[cryptography] The consequences of DigiNotar's failure

Jeffrey Walton noloader at gmail.com
Sun Sep 18 05:30:04 EDT 2011


On Sun, Sep 18, 2011 at 5:18 AM, Ian G <iang at iang.org> wrote:
> On 18/09/11 8:38 AM, Jeffrey Walton wrote:
>>
>> On Fri, Sep 16, 2011 at 1:07 PM, M.R.<makrober at gmail.com>  wrote:
>>>
>>> On 16/09/11 09:16, Jeffrey Walton wrote:
>>>>
>>>> The problem is that people will probably die
>>>> due Digitar's failure.
>>>
>>> I am not the one to defend DigiNotar, but I would not make such
>>> dramatic assumption.
>>
>> I don't think DigiNotar has any defenders remaining :) As for the
>> dramatic assumptions, I believe past performance is indicative of
>> future expectations: http://en.wikipedia.org/wiki/SAVAK and
>> http://en.wikipedia.org/wiki/SAVAMA. (Sorry about the lame wiki
>> reference, I probably should have found a UN human rights report).
>
>
> I don't think there is any doubt that people can die because of breached
> communications.  No need to look at the Iranians, just look at the US CIA,
> and the intel feeding into drones.
>
> The question of causality is one that is very difficult to determine, absent
> some pattern revealed by WikiLeaks (who have been accused as well).
>
> However causality is also very important.  Without some historical pattern
> of facts, we're all speculating to a greater or lesser degree. How confident
> are we of that?
>
>>> No one actively working against a government that is known to engage
>>> in extra-legal killings will trust SSL secured e-mail to protect him
>>> or her from the government surveillance.
>
> This is a sadly inaccurate statement.  Most people working actively and
> aggressively against unconstrained governments know diddly squat about tech.
>  The communities have frequent roll-over, frequent recruitment. The techies
> working with them are under considerable pressure to deliver, and often make
> basic mistakes.
>
>
>> Perhaps I don't appreciate all the pressure and options, but I believe
>> an [external] email service using HTTPS is one of the safer options
>> available when observing due dilligence.
>
> Yes, definately.  Open question:  did the 9/11 guys use HTTPS?  Or just
> HTTP?  I'm still searching for a case where it makes a clear difference.
>
> (Their main counter-intel coup was to understand that the threat model
> better than their enemy.  Their technique was to open an ordinary Yahoo
> style account, share the account, then open up a draft email, and share
> that!  Never send it, just edit and delete, over and over.  The NSA which
> were presumably hoovering all sent emails ... never saw a thing.)
>
>> Its kind of like the poor
>> man's cloud (and corporate america is flocking to the cloud, in part
>> due to the additional layer of liability offload).
>
> ! OK, I'll bite.  How does one offload liability by using the cloud?


On Sun, Sep 18, 2011 at 5:18 AM, Ian G <iang at iang.org> wrote:
> On 18/09/11 8:38 AM, Jeffrey Walton wrote:
>>
>> On Fri, Sep 16, 2011 at 1:07 PM, M.R.<makrober at gmail.com>  wrote:
>>>
>>> On 16/09/11 09:16, Jeffrey Walton wrote:
>>>>
>>>> The problem is that people will probably die
>>>> due Digitar's failure.
>>>
>>> I am not the one to defend DigiNotar, but I would not make such
>>> dramatic assumption.
>>
>> I don't think DigiNotar has any defenders remaining :) As for the
>> dramatic assumptions, I believe past performance is indicative of
>> future expectations: http://en.wikipedia.org/wiki/SAVAK and
>> http://en.wikipedia.org/wiki/SAVAMA. (Sorry about the lame wiki
>> reference, I probably should have found a UN human rights report).
>
>
> I don't think there is any doubt that people can die because of breached
> communications.  No need to look at the Iranians, just look at the US CIA,
> and the intel feeding into drones.
>
> The question of causality is one that is very difficult to determine, absent
> some pattern revealed by WikiLeaks (who have been accused as well).
>
> However causality is also very important.  Without some historical pattern
> of facts, we're all speculating to a greater or lesser degree. How confident
> are we of that?
>
>>> No one actively working against a government that is known to engage
>>> in extra-legal killings will trust SSL secured e-mail to protect him
>>> or her from the government surveillance.
>
> This is a sadly inaccurate statement.  Most people working actively and
> aggressively against unconstrained governments know diddly squat about tech.
>  The communities have frequent roll-over, frequent recruitment. The techies
> working with them are under considerable pressure to deliver, and often make
> basic mistakes.
>
>
>> Perhaps I don't appreciate all the pressure and options, but I believe
>> an [external] email service using HTTPS is one of the safer options
>> available when observing due dilligence.
>
> Yes, definately.  Open question:  did the 9/11 guys use HTTPS?  Or just
> HTTP?  I'm still searching for a case where it makes a clear difference.
>
> (Their main counter-intel coup was to understand that the threat model
> better than their enemy.  Their technique was to open an ordinary Yahoo
> style account, share the account, then open up a draft email, and share
> that!  Never send it, just edit and delete, over and over.  The NSA which
> were presumably hoovering all sent emails ... never saw a thing.)
>
>> Its kind of like the poor
>> man's cloud (and corporate america is flocking to the cloud, in part
>> due to the additional layer of liability offload).
>
> ! OK, I'll bite.  How does one offload liability by using the cloud?
The provider is another entity in the legal entanglements, which
offers yet another level of indirection.

Pre-cloud: Company A houses your data. Company A is breached, and
company A is exposed to legal liability. Post-cloud: Company A uses
Company B's cloud service. Your data is breached, and its not clear if
the loss occurred at company A or company B. Since you can't prove who
is responsible for the loss, neither company is subject to a tortable
action.

By the time dust settles on data breaches, any attempts to certify a
class action are thrown out because members of the class cannot show
loss (and future loss is not considered). Its only going to get worse
when cloud providers are added to the mix.

Jeff



More information about the cryptography mailing list