I have been in a couple of meetings recently where people have been talking about the attraction of the simple, single box search as the ultimate goal. To this, my response is 'yes, and what else?'. In Google's case, pagerank has been the principal 'what else'. Going forward they have interesting questions to face about how to rank materials which do not fit the web-page model. The improvement of search, and the improvement of ad placement, are a major focus for them, as indicated in the much discussed Google analyst day presentation [pdf]. A simple box is one part only of their formula: good results and good ads are necessary for them.
Interestingly, in Amazon's cases their results are their advertizing. Each result represents a potential purchase. This is one reason that it is useful for them to make APIs to their results available. And it is one reason that their presentation strategy is to offer a rich texture of suggestion on their results pages. You are hit with many hints about potential items of interest, and this data is created in multiple ways (mobilizing the edge of reader contributions, mining the 'intentional' data from user purchase and browse patterns, mining the text of books). An Amazon page has many 'suggestions', using a variety of approaches.
I think we will see more 'simple search' but supported by smart results and rich browse. Whenever somebody says that people need a simple single box to search, try asking 'yes, and what else?'.