[cabfpub] [cabfman] Improving the security of EV Certificates

Fri Dec 20 01:23:10 UTC 2013

Hi Kirk,

The numbers were hardly meant to be "representative". Pinning is far more
expensive than $100, and I have no doubt that for some CAs, they may find
their inability to change issuance pipeline, or to maintain a consistent
view of it, may see their costs far exceed $10,000.

The purpose of these numbers, and breaking them down into smaller ones, was
to show that the math of CT vs pinning is that CT, even if more expensive
upfront for CAs, is a far greater cost savings because it affects *every
single secure site operator*. The force multiplier of having valid
detection mechanism available for every website, versus requiring the
significant engineering and technical investment into pinning (as well as
the unforeseen costs of getting pinning wrong) mean that in any world in
which there are more than a dozen or so SSL sites, CT is a far cheaper
option.

I don't mean to resurrect the "pinning as viable" thread - surely because
it's not and they solve different issues - but the costs of pinning are
directly born by the site operators, yet the risks are themselves
introduced by the continued failures and misbehaviours of CAs. Surely
that's putting the cost on the wrong actors - a cost that is still far, far
greater than CT will be.

On Wed, Dec 18, 2013 at 8:07 PM, kirk_hall at trendmicro.com <
kirk_hall at trendmicro.com> wrote:

>  Ryan – out of curiosity, how did you come up with an estimated cost on
> CAs of CT implementation of $10,000 per year per CA?
>
>
>
> *From:* public-bounces at cabforum.org [mailto:public-bounces at cabforum.org] *On
> Behalf Of *Ryan Sleevi
> *Sent:* Wednesday, December 18, 2013 1:11 PM
>
> *To:* Eddy Nigg (StartCom Ltd.)
> *Cc:* CABFPub
> *Subject:* Re: [cabfpub] [cabfman] Improving the security of EV
> Certificates
>
>
>
> Hi Eddy,
>
>
>
> Thanks for allowing this to be posted to the public list. I've attempted
> to reply to your points below.
>
>
>
> On Wed, Dec 18, 2013 at 9:55 AM, Eddy Nigg (StartCom Ltd.) <
> eddy_nigg at startcom.org> wrote:
>
>
> On 12/18/2013 07:28 PM, From Jeremy Rowley:
>
>  Pinning is more risky for unsophisticated users who could brick their
> systems.
>
>
>
> I don't think so....such users would never use it, the same way such users
> would never investigate a log or list of certificates.
>
>
>
>  Plus, pinning becomes a market divider so I’d worry about anti-trust
> violations if the recommendation came from the CAB Forum.
>
>
>
> How come, can you explain?
>
>
>
> Much in the same way that some CAs object to CAA. Any notion of having the
> CA/Browser Forum encourage that users use technological means to say "I
> have relationships with these CAs, and I would prefer [other CAs] not
> issue" seems to scare some members. I do not believe this is at all
> warranted, and is more likely a mischaracterization of the technology, but
> that's something to be deferred for your own legal counsel.
>
>
>
> Both pinning and CAA are ways for customers to take control of the "Every
> CA is capable of issuing for every site" issue, which we see year after
> year, incident after incident to be the root cause of the continued erosion
> of trust in the CA ecosystem.
>
>
>
>
>
>
>  Transparency, in my view, is better because it requires a change only by
> the CAs and browsers, not by the users.  The information is then available
> for any researcher to digest and evaluate, not just the end user.
>
>
>
> It's mostly the competing CAs!!!, software vendors, Netcraft and friends,
> and some researchers that care about it (EFF, Qualsys and some others).
> It's the same crowd that would use pinning too.
>
>
>
> This is not true at all.
>
>
>
> Auditors are not equivalent to site operators. Site operators carry great
> risk in pinning and getting it right - it certainly provides preventative
> benefits, but with an operational complexity cost. Further, organizations
> like EFF, Qualsys and others cannot defer those costs on behalf of the site
> operator. It's a risk they bear on themselves.
>
>
>
> Pinning offers the ability for anyone, without risk to their operational
> capability, to look to examine for misissuance - past or present. This
> presents zero risk for the site operator, at a cost born by the CA, whereas
> today's system of global issuance capability with no auditing is a
> zero-cost-to-the-CA, with risk and cost born by every single site that
> would ever wish or does use SSL.
>
>
>
>
>
>
>  A headache is when another DigiNotar is compromised , issues a couple
> thousand certificates fraudulently, and covers it up for several months.
>
>
>
> Truly a problem, but may be attacked from a different angel (how about
> different approach  to auditing?). I mean, we are doing our utmost to
> comply to all the various requirements and much more than that - a price we
> are willing  to pay because for this we are here. Now this proposal has a
> significant price tag for something that hasn't been tested and used over
> time with the "only" goal to detect the next DigiNotar.
>
>
>
> Every single public CA security incident we have seen in the past 3 years
> would have been detected immediately from a system like CT. Trustwave,
> Diginotar, Turktrust, and most recently, ANSSI were all detected through
> luck and vigilance, and only because they happened to affect a large site
> whose engineers are using every means capable to them to attempt to detect
> such mis-issuance.
>
>
>
> For all we know, there may be thousands of other misissuances from
> existing CAs - whether well-meaning or not - that have gone unnoticed,
> simply because they did not decide to affect one particular site. CT makes
> it possible for anyone - from Joe Schmo on the street with his $10
> certificate, to the multi-billion dollar multi-national with engineers
> committed to dealing with just this issue - to detect misissuance.
>
>
>
> I think you're pretty grossly understating the benefit here.
>
>
>
>
> IMO pinning can achieve the same way cheaper (for me). And again, if we
> could combine revocation for example, the benefit would be much bigger and
> trade off the expenses/efforts...
>
>
>
> Assume the cost of pinning is $100/year/site.
>
> Assume the cost of CT is $10,000/year/CA.
>
>
>
> By any measure, CT has a far bigger benefit with far fewer expenses,
> efforts. And that's using a very generous understatement of pinning's costs
> (operationally), which is likely several orders of magnitude higher as the
> profile of the site increases. And even if CT itself carries with it
> several orders of magnitude more cost, the math *still* works out, because
> there are *millions* of sites that use SSL, and *millions* more that
> should, and if your idea is pinning is right for everyone, then you're
> asking inordinate amount of costs to be born for all users, many of which
> who may and are happy with "detection" over "prevention" for their
> risk/reward profile.
>
>
>
>
>
>    Regards
>
>
>
> Signer:
>
> Eddy Nigg, COO/CTO
>
>
>
> StartCom Ltd. <http://www.startcom.org>
>
> XMPP:
>
> startcom at startcom.org
>
> Blog:
>
> Join the Revolution! <http://blog.startcom.org>
>
> Twitter:
>
> Follow Me <http://twitter.com/eddy_nigg>
>
>
>
>
>
>
> _______________________________________________
> Management mailing list
> Management at cabforum.org
> https://cabforum.org/mailman/listinfo/management
>
>
>
> TREND MICRO EMAIL NOTICE
> The information contained in this email and any attachments is confidential
> and may be subject to copyright or other intellectual property protection.
> If you are not the intended recipient, you are not authorized to use or
> disclose this information, and we request that you notify us by reply mail or
> telephone and delete the original message from your mail system.
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.cabforum.org/pipermail/public/attachments/20131219/0aa5055e/attachment-0003.html>