I've been trying to avoid commenting on this, because it doesn't seem
useful to argue about this when PEM is being deployed. But this is
getting out of hand.
Our customers (which are businesses, not individuals, and many of which
are Fortune 1000 companies) have expressed a need for authentication,
ahead of privacy. (In fact, they are also interested in authorization,
along the lines of what Bob Jueneman is proposing, but in a much more
structured fashion. But that's outside the scope of PEM.)
The NIST OIW (and EWOS, their European counterpart) went through this
exercise with X.400 security, and came up with the same result:
authentication and integrity were the most important services, followed
by confidentiality. This is yet another large constituency, whose
members consist of vendors, carriers, and large users, who seem to
think differently from your poll results.
PEM can provide confidentiality, so your only problem appears to be
that PEM doesn't provide enough anonymity. If the PERSONA mechanism
doesn't provide what you want (and I don't understand why it doesn't,
from a technical perspective), you are free to define another PCA with
a more suitable policy, or use out-of-band key distribution methods
a la RIPEM or PGP. I can't believe that, in a *business context*, you
would believe an uncertified public key arriving in a message, though;
this seems to not really be "prudent business practice." (My priorities,
obviously, are with business rather than individual users, since that's
where the money is.) You are certainly welcome to *participate* in the
next round of PEM enhancements, so I would suggest working with the group
instead of being at odds with (some significant percentage of) the
members.
PEM has taken 6 years to deploy (mainly) because of the need to
accommodate the need for a scalable infrastructure for authenticated
public key distribution. If authentication is to be provided *at all*,
this infrastructure is needed. I've been on the mailing list for
about 4 years, and the changes to RFC 1421 have been minimal compared
to the changes to RFC 1422. (I would hope you are somewhat concerned
about scalability, if you hope to sell a lot of product and/or
interoperate with the rest of the world.)
Regards,
Rich