Looking at it in more detail as I research and write the piece I've found the API an extremely interesting piece of work. The query URI structures used in the XML/HTTP version of the API are very interesting, as it's clear that they've been designed to be both relatively easy to design and use and to map relatively easily to the backend SQL queries that Amazon will be using on its databases. A query URI for a search that's exploring just one area of the Amazon site looks like this:
http://xml.amazon.com/onca/xml?v=1.0&t=webservices-20&dev-t=[Developer's ID goes here]&BrowseNodeSearch=[browse node goes here]&mode=[product line goes here]&type=[lite or heavy] &page=[page # goes here]&f=xml
You can see just how the resulting code on the Amazon side will operate. It will first authenticate using the Developer's ID, place restrictions on the search scope using the browse node information, and finally chooses the type of document returned (along with the page number in the query response). This type of request will allow you to see the most popular items in a product class - so you could use it to deliver a regularly updated list of top selling books or CDs for a web site.
It's an interesting piece of work, and something I'm going to need to keep in mind as a design pattern for future KM work.