[messaging] collaborative random number generation

[carlo von lynX]

On Tue, Dec 08, 2015 at 02:47:31PM +0100, Jeff Burdges wrote:
> A shared random number does however appear useful if you want to hide
> which zones are popular.  At present, certain adversaries could simply
> identify zones that were popular at certain times from certain parts of
> the world and then hack some querier to determine the zone's contents.

Oh, *that* is an interesting scenario. So the use of plain pet names
is in any case not so smart - what it takes is a shared secret that
you throw into the look-up maths: private identity pubkey + secret + petname.
The secret is shared among all legitimate subscribers of the private identity,
but I don't see why it would need to be "random". Not sure if this is already 
implemented in the GNS design or would need to be done on top. The sharing
in any case is trivial - it's just an extra parameter in the subscription
process. No need to come up with any byzantine consensus scheme AFAIU.

For completeness, the psycstore approach would not suffer from this kind
of attack. It has to deal with the challenge of anonymization of 
distribution paths instead, which currently isn't implemented - so
running a sufficient number of GNUnet nodes could allow you to guess
metadata of subscribers. Luckily there have been several functional
implementations of anonymous multicast in the science community, so
GNUnet is working on plugging one of those into the backend.


-- 
  E-mail is public! Talk to me in private using encryption:
         http://loupsycedyglgamf.onion/LynX/
          irc://loupsycedyglgamf.onion:67/lynX
         https://psyced.org:34443/LynX/