Re: email at scale

On 5/14/2012 3:59 PM, paul vixie wrote:
> On 5/14/2012 1:24 PM, Dave Crocker wrote:
>> On 5/14/2012 12:43 AM, paul vixie wrote:
>>> the people who are willing to put that much effort into their e-mail
>>> communications are few and far between. we can build for utility at
>>> scale, or privacy and authenticity at scale, but not both.
>>
>> I think we do not have an existence proof for privacy and authenticity
>> at scale.
>
> i would have said that your messaging work at compuserve qualified as
> such, dave. in fact that was one of the visions in my head when spoke
> the words, 'walled garden'.

1.  I never worked at compuserve, but I'll guess you meant MCI.

2.  MCI Mail had no interested distributed security technologies that 
are relevant here, that I can think of.

Please note that I'm imposing a particular meaning for "scale" that has 
has less to do with number of users than with number of independent 
administration and lack of central control (except perhaps a central 
control for a registration hierarchy.)

>> A common view is that good security cannot be easy to use.  It might
>> even be true.
>
> i know that incompatibility, i just think in the other direction.
> anything that's easy for a human to use will also be easy for all of the
> malware infesting that human's devices to use.

There's an inherent and even obvious logic to that view.  And I can't 
provide anything like an adequate contrary proof.  But I believe it 
isn't true.  I'm pretty sure I mean that as an engineering, rather than 
religious, belief.

There's probably also some adjustment to the definition of 'security', 
and no, I can provide details for that either.

(and thus, neither secure
> nor securable). something that's easy for way-way-way-smarter humans
> (for example, my kids and their friends) is likely to borderline
> unusable by me (and maybe even by dave). the tension is, as one ratchets
> up the minimum skill level required then security goes up but utility
> (by definition) goes down.
>
> anyway this isn't history (sadly).

except the long history of failure in gaining adoption for strong 
security technologies over the open Internet, for a sufficient range of 
features.

d/

--

-- 
  Dave Crocker
  Brandenburg InternetWorking
  bbiw.net