ietf-smime
[Top] [All Lists]

RE: Nonrepudiation and what to do about it

1999-08-18 14:09:07
I agree in part and disagree in part.

First non repudiation must be an attribute of the signed document.

For this reason I don't see a need to modify the NR semantics of the
certificate - if indeed this is possible. I don't see that the working group
can increase the legal weight applied to this bit at this point.


The second issue is that NR is not a binary attribute. The issue for the
recipient is whether they can trust the degree of certainty in the assurance
provided by the signature. This will inevitably be a complex function of
many factors, not least the CPS governing issue of the certificate in
question.

If anyone thinks they can compress a CP or CPS into one bit...


All the NR bit can do is signal 'the circumstances of issue for this cert
are such that it is not intended to create contracts'.

Since NR is complex attribute the most appropriate format for the document
indicator would be an attribute with a URI as the parameter. The URI could
then identify the governing semantics.


Again, I think folk are overestimating the extent to which complex legal
issues can be eliminated through bits on the wire. I see an issue raised and
a 'solution' advanced which does not solve the problem identified as far as
I am concerned.

Three things are certain: death; taxes and lawyers.

        Phill


<Prev in Thread] Current Thread [Next in Thread>