Thursday evening quarterbacking...

Brad Knowles (brad@his.com)
Fri, 23 Feb 1996 03:55:50 -0500 

Folks,

    Since it's not Thursday morning (more like Friday morning),
I couldn't use Dave's subject unchanged.

    I would like to put down in writing my slightly modified
views on what I believe that our priorities are for my employer,
from a purly technical perspective.  This is based on the slides
I briefly presented or marked up (oops...), and without any
necessarily particular order:

        * Protection of installed base
        * Backwards compatibility
          (i.e., signed messages are readable by non security-aware MIME
           or RFC822-only MUAs)
        * Non-duplication of data
        * Support for both client and server models of operation
        * Strong encryption [1]
        * Exportable [1]
        * Encryption algorithms are unclassified and publicly reviewed [1]
        * Reference implementations are freely available [2]
        * Scale
        * Scale
        * Scale

[1] Important for me personally as a "card-carrying" member of
the EFF and other privacy advocacy or civil libertarian groups,
and I think also important to our users, whether they realise it
or not.

[2]  Mentioned at the workshop, but not actually written on my slide.


    As I mentioned at the workshop, we believe that MOSS (RFC
1848) and PGP/MIME end up as the two standards (somewhat
incomplete as they are, in our eyes) that are most closely in
line with our desires, and therefore rate "plus" signs on our
chart.  There is no change on these two from what I presented.

    PGP represents the installed base we are most concerned with
(most of our users are home users, and of the home users that
care, most seem to care about PGP), but I do not believe that
the existing PGP standard is where we (as either IMC or my
employer) want to go.  For that reason, PGP would rate a "minus"
(I had rated it a "nothing").

    The existing S/MIME specification violates a couple of the
above points, and to the degree that this standard is likely to
change to be closer to our criteria, I'd be willing to rate it
as a "minus" as well.  This is a slight improvement over the
rating I gave at the workshop, which was a "nothing".  However,
the availability of the RIPEM reference implementation is
something I consider quite crucial, however -- without that, I'd
have to rate it as "nothing".  And I'd still like to see the
reference implementation be exportable outside the U.S.

    I know very little about MSP, and know literally nothing
about MSP/MIME, so I don't feel I can judge yet what kind of tag
that we would attach to it, although I'd be inclined to say that
I doubt it would rate higher than a "minus", and maybe lower
(50/50 chance of nothing/minus?).  As I said at the workshop, I
used to work for DISA, and DMS was one of my favourite sacred
cows to attack.  I have a hard time believing that our
government (and especially our military) could accidentally
manage to do something right, but I am willing to be convinced.
Until I know more about the specifications, I'll have to withold
judgement and say that, by default, it still rates a "nothing".



    Now, I have some ideas that ocurred to me after the
workshop, so instead of overloading this subject, I'll send that
under separate cover....

--
Brad Knowles,                                       MIME: brad@his.com
    comp.mail.sendmail FAQ Maintainer      <http://www.his.com/~brad/>

The comp.mail.sendmail FAQ is at
<http://www.his.com/~brad/sendmail/>