Steve Crocker writes:
It may not be what you intended, but since it's
clear that the present course of using DNs
unconnected to email addresses is a bust, and
since it's clear that we need a trust hierarchy
that is built from the bobttom instead of the top,
perhaps you really are suggesting we bring RIPEM
and/or PGP into the IETF. This may be the most
fruitful approach. Non-repudiation and other
advanced concepts can be left for another day.
We may have to agree to disagree, and strongly at that.
It is NOT clear to me that using DNs that are not
connected to email addresses is a bust. It IS clear
that the existing implementations that provide little
if any support for such a binding are not being received
very well. It is also clear that a Unix implementation
on top of a mail handler is not going to meet the needs
of an ever increasing population that is oriented towards
Windows or the Mac.
Perhaps one of the perceived virtues of RIPEM is that it
is much easier to install, works on a much wider variety
of platforms, and doesn't have to be integrated with a
mail handler to function, as opposed to current PEM
implementations.
It is also NOTat all clear to me that "we need a trust
hierarchy that is built from the bottom rather than the top."
With the exception of the IPRA which is not yet operational,
there are a sufficient number of PCAs available, and there
are CAs that are beginning to sign up. There is also
sufficient support for residential persons and persona
certificates that those concepts can at least be tried.
I agree with Steve Kent that self-signed certificates are
an oxymoron. If you intend to disregard issues of
nonrepudiation and don't care about the question of
scaling such approaches beyond your personal circle of
friends, why not just use a private key DES system?
There are certainly a number of DES encryption systems
available in various utility packages. Just call up your friend,
exchange the key over the telephone, encrypt your mail,
BINHEX or uuencode it, and send it.
I have been told by a number of people that there is a pressing
need to have digital signatures for nonrepudiation in applications
such as EDI. If the PEM community has made a mistake, it
is perhaps in ignoring applications such as internal forms
management and EDI while concentrating too much on academic
use of the Internet for e-mail. PEM certainly doesn't answer
all of these questions, but it comes a lot closer than RIPEM or
PGP.
I hope that after five or more years of hard work, that we don't
suddenly throw the effort out the window just when decent
implementations are about to become available.
Bob