Does Browsing Beat Searching?

Craig Bellamy points to Xfer, a for-pay reference tool that puts browsing in books before searching (as in Google). Xfer offers a free trial but except a useful concept map visualization of its content (see results for search word: collaboration) I found it hard to see the advantages of browsing through the proprietary reference books and dictionaries made available here to using Wikipedia. In Xfer I did browse through the Penguin dictionary of sociology and found terms like political participation and Durham's social fact. Browsing in a reference tool like Xfer has a similar effect to finding a book that is placed next to the one you looked for in a book store. You could not be find this with Amazon's autolink feature "people who bought this book also looked at..." But such random sampling or browsing, which often serves as argument in favor of the physical bookstore, can also be achieved by database-driven online encyclopedias that can feature randomly chosen entries on their front pages, for example. What is the advantage of Xfer to, for instance, Listible or CitULike (the del.icio.us for academics)? Accuracy would be an argument perhaps. But it'd have to be seen if entries are in fact less faulty. The comparison between Wikipedia and Brittanica only showed small degrees of difference in terms of accuracy. Finally, there is the fact that "open," participatory reference repositories are more up to date. I'm not convinced that browsing beats searching as a white paper on the Xfer site claims.