I have written before about the advantages and disadvantages of including an owner override feature into trusted computing architectures. In the earlier posting, I argued that an owner override solution is "over-inclusive" because it limits the functionality of TC. TC becomes useless for DRM applications. But this does not only affect DRM. It also affects distributed computing applications which depend on a secure authentication of the client applications so that malicious users can be identified and excluded from the system. I just came across some documents that describe this problem for distributed computing in a much more detailed way. See here, here and here.
