Re: [perpass] draft-farrell-perpass-attack architecture issue

[Stephen Farrell <[e-mail]>]

On 01/14/2014 10:00 PM, Melinda Shore wrote:
> On 1/14/14 12:45 PM, Fred Baker (fred) wrote:
>> So the question in the shepherd's report should not be "tell me you
>> thought about the EU Data Retention Initiative and whether your
>> protocol's data identifies an individual". It should be "what
>> personal, equipment, or session identifiers, encrypted or otherwise,
>> are carried in your protocol? How might they be correlated with
>> offline data or otherwise used to infer the identity or behavior of
>> an individual?"
> 
> I agree - I think this is a useful framing, beyond the question
> of actual traffic inspection.  It's pretty clear that there's
> been a lot of data mining, as well, and we haven't thought very
> carefully about what we may be leaking inadvertently.  This is
> particularly a concern as efforts like geonet start to ramp
> up.

I do like the idea that shepherds would report on this topic
(or more generally on security and privacy) in their write-ups,
but have a genetic dislike of the way we used to have a
1000-point questionnaire for shepherds to fill in. And a lot
of the current shepherd write-ups we get tend to be out of
date wrt e.g. IPR so I'm pretty convinced that we shouldn't
hardcode shepherd write-ups into RFCs on this topic, since
that level of process is liable to change relatively often.
OTOH, as a "new" thing for WGs to consider, it might be
quite useful if shepherds are prompted to not forget about
pervasive monitoring.

So I'm in two minds here really.

I figure that this is something where we'll have to learn as
we go. Maybe we should look at a tool that randomly (but
not uniformly randomly) picks a small number of hard questions
from a long list and asks the shepherd to answer those. Sort
of a write-up bingo;-)

I'd be interested if someone wanted to start work on some
WG-chair/shepherd guidance for how to consider pervasive
monitoring. That'd likely take a while to get baked, and
would maybe end up not (just) as an RFC, but as training
material and/or an IESG statement or something, but could
easily start as an I-D. Any takers?

S