[cabfpub] [EXTERNAL]Re: Ballot 218: Remove validation methods #1 and #5

Thu Jan 4 17:44:21 UTC 2018

On Thu, Jan 4, 2018 at 9:48 AM, Bruce Morton <
Bruce.Morton at entrustdatacard.com> wrote:

> Hi Ryan,
>
>
>
> Here are some details on how we perform this method.
>
>
>
> For an OV certificate, we perform method #1 as follows:
>
>
>
> 1.      Order is received with the subject name, SANs, a certificate
> requester and an authorization contact. The authorization contact must be
> employed by the organization in the subject name.
>
How can you vet that? You've not vetted anything yet, so how can you vet
the Applicant's employment status? I think you're trying to hint at this in
4, but that's unclear.

> 2.      BR 3.2.2.1 is performed to confirm the identity of the
> organization. This task is done with using a QIIS.
>
Note that the concerns about QIIS apply here - as we've seen with
situations like "Stripe, Inc" in Kentucky. At this point, you've verified
that some organization exists, but that's based entirely on
Applicant-provided data (namely, the Applicant provided the
organization-to-vet in 1, which is under their control - e.g. "Stripe, Inc"
in Kentucky)

> 3.      BR 3.2.2.4.1 is performed using registrar information to confirm
> the organization has registered the domain name. The QIIS is used to
> identify relationships if the domain is registered to a parent or
> subsidiary.
>
As Jeremy highlighted, this is both insufficiently described and inherently
dangerous. That you have is that _an_ organization with a name such as
"Stripe, Inc" requested a certificate. Are you obtaining the full
jurisdiction of incorporation information from the registrar (I would be
curious which registrars they are, given the various collection policies
that contractually apply)? If not, then you're back to relying on the
'fuzzy match' - for example, does the registrar say "Stripe, Inc" at this
address applied for it? As shown with both the QGIS and QIIS situations,
this is not a particularly robust method - and that's assuming y'all were
trying to "do it right"

> 4.      BR 3.2.5 is performed by finding contact information for the
> organization using a QIIS. The authorization contact is called at the
> organization using this information. The authorization contact is asked to
> confirm that a certificate with the organization name, using the domain
> name(s) requested, can be issued to the certificate requester.
>
This doesn't link the Applicant's employment status - just the
authorization. Further, given that Entrust has argued strongly in favor of
reusing past validations, I'm guessing that this authorization is then
applied to subsequent requests. Further, given the BRs allow you to reuse
data and documents, do you use such authorizations for additional domain
names?

