[Servercert-wg] [EXTERNAL] State or Province
sleevi at google.com
Thu Sep 5 12:10:35 MST 2019
On Thu, Sep 5, 2019 at 3:03 PM Tim Hollebeek <tim.hollebeek at digicert.com>
> So, there have been a bunch of extremely useful suggestions. Thanks to
> everyone for engaging constructively.
> I think Jeremy is right that we need to get to a MUST eventually, but I
> find Wayne’s argument about the value of a SHOULD compelling. We can
> revisit the issue once everyone has a chance to provide feedback about the
> compliance challenges in corner cases.
> Some people are rather skeptical of SHOULD requirements, but I’ve always
> found them useful for expressing best practices, and more importantly, for
> expressing goals for future requirements.
While I don't disagree with that, I think the Forum has shown repeatedly
that when the objective of the SHOULD is getting feedback from CAs about
the operational challenges, so that we can avoid a MUST that breaks things,
we never actually succeed at that.
That is, we only make progress when discussing a MUST, as we're doing now.
If we added a SHOULD, I have zero faith that CAs, consistently and as a
collective whole, maintain accurate logs about when and why they violated
the SHOULD. Some do, some treat the SHOULD as a MUST, but without that
uniformity, we don't make any progress or advance any understanding.
An idea that I continue to raise, for every violation of a SHOULD, that
actually helps ensure we make continued and timely forward progress, is to
require disclosure, to the Forum, for every situation where the CA deems it
appropriate to violate the SHOULD. This is objectively auditable, does not
require membership in the Forum itself, aligns with how we handle
legal/jurisdictional conflicts, and quantifiably and objectively measures
any complexity or challenges in ways that help develop better criteria.
> I think Doug is also correct that there are a number of large and
> important countries where the situation is less complicated where we could
> probably go directly to a MUST. I’m sure some people will be less than
> satisfied with non-uniformity of requirements around the world, but I think
> that just reflects the reality of the situation …
> The UPU is also an interesting suggestion. Apparently it has been
> discussed internally here as well. I’d have to look more closely at the
> advantages / disadvantages of each list. If there’s a better data source
> or standard available, it should be considered. ISO 3166-2 is simply one
> of the best known, and has the advantage that we already rely on ISO
> 3166-1, so hopefully that would reduce the risk of divergences or conflicts.
Perhaps, rather than focusing on the list, we should focus on the question
Joanna highlighted? Namely, what do we 'intend' for these fields? This
seems essential to any further discussion about the merits of which list
(which, I agree, are important for reasons you've mentioned)
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the Servercert-wg