It turns out that cutting inventory by 10% and facing books out (instead of just showing spines) increased their sales by 9%. This is counter to Long Tail thinking, which says that more choices and more inventory tend to increase sales.
This highlights a real point that online and off-line thinking and merchandising are inherently very different. While the online world is currently focused on search as the primary navigation method, the off-line world is still all about browsing and discovery. It makes perfect sense that when there is less information to consume in a book store, where browsing is the primary mode of discovery, sales will go up. Make it easier to discover new books by facing covers out to the consumer and obviously sales will go up.
On the web, however, the long tail is still king. I doubt that Borders has decreased its online inventory in any way. In fact, I'm sure they try and increase it so that any possible search gives them a sales opportunity.
What the internet does need is a hybrid approach. Sometimes you just want to browse and discover. Search provides a poor interface to this type of discovery. Companies like SearchMe are working on this. I'd like to see more sites embrace this type of interface. I want it for NetFlix's "new releases" (something akin to browsing BlockBuster's new release wall) and it would be also great for clothing sites. Searching is great, but sometimes an experience similar to flipping through a catalog is the right approach.