>
>
>
>
> Thanks, Bruce.
>
>
>
> *From:* Ryan Sleevi [mailto:sleevi at google.com]
> *Sent:* January 4, 2018 12:09 AM
> *To:* Bruce Morton <Bruce.Morton at entrustdatacard.com>
> *Cc:* Rich Smith <richard.smith at comodo.com>; CA/Browser Forum Public
> Discussion List <public at cabforum.org>; Kirk Hall <
> Kirk.Hall at entrustdatacard.com>
> *Subject:* Re: [EXTERNAL]Re: [cabfpub] Ballot 218: Remove validation
> methods #1 and #5
>
>
>
>
>
>
>
> On Wed, Jan 3, 2018 at 5:27 PM, Bruce Morton <
> Bruce.Morton at entrustdatacard.com> wrote:
>
> I disagree.
>
>
>
> Removing, changing and adding back in method #1 is not a productive
> exercise. This method has been used for probably 20 years and yet we never
> see any notifications, articles, alerts, etc. of how this method was
> defeated by an attacker.
>
>
>
> I think it's exceptionally dangerous to rest on that, particularly since
> CAs such as Entrust don't make available to the public their processes and
> controls to inspect whether or not they're vulnerable. I am greatly
> appreciative to Jeremy sharing the case of customers from a CA they
> acquired not being validated to the level that DigiCert holds itself - but
> that's hardly to be expected, unless we are to suggest DigiCert should buy
> out every other CA.
>
>
>
> It was this philosophical opposition that resulted in MD5 being exploited
> 'in the wild' - CAs ignoring the literature and research and demanding
> 'proof it applied'
>
>
>
> Does Entrust (or the CAs it has acquired) use this method? Can you share
> the details that are used to do this?
>
>
>
> I stand by the assertion that while it may be possible to restrict what is
> done under 3.2.2.4.1 to be 'secure', that is not what it is in the
> language, and what is presently executed is demonstrably insecure. If we
> are to suggest that we, as an industry, care about security of our users,
> then we should make tradeoffs that favor security.
>
>
>
>
>
> Note, I agree that method #1 can be approved in the BRs, but please advise
> which CAs have not already improved this method in practice? If a CA finds
> a BR requirement to be weak, they should either not use it or improve the
> process in their own practices. I assume that many BR requirements were not
> intended to have loopholes, but were written to allow competitiveness in
> the way they are adopted.
>
>
>
> Fundamentally, I disagree with this framing. System security works by
> ensuring the minimum level of security is appropriate - not that every CA
> will be smart enough, well-versed in the nuance enough, and/or financially
> motivated enough to 'not be creative'.
>
>
>
> I think that the current ballot 218 is bypassing the working group process
> where a working group was created by ballot to improve the validation
> methods. Is this the intension?
>
>
>
> This is not, nor has it never been, required by our Bylaws. There have
> been suggestions from some CAs to try to do this, but this has historically
> turned out to be a stalling tactic.
>
>
>
> Does Entrust employ 3.2.2.4.1? If so, given that it's required to track
> the validation methods it uses, can you share approximately how many or
> what percentage of certs you use it for, and how you use it? This can go
> leaps and bounds to providing meaningful data about the potential impact to
> the ecosystem from disallowing it, without the deliberative delays.
>
>
>
>
>
> If we are going to support abrupt ballots, then I would suggest that they
> at least be split into one topic and discuss method #1 and #5 independently.
>
>
>
> Finally, what is the rush? Why can’t this change be discussed at the
> bi-weekly CAB Forum meeting or Validation Working Group meeting at least
> once before a pre-ballot is produced? And why effective March 1, 2018? Not
> only has method #1 been highly effective for 20 years, but we have also
> just updated the validation methods to support ballot 190. More time would
> allow CAs to add changes to their release cycles and allow Subscribers to
> learn new validation processes they will now have to adopt.
>
>
>
> Security comes first. That sounds like a considerable stalling tactic, to
> be honest. Is this because Entrust specifically uses 3.2.2.4.1, or are you
> opposing it on procedural grounds? This, too, will help understand both the
> substance of the objections and potential paths to addressing them.
>
>
>
>
>
> I am open to change BR requirements, but do not support ballot 218.
>
>
>
> Bruce.
>
>
>
> *From:* Public [mailto:public-bounces at cabforum.org] *On Behalf Of *Rich
> Smith via Public
> *Sent:* January 3, 2018 4:44 PM
> *To:* 'Ryan Sleevi' <sleevi at google.com>; 'CA/Browser Forum Public
> Discussion List' <public at cabforum.org>; Kirk Hall <
> Kirk.Hall at entrustdatacard.com>
> *Subject:* [EXTERNAL]Re: [cabfpub] Ballot 218: Remove validation methods
> #1 and #5
>
>
>
> I agree with Ryan on this and stand by my endorsement of this ballot to
> move forward.  I’m not opposed to adding 3.2.2.4.1 back in if it can be
