[Cryptography] A PKI without CRLs or OCSP

Peter Gutmann pgut001 at cs.auckland.ac.nz
Thu Oct 27 21:51:21 EDT 2016


alex at alten.org <alex at alten.org> writes:
>Quoting ianG <iang at iang.org>:
>> It's all online now.  Need another design.
>I remember pointing this fact out during a cryptographic system design
>meeting back in 1996!!!

Lots of people tried to address this, over a long period of time, but it never
made any difference.  Watching the PKIX standing committee operate was a weird
experience, it was more like a group trying to formalize religious dogma than
create a technical standard.  Almost anyone with any practical, hands-on
experience either left or went into read-only mode (I was surprised at some of
the off-list responses I got at times, people who I thought had bailed years
ago were still reading the list but had given up trying to contribute).  What
was left was the same sort of people, or in some cases literally the same
people, who had made OSI the success it was (as someone once observed, and
this is an approximate quote "it's a good thing they're all tarpitted in PKIX,
imagine how much damage they could be causing if they went elsewhere").

Unfortunately I don't think it'll ever change.  The CA business model is, as
Ian has pointed out, the problem.  It's based around selling as many
organisations as possible a bag-o-bits, billable yearly, and neither the CAs
nor browser vendors have any interest in changing that.  No matter how clean
or elegant or functional your newly-proposed technical solution is, you need
to get the CAs and browser vendors out of the loop, and I can't see how you'd
do that.

Peter.


More information about the cryptography mailing list