search box that you might not even notice because the collection is so accessible via browsing. That's worth emulating. [Web4Lib Archive: [WEB4LIB] NOTAmazon OR Google as a "corporate" role model: iTunes]
At OCLC we have been looking at a dual approach: we are exposing metadata for search through the major search engines in our OpenWorldCat program. When a user clicks on a search result they come back to our rendezvous page, where they have various onward options. One of our ambitions over the coming few months is to create richer browsing options at that stage. We want to really release the value of the accumulated investment in bibliographic data by making it work harder. At the moment, if you come in through a search engine, you can click on subject headings, authors, and related editions. We want to expose much more structure, and make this a richer experience. This builds on prior work, including FictionFinder. In this way, we hope that we can combine effective search and browse features. Put search where the users are, and offer them a rich browse experience when they click through. The screenshot is a work in progress: it is an actual response to a query against a database. Note the tabs: we are also experimenting with our MetaWIKI infrastructure to capture user contribution. This is really a research activity at the moment, but we hope to see some pilot activity later in the summer.