Ned,
The issues that arise as one tries to associate sematics with
even simple signed objects are non-trivial. Addison Fischer has
written some very good, pragmatic papers on these issues, in the
business arena, with a focus fiduciary applications. The more general
context we are discussing here, with annotations and timeliness of the
annotations is even more complex. I suggest that we not try to solve
this problem until we get the syntax straight.
By the way, there are alternatives to a trusted third party
archive server. A timestamp notary is one example, and Bellcore has
developed a clever technique that uses one-way hashes, but not
signatures, for the timestamp registration function. I think the goal
for PEM and MIME-PEM is to provide a framework in which these
anciliary facilities acn be invoked, but not attempt to put too much
of the semantics into this set of standards.
Steve