[cryptography] The next gen P2P secure email solution

grarpamp grarpamp at gmail.com
Tue Dec 24 02:45:17 PST 2013


On Tue, Dec 24, 2013 at 5:09 AM, danimoth <danimoth at cryptolab.net> wrote:
> A problem which could rise is the 'incentive' for peers to continuosly
> providing bandwidth and disk space to store messages. I'm a simple dude,
> with a mailflow of ~5 email per day. Why I should work for you, with
> your ~10000 mail per day for all your mailing list?

I think this is one of many design choices to be made.

Extra bandwidth is hard to avoid, unless the topology is
point(sender)-to-point(recipient). Yet with that, there is no effort made
to hide who is physically talking to who. We want to try to defeat this
type of analysis, so we can't be simply point-to-point.
 ie: bittorrent and today's email are point-to-point, no multihop.

Next is storage (mix) vs. latency (tunnels). This seems less clear to
me when up against analysis. Filling circuits (tunnels) with chaff
seems interesting. And with deliverey directly to your recipient over
some tunnel circuit, you don't have to build in complex message redundancy
protocols (more storage float outstanding) to ensure your message 100%
gets there when 90% of the nodes go offline taking your stored message
with them. You also get direct realtime delivery confirmation too.

> Somewhere on this list (or p2p-hackers?) there was a post of mine,
> regardings an economic incentive between peers, which could be a
> solution, but as always technical problems arose, like pricing the
> services and a fair exchange between peers.

The question arises, how does one provide free anonymous transport
to those anons who simply can't pay because they are anon? How
do you 'get users' when the mentality is 'for free'? Bittorrent/Tor are
free and seem to work ok. Though it's also probably not unreasonable
to suggest (and harder to enforce) that you get 1:1 what resources you
donate to it. ie: I need to push 1GiB this month, so I need to provision at
minimum 1+Nx1GiB to do that... 1 for me, Nx1 for the net due to my use
(where N is some impact ratio inherent in the design of the net, such
as number hops.)



More information about the cypherpunks mailing list