Re: [ietf-privacy] Is there an official working definition for Privacy Online?

[Peter Schoo <[e-mail]>]

Am 05.05.16 um 17:12 schrieb Stephen Farrell:
> 
> 
> On 05/05/16 15:53, Alissa Cooper wrote:
>> +1. If people want to consider privacy as a heading under which we
>> group a bunch of different kinds of attacks, that works perfectly
>> well I think.
> 
> In the case of privacy, not all the bad things are correctly
> described as attacks IMO. E.g. leaving sensitive data in a
> log file for too long is not in itself an attack, but can be
> risky. Only emitting packets when a user is present similarly.

Descriptions of privacy (and security taken as example for how to make a
privacy definition) are discussed on different levels. Robin's two-liner
needs to be applied and interpreted. Effect is that it raises
understanding, but takes time and effort to seriously discuss it, which
is good to do.

On the other hand, it helped to establish, as Christian wrote,
categories for security attacks -- denial of service, information
disclosure, spoofing, elevation of privilege, etc. Agreed attacks
categories form common ground and makes it more useful in standards.
Something similar would be useful for privacy too, as the set of
potential threat is blurred or the understanding of threats and impacts
changes with deeper discussions, more thorough investigations.

Why are these differences? Sure, the nature of security and privacy are
somewhat different.

> I'm not even sure the risk analysis method we use for security
> is the best way to try address privacy in IETF work. 

and that's why many apply privacy impact assessments, which have a
different focus when you analyse a system/service/whatsoever, e.g.
concerning time. I argue that security threats typically do not last as
long as privacy threats, they are bound to communication sessions or
periods of subscription - more overseeable. Clever data fusion can
create threats the day after tomorrow.

Security discussions sometimes comes with properties or services that
shall be provided, e.g. authentication, authorisation, confidentiality,
integrity, availability, or CIA etc. This part of a taxonomy is not
present in the privacy discussion. Like we detail security in
(permutations of) CIA, I miss detailing privacy here. Today privacy
papers often apply Anonymity, Unlinkability, Undetectability, Plausible
deniability and Confidentiality.

All of the security definition approaches are helpful, two liners,
attacks and properties. Concerning privacy we are still learning about
threats and how to categorize them, to make these categories useful for
standards. Though it might be different with finding a suitable
taxonomy, i.e. relevant to IETF, that help detailing privacy aspects.

BR
	Peter

-- 
Peter Schoo, [e-mail]