[cabf_validation] 2023-05-18 Minutes of the Validation Subcommittee [DRAFT]
cclements at google.com
Fri May 19 19:06:50 UTC 2023
Meeting Date: May 18, 2023
Attendees: Aaron Poulsen (Amazon), Aneta Wojtczak-Iwanicka (Microsoft), Ben
Wilson (Mozilla), Bruce Morton (Entrust), Cade Cairns (Google), Chris
Clements (Google), Clint Wilson (Apple), Corey Bonnell (DigiCert), Daryn
Wright (GoDaddy), Dimitris Zacharopoulos (HARICA), Dustin Hollenback
(Microsoft), Eva Vansteenberge (GlobalSign), Inigo Barreira (Sectigo),
Johnny Reading (GoDaddy), Joseph Ramm (OATI), Luis Cervantes (GoDaddy),
Michael Slaughter (Amazon), Michelle Coon (OATI), Nate Smith (GoDaddy),
Rollin Yu (TrustAsia Technologies, Inc.), Ryan Dickson (Google), Tobias
Josefowitz (Opera Software AS), Trevoli Ponds-White (Amazon), Wayne Thayer
Corey Bonnell greeted participants, started the recording, read the
attendance (above), and read the “note well.”
Approving previous meeting minutes:
April 20th meeting minutes were approved
May 4th meeting minutes were approved
Corey summarized the meeting’s planned agenda:
Check-in on multi-perspective domain validation work
Delegation of domain validation to other parties
Should we meet on June 1st (week before F2F)?
Check-in on multi-perspective domain validation work:
The Chrome Root Program is trying to wrap up suggested comments in a
doc hopefully by tomorrow.
Ryan Dickson will be transitioning the proposed draft language into
GitHub, such that it can be circulated more broadly with this group.
This is for collecting more feedback and more perspective to help
steer us down the path of a ballot.
Delegation of domain validation to other parties has been a hot topic
and first and foremost in Corey’s mind.
Corey asked if others had topics they would like to propose. None
We have 1.5 hours scheduled for Thursday morning.
Corey asked if an hour and a half was too much for this one topic
or if we see this as a discussion that would take
significantly longer than
an hour. His initial thinking was that an hour and a half was okay.
Ben Wilson asked if we have looked back at past face-to-face
discussion items that were not fully discussed and may need additional
conversation or if this one topic is indeed all we have to discuss.
Corey mentioned not having the list that was circulated previously
in front of him, but he thought the next steps were clear and did not
recall if we actually had any outstanding discussion.
Corey does recall there being some discussion around some
improvements to the subscriber language.
Ben stated that was correct and that he and Dustin Hollenback were
talking about it before this call. They were working on it
until about a
month ago and then got busy with other priorities. They need
to check and
see if there is something they can present for discussion.
Corey will allocate some time for Ben and Dustin to present those
Wayne Thayer asked to revisit the first topic of delegation of domain
validation. He thinks the scope of that conversation could determine how
much time we talk about it. If we are just talking about what are the
conditions under which it’s okay to delegate domain validation, that’s
pretty easily scoped or pretty narrowly scoped. He likes to think of the
topic from the broader picture of how we make the whole process more
automated, which could turn into a much bigger discussion and opens more
opportunities for brainstorming.
Trevoli Ponds-White suggested if we want to scope the conversation
down to the clear desire by people to automate certificates
then we should
frame it as such. It would be good to steer the discussion.
For example if
we want to discuss when is it okay to delegate, we’re actually saying
domain validation has to be delegated and these are the
we’ve agreed to as a group and people can comment from there.
Corey said it sounds like there is a desire to widen the topic a
bit. In conversation weeks ago we were going to look at and
steps on how we wanted to improve the Baseline Requirements
and whether or
not CAs need some guard rails around delegation and domain
we could just have very surgical change to the BRs, which
would mainly just
be changing the definition of random value.
Trev clarified a desire to present our conclusions and allow
people to provide feedback. Not start another discussion that
had from scratch.
Wayne stated on the broader issue of how to automate domain
validation, they have not really reached any conclusions to present.
Clint Wilson believes we’ve agreed on a very high level, but we
haven’t done any of the granular “how” does domain delegation
how does it happen securely. We’ve agreed with the
overarching “this is a
topic that we want to talk about” but all of the work is
still yet to be
Trev stressed the importance of having a goal for the discussion
and that we reach that goal. We should try to present
to the guard rail concept mentioned by Corey. We should present a
conclusion and then discuss it because that is easier than
some thoughts, what do people think?”
