Reverse engineering Google and Amazon

I wonder if it would be possible to study the output of Google’s search results and Amazon’s recommendations and determine the algorithms & methods they use in producing that output. The key components in both systems are the relationships between the items (Web pages in the case of Google and merchandise in the case of Amazon) in the databases, but the details of all the different factors that make up those relationships are closely-held corporate secrets. Precisely how Amazon makes their recommendations might be very valuable information for other online shopping sites to have, especially large offline retailers that are looking to establish themselves online and have R&D money to burn.