[Freedombox-discuss] Establishing Communication between Freedomboxes

[bertagaz at ptitcanardnoir.org]

On Sun, Jul 10, 2011 at 10:05:54PM -0400, Daniel Kahn Gillmor wrote:
> On 07/09/2011 10:13 AM, bertagaz at ptitcanardnoir.org wrote:
> > Well, in the GnuPG scenario I quickly explained, handling of "disruptive
> > behaviors", being abuses of a community policy or spam would be handled
> > collectively by the community I guess.
> 
> Yes; it's the details of the mechanism for doing this that need to be
> hammered out.
> 
> > If such a case happens, people
> > would just have to revoke their signature of the offending peer, thus
> > getting him outside of their WOT.
> 
> I'm not sure this is the right approach.  It seems to mix identity
> certification ("this is the key that belongs to person A") with policy
> certification ("this keyholder abides by the community principles laid
> out in document Z").
> 
> I've already written a bit about why i think mixing these two senses of
> certification are problematic.  For one thing, it makes the act of
> certification much more complicated, which tends to reduce the
> likelihood of identity certification, making it harder to communicate
> securely without having met in person (because the WoT of certifications
> is much more sparse, so key-identification is more difficult).  For
> another, it leaks more social information into the public than people
> may want leaked (that is, "I think that person A properly abides by
> manifesto Z" is a lot closer to "person A and i share some common
> values" or "person A is my friend" than "This key belongs to person A").

Yeah, I'm aware it's mixing two different things in the GnuPG signature
mechanism. I only gave a quick thought to the raised question, and this
seems to be the easiest answer (well at least at the first glance, sure it
brings in fact fuzziness).

I guess we're reaching the limit of GnuPG in this area, and that's the
problem when trying to bind social relationship on top of this tool. Maybe
it's just a nice dream, or at least it's not really possible without
adding some companion app filling the gap. Probably a smarter and more
user intuitive interface is a fist shot in this direction, and could help
adding features.

> I'd suggest that distributed approaches to community membership (and its
> flip side, distributed approaches to community ostracism) are topics
> worthy of research and experimentation, particularly with
> anti-surveillance goals in mind.

That's kind of a challenge, but an important one. :)

bert.