I've been struggling to work up a "can-do" talk for an upcoming conference on video in New York, and at the same time musing about the things I learned and saw a week ago (in New York) in my whirlwind publishing conference extravaganza.

One of the things that I keep coming back to is the glue that Google has spread around the world that makes their site so sticky. Hardly the first one to do so, Nick Carr recently observed how the big get bigger in Web 2.0 (this is actually a theme that Tim O'Reilly has hit upon several times as well). Of course, this means that for things-that-libraries-care-about, like books and so forth, Google will increasingly be both a discovery and services platform.

Today, Google announced that they were combining all of their silos into their primary search interface. Despte the huge user interface challenges associated with this strategy, Jeff Jarvis observes in BuzzMachine that the impacts of this move will probably be pretty profound, and will almost certainly cause additional stickiness.

If you think for a moment about the incredible things that Google can do in their Book Search product - try running this book search on Drew Faust for example - then there are increasingly few places to "make a impact" in the areas of search, discovery, and access. One of the things that Google has not done, so far at least, is to fully widgetize Google Book Search. There is only a limited degree to which one can embed their page turner into external applications. The Random House widget (BISG presentation) has some very nice and appealing features that encourage it to be viral, and customizable by other users such as bloggers. The Harper Collins widget (IDPF presentation) also has some of this functionality, and I think overall widgets will see continued evolution for marketing, sales, and ultimately data integration.

I have to imagine that digital libraries should be aggressively seeking some of this kind of functionality too. For example, we should have lightweight DSpace and Fedora repository widgets that permit users to embed high value digital library content into other locations, such as blogs, and that provide basic functionality for those objects, like page turning, video display, and so forth. (I would imagine that Fedora's default action class would particularly well lend itself to such an applet). There might be embeddable descriptions associated with the objects.

One of the challenges for widgets is that they must be easily found, or their naming and creation obvious. Otherwise, they simply will not get used. This means that digital library widgets have to permit easy entry into the workflow of the collaborative user. Workflow insertion is a very difficult problem for content distributors - or DADs, as Mike Shatzkin describes them.

Libraries particularly have been very passive in content creation and distribution, at best adopting a "come and get it" mentality. This is not a winning approach. Instead, digital libraries need to be pushy, actively encouraging their content to be re-used, re-combined, and separated from its source. Widgets can help.