For example one of the conclusions could be you have to
determine the level of rules that you want to have around
this, you can’t
constrain subscribers from going to other service
providers and constrain
what they do. You can only constrain the CAs, so it
doesn't make sense to
constrain the CAs so much that what they are offering is
than what a subscriber could get from someone that's not a CA.
Wayne proposed the prompt that it is okay for domain validation to
occur without the subscribers involvement.
Corey agreed this is good framing. It also provides us the
opportunity to give more context and background on some of
that we’ve had in this group over the past couple of
months that the wider
group may not be familiar with. We can lay out the
rationale and then open
up the floor to any conversation.
Trev wanted to add to what Wayne said, the subscriber does not
directly need to be involved, but rather they can opt-in.
It’s not that the
subscriber is not involved and validation happens without
but rather they can opt-into it.
Corey agreed that it is useful clarification and this probably
needs to be discussed but in his mind there should be an
action undertaken by the applicant to allow that delegation.
Ben suggested they could agree to it in the Subscriber
Corey will create some slides to help stimulate conversation at
Trev asked what we want to close out in the face-to-face conversation.
Corey thinks coming out with concrete language changes to the
Baseline Requirements might be a bit of a stretch. At least having a
direction on what guardrails the CA needs to have, if any, in terms of
Trev suggested a general agreement that opt-in is okay and fewer
guard rails makes the most rational sense because anything else is
unenforceable. These two seem like blockers before we can change the
Corey said there was general consensus on previous calls.
Wayne suggested the tension is when you say, okay, the logical
conclusion of this is that you don’t have to perform domain validation
every X days. You are essentially saying you can do it once and with
certain provisions you’re done. When you say it that way, a
lot of people
are going to say “no, we can’t do that”. To foster
discussion, what does it
mean to say that you can assent to automated renewals?
Corey agreed that phrasing is important. He does not see the
delegation as an allowance for the CA to skip the domain
It was just that they would be performing all the actions,
the CA would potentially be publishing the random value. They
be performing the same mechanical steps that are required for domain
validation (DNS lookups etc.) but it sounds like there might
interpretations of this.
Wayne clarified that once you say an applicant or subscriber can
agree to the use of their domain on a permanent basis until
This has implications, such as why is the CA creating a
random value, why
are they not checking CNAME? At some level we’re saying a lot
of what CAs
do during a renewal scenario today doesn't make a lot of
sense. It’s easy
to say, “oh yea, that’s fine” but once you start thinking through the
implications, some people may become uncomfortable.
Michael Slaughter suggested thinking about the topic as two
Does this concept of delegated domain validation fit within the current
construct of the Baseline Requirements as is today with maybe a few tweaks?
Is this allowable?
Is there a way to make the Baseline Requirements better align with this
concept of domain delegation with changes to some of the fundamental ways
we look at domain validation in this new way? We don’t have to answer both
questions to make progress in the exact same way.
Michael clarified he thinks there is room for targeted surgical changes
to the BRs to make this concept of delegated domain validation more clearly
allowable and perhaps add additional guardrails where it makes sense. The
thoughts provided by Wayne lead us into some of the deeper fundamental
questions and ways we look at domain validation, which is fascinating, but
these can be explored on two different threads.
Wayne agreed. The conversation about tactically making changes to
clarify what's allowed with CNAME validation is relatively easy and we
should have that conversation and fix that. There is a more interesting
conversation to be had, but this can be treated as two separate threads.
Corey asked if the group wants to first tackle the surgical approach
versus the higher-level discussion with the potentially larger changes. Do
we want to have the wider discussion or the more surgical discussion?
Wayne believes the surgical approach is pretty well understood and we
just need to write a ballot.
Michale suggested we state the conclusion upfront to the group and say
this is what the validation subcommittee discussed and this is what we
concluded and let the group provide feedback or raise points that were not
previously discussed, and then move on to the larger discussion that is
potentially more interesting.
Corey suggested we still need to discuss CA guardrails and we can do
that in the context of the face-to-face.
The topic around Subscriber Agreements from Ben and Dustin will go first
and then we’ll discuss delegated domain validation.
We can plan for 15-20 minutes for this topic.
Should we meet on June 1st (week before F2F)?
Historically we skip the week before so that everyone has time to
prepare for potential travel.
Corey asked if there were any objections with canceling the next
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the Validation