On Wed, Aug 06, 2014 at 07:49:13AM -0700, Dave Crocker wrote:
Of the various terms that were originally suggested, the one that has
the simplest, clearest and most useful meaning is "best effort".
Opportunistic is clearly a much sexier word, but the continuing lack of
coherent community understanding of its meaning makes it problematic. At
the least, it means that it will not be particularly intuitive for the
rest of the world.
Perhaps you're projecting your own surprise at the meaning of the
term onto the community at large.
It's always self-comforting to choose an ad hominem counter-argument.
Please try to refrain from repeating that indulgence.
Apologies, no ad-hominem attack intended. I am trying to say that
it seems that the problems with the text are far from universal, but
if it fails to reach some people, perhaps we can do better, without
losing yet a different group of people...
Is the problem at all partly the possibility that you're bringing
a prior conception of what "opportunistic security" might mean to
the table? Or is it that I am simply failing to explain the term?
Either way, now that you've seen various formulations of the idea
of a range of security levels, dynamically tailored to the purported
capabilities of the peer, what would you like the draft to do
differently? Would all become light if I in fact added something
like this at the beginning of the introduction:
Opportunistic security: A security protocol in which the ability
to communicate is prioritized over absolute security. This is
achieved by replacing a fixed all or nothing security level
expected of all peers with a range of security levels, such
that the minimum acceptable level is tailored to the purported
capabilities of the particular peer system. Provided the
communicating peers are not misconfigured to promise greater
capabilities than they can correctly deliver, security does
not get in the way of the ability to communicate.
While, in order to address Pervasive Monitoring (PM [RFC7258]),
opportunistic security aims to always achieve at least
unauthenticated encryption, with legacy protocols or infrastructure
it may be acceptable to fall back to cleartext with peers that
are not encryption capable.
Opportunistic security protocol designs are strongly encouraged
to strive for more than just unauthenticated encryption.
Designs should, if possible, enable peers to advertise (in a
downgrade-resistant manner) support for authenticated communication
to thwart active attacks. When peers advertise such capabilities,
it is expected that opportunistic security protocols will
require greater security with those peers, and will refuse to
communicate when the expected security level is not achieved.
--
Viktor.