It seems, though I haven't researched this, the "nesting" issues would be described in the WSDL document defining a web service. At least as far as providing or denying authorization to distribute values to unknown entities.
Also, while I understand how the semantics of information can become corrupted along a chain of web service consumers/servers (ie. the "telephone" game), in theory, each web service consumer has an implicit contract with its server. (using your model: Consumer has a contract with WS1. WS1 has a distinct contract with WS2.)
Thus, even though a web service may also be a consumer as part of its task to provide a service to the originating consumer, the first web service has a "contractial obligation" to provide its service to the originating consumer as accurately as possible. So for example, the issue of authentication, Should WS2 have a need for credentials from the Consumer? Or should WS1 just have a "contract" with WS2 that states that WS1 can share the results of any properties or methods it receives from WS2 with any if it's (WS1's) consumers? Yes, the question remains, where is this "contract" documented and how are these terms validated proactively. Perhaps this is what you meant when you stated "provided the design pattern compensates for potential problems" .
-jason
www.jjkpryde.com
