I’m curious about the “preharvested” aspect of it, as that makes it seem, to me, artificial, but again, I’m new to this product. People use Google because it works well – I could see providing almost a Google Custom Search Engine, initially limited to only the library catalog, subscription databases, and selected websites, a useful approach. But you’re right, that only covers a segment of the information world, and may not scale up in real-time.

The challenge is in creating a tool that encompasses all media sources, but filters out the chaff. Like you said, manually limiting the search to preselected resources (I like your “bigger silo” phrase) seems to only address a portion of that.