Re: [saag] Channel binding is great but not a silver bullet

[Nicolas Williams <[e-mail]>]

On Mon, Mar 02, 2009 at 01:09:06PM -0500, Jeffrey Hutzelman wrote:
> --On Monday, March 02, 2009 06:57:32 PM +0100 Alan DeKok 
> <[e-mail]> wrote:
> 
> >  Take Starbucks, for example.  They have WiFi.  What do the employees
> >know about it?  Pretty much nothing.  What do the people who installed
> >it know about it?  Pretty much nothing.  What does the network operator
> >*running* the network know about it?  Pretty much nothing.
> 
> Actually, I believe in the specific case of Starbucks, the network operator 
> running the network is T-Mobile, and so probably actually _does_ know 
> something about it.  At least, at the macro level.
> 
> But the point stands -- coffee houses are not network operators and they 
> are not federated identity service providers.  They are purveyors of 
> concentrated liquid evil, and the occasional cup of tea.

Coffee houses most definitely sell Internet access (or just give it
away).  And the point wasn't specific to coffe houses.  Any access point
could provide an enrolment bootstrapping service via federation if the
market wanted that sort of service...

... but the market won't demand that because people will generally be
happy to use existing relationships, such as phone service, for
enrolment.  And those that don't will settle for using the existing
weak-PKI and will test for MITMs by trying to use their accounts from
many different places in the network.

I used coffee houses as an example of how anonymity could still be had
while solving the enrolment process without a PKI.  I didn't mean to
propose that as the primary form of enrolment.  Quite the contrary; in
practice using SMS should do quite well.

Nico
--