> made much more secure and brought up to equivalent level with the other
> methods, but I also have my doubts as to whether or not that is possible in
> the broad sense across all TLDs and registrars.  That being the case I
> think the best course is to remove it for now because in it’s present form
> is extremely weak and add back later if and when it has undergone
> sufficient revisions to make it secure.
>
>
>
> Regards,
>
> Rich
>
>
>
> *From:* Public [mailto:public-bounces at cabforum.org
> <public-bounces at cabforum.org>] *On Behalf Of *Ryan Sleevi via Public
> *Sent:* Wednesday, January 3, 2018 2:24 PM
> *To:* Kirk Hall <Kirk.Hall at entrustdatacard.com>; CA/Browser Forum Public
> Discussion List <public at cabforum.org>
> *Subject:* Re: [cabfpub] Ballot 218: Remove validation methods #1 and #5
>
>
>
> Hi Kirk,
>
>
>
> We had two endorsers for the discussion. As I mentioned, there's nothing
> inherent in needing to direct this to VWG. As DigiCert has pointed out,
> there are CAs today that are doing validations that are patently insecure.
>
>
>
> While we can understand and appreciate that some members may wish to
> introduce new validation methods that are limited in scope and
> applicability (for example, Mads' example only applies to a limited subset
> of ccTLDs, and cannot be done safely generically), in order to reduce that
> risk, I think it's entirely appropriate to take the necessary steps to
> ensure the safety of the Internet at large.
>
>
>
> This does not prevent or inhibit the issuance of certificates that have
> appropriate controls - that is, these methods could be argued as an
> 'optimization' - and thus we should not unduly delay progress. Regarding
> passing it to the VWG, could you indicate where you saw that was suggested?
> The only mention of it I saw was from you, on a separate thread, and I'm
> curious if perhaps I've missed additional discussion. Certainly, our
> workmode does not require sending such discussions "to committee"
>
>
>
> On Wed, Jan 3, 2018 at 3:10 PM, Kirk Hall via Public <public at cabforum.org>
> wrote:
>
> Tim, I thought this issue was going to be discussed first by the VWG, as
> several CAs have indicated they would like to keep (but improve) Method 1.
>
>
>
> *From:* Public [mailto:public-bounces at cabforum.org] *On Behalf Of *Tim
> Hollebeek via Public
> *Sent:* Wednesday, January 3, 2018 11:22 AM
> *To:* CA/Browser Forum Public Discussion List <public at cabforum.org>
> *Subject:* [EXTERNAL][cabfpub] Ballot 218: Remove validation methods #1
> and #5
>
>
>
>
>
> Ballot 218: Remove validation methods #1 and #5
>
>
>
> Purpose of Ballot: Section 3.2.2.4 says that it “defines the permitted
> processes and procedures for validating the Applicant’s ownership or
> control of the domain.”  Most of the validation methods actually do
> validate ownership and control, but two do not, and can be completed solely
> based on an applicant’s own assertions.
>
>
>
> Since these two validation methods do not meet the objectives of section
> 3.2.2.4, and are actively being used to avoid validating domain control or
> ownership, they should be removed, and the other methods that do validate
> domain control or ownership should be used.
>
>
>
> The following motion has been proposed by Tim Hollebeek of DigiCert and
> endorsed by Ryan Sleevi of Google and Rich Smith of Comodo.
>
>
>
> -- MOTION BEGINS –
>
>
>
> This ballot modifies the “Baseline Requirements for the Issuance and
> Management of Publicly-Trusted Certificates” as follows, based upon Version
> 1.5.4:
>
>
>
> In Section 3.2.2.4.1, add text at the end: “For certificates issued on or
> after March 1, 2018, this method SHALL NOT be used for validation, and
> completed validations using this method SHALL NOT be used for the issuance
> of certificates.”
>
>
>
> In Section 3.2.2.4.5, add text at the end: “For certificates issued on or
> after March 1, 2018, this method SHALL NOT be used for validation, and
> completed validations using this method SHALL NOT be used for the issuance
> of certificates.”
>
>
>
> In Section 4.2.1, after the paragraph that begins “After the change to any
> validation method”, add the following paragraph: “Validations completed
> using methods specified in Section 3.2.2.4.1 or Section 3.2.2.4.5 SHALL NOT
> be re-used on or after March 1, 2018.”
>
>
>
> -- MOTION ENDS –
>
>
>
> For the purposes of section 4.2.1, the new text added to 4.2.1 from this
> ballot is “specifically provided in a [this] ballot.”
>
>
>
> The procedure for approval of this ballot is as follows:
>
>
>
> Discussion (7+ days)
>
>   Start Time: 2017-01-03  19:30:00 UTC
>
>   End Time: Not Before 2017-01-10 19:30:00 UTC
>
>
>
> Vote for approval (7 days)
>
>   Start Time: TBD
>
>   End Time: TBD
>
>
>
>
> _______________________________________________
> Public mailing list
> Public at cabforum.org
> https://cabforum.org/mailman/listinfo/public
>
>
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.cabforum.org/pipermail/public/attachments/20180104/445ba7e8/attachment-0003.html>