Nelson brings up some key points on conflicts inherent
in the Mozilla situation.  I think he's right, but also,
I don't see how to necessarily solve or deal with these
issues easily, unless some treasured assumptions are
challenged.  Some thoughts follow...

The core of the problem is that Mozilla is working for
the CAs, for free.

Microsoft, which also distributes CA root certs inside
its browser (and accepts CA-signed certs) charges each
CA for the privilege.  Thus Microsoft can cover itself
for any support issues it has to deal with.  It can also
choose to set its fees at different rates, depending on
how broken each CA is.  Even better, Microsoft can bend
its own offering to the CA that pays the most money, so
if some CA wants broken certs, they can have them.  For
a fee, why ever not?

Mozilla does not have that advantage.  Neither has it
the incentive to promote one CA over another, nor all
CAs over the OpenSSL self-issuers.  It simply hasn't
got the easy money choice, so it has to decide what is
"good" in certs, some other way.

Hence Nelson's dilemma in deciding how to handle a
cert that isn't laid out to standard, but is usable
none the less.

About all that Mozilla can do is try to present each
cert, in its best light.  If a cert is non standard,
then say so - on the browser, and on the website if
there is sufficient information, or a test page such
as proposed by Jean-Marc.

I fully agree that if a CA gives too much trouble, then
MF should consider dropping it - but I wonder how likely
that is to be a "fair" process?  We can easily craft a
set of rules when it comes to CACert, but what happens
when Verisign starts being hard to deal with?

It all comes back to the little lock symbol.  The one
binary decision, good or bad, all compressed into that
one lonely icon, is what makes it so difficult.

What can't be done is to hide all the available information
and try and mash it all into one tiny little lock symbol.
At some point something has to give, and there will be
no easy answer.  There just isn't any way to compress all
those difficult choices into one binary decision, and be
confident that you are delivering "trust".

The only way forward is to engage the user in the protocol,
as only the user has the interests best in mind.  This has
been tried, with good results, in Mozilla, even:  the
paper by Ye and Smith presents their experiments and
results with modifying the browser to deal with security
attacks.

     Ye and Smith, Trusted Path for Browsers,
     11th Usenix security symp, 2002.
     http://www.cs.dartmouth.edu/~pkilab/demos/countermeasures/

That's not the only way.  But, it is clear that the only
way to deal with the complex security choices facing
browsers is something to do with presenting the user with
more information.

iang


Nelson Bolyard wrote:
In the past, the largest single support burden for NSS and PSM
developers and QA folks has been the relentless stream of "bugs"
filed by OpenSSL users who made certs that don't work, and who just
assume that the problem must be a mozilla bug.  This has primarily
been a problem with certs from CAs not in mozilla's trusted CA list.

There have been rather few problems with certs from trusted CAs,
which is just as it should be, and is how it must remain.

It simply MUST NOT become the case that everyone who has a problem
with a cert from a trusted CA receives their customer support from
mozilla.  CAs cannot rely on bugzilla.mozilla.org as their bug
tracking system.

So, I think mozilla foundation's trusted CA meta-policy needs to
say something to the effect that a trusted CA cannot create a
support burden for mozilla and its developers.

Further, I believe the policy needs to state that trusted CAs MUST
provide all technical support for (a) their "customers" (the parties
who obtain certs from them), and (b) their "relying parties" (people
who rely on their certs, e.g. while web surfing).

Finally, I think that the policy should state that a CA whose certs
cause too many problems maybe removed from the trusted CA list.
We need to spell out what the threshold is, and it needs to be quite low.
_______________________________________________
mozilla-crypto mailing list
[EMAIL PROTECTED]
http://mail.mozilla.org/listinfo/mozilla-crypto

Reply via email to