🚀 go-pugleaf

In article <9208272105.AA12700@news.cis.ohio-state.edu> Marc.Ringuette@daisy.learning.cs.cmu.edu writes:
>I presume we all agree that we should move towards some form of secure
>email and some means of exchanging public keys.  There are a number of
>other cryptographic services we could ask for.  I presume we do NOT all
>agree on exactly what form they should take.

	Movement has been underway for some time now.
Privacy-Enhanced Mail (PEM) should appear in the Internet within the
year (famous last words :-).  Older RFCs for PEM are online.  A newer
set was just/is about to be released.

>How do you think we should take advantage of cryptography in the next
>generation network?  What tools would you use?  What organization?  Is
>there anything you want particularly to avoid?  I think it's important
>for us to develop a more accurate vision of what it is that we want.
>Otherwise, how will people know what to implement for us?

	All good questions.  The IETF has several groups actively
working on Internet Security, some under the Security Area and some
under other areas.  

Ran
atkinson@itd.nrl.navy.mil

I think demonstrating the delivery of postings is overkill, as well as
being difficult.  Given a moderator for a newsgroup, one could implement
a simpler scheme as follows:

Let	M(n) be the n-th posting to the group
	D(n) be a digest (e.g. by MD5) of M(n)

When the moderator accepts a posting M(n+1), he appends D(n) to it, signs
the result, and returns a copy to the poster.

The poster is now in a position to prove that he submitted the successor
to M(n).  In addition, any reader who receives M(n+2) without receiving
M(n+1) will detect the omission.  (What he does about this is up to him,
but at least he knows about the problem.)

If the moderator signs two different postings as successors to M(n), and
either author gets ahold of a signed copy of the other posting, then he
is in a position to discredit the moderator.  If there is any means (even
non-deterministic) of making public signed postings (e.g. posting them to
other newsgroups, uploading them to random BBS's, printing them in the
Personals columns of newspapers, etc.), then such inconsistencies are
subject to detection.

Furthermore, if the moderator does fork the newsgroup in such a fashion,
the appended digests prevent seamless merging of the resulting groups.

It remains the case that the moderator can refuse or alter a submission
to the group, but this is traditionally one of his prerogatives.  (Note
that having the moderator sign D(n) without seeing M(n) is unnacceptable.)
If he alters or fails to sign it, the poster will of course know.  If he
signs and suppresses it, he compromises himself as soon as he signs the
next posting.  Of course, he can refuse all postings thereafter, but this
is again his prerogative.

Finally, a moderator must still be chosen; but now his performance (or,
rather, his performance modified by the reliability of the newsfeed) can
be effectively monitored.  (If you miss a posting, I guess you query
archive servers and/or the moderator for it.  If they don't reply, I
guess you start complaining in other newsgroups.  If every moderator
refuses your postings, I guess you get paranoid...)

--
Jeff Keller <keller@cse.ucsc.edu>	CIS Board, UCSC
"If X is the answer, what was the question?" --David Fuller

This thread should really be called "secure email" not "secure
network" since the poster really only commented on email and the
latter has all sorts of other implications.

The more I learn about secure email the more I'm convinced we're going
to have to try it a while before we can wring the bugs out.  One can
make a laundry list of features one might want, but I don't think
there's a good basis for doing a cost/benefit tradeoff on the various
alternatives. Until we've had some pilot systems fielded and get to
'kick the tires' for a while, the technical community isn't going to
have the experience needed to make good engineering decisions.

Rick.
smith@sctc.com        arden hills, minnesota