null
[XMLHack] Amazon goes REST, Google goes SOAP. Without having seen the DTD, and based solely on Simon St.Laurent's article, it seems to me that a WSDL could be produced which uses http:binding verb="GET" and soap:binding transport="http://schemas.xmlsoap.org/soap/http" style="document", much like the default mappings that .NET provides today.
So: both company's protocols are layered over HTTP. All verbs are stateless. Both are based at least partly on XML. If you accept Roy's interpretation of REST as a strictly an architectural style, both are REST to various degrees and both are request/response.
Amazon's interface may be more suitable for traditional approaches to caching (though requesting that developers limit themselves to once per day per query effectively nullifies this). Amazon's interface may also be simpler in the starting-from-zero sense. But as Google's interface builds upon infrastructure that is widely deployed, the net effort to get started is may be less for many. Significantly less.