[cryptography] Potential funding for crypto-related projects
michael at briarproject.org
Wed Jul 3 15:07:32 EDT 2013
-----BEGIN PGP SIGNED MESSAGE-----
I'm no expert either but I'll try to answer to the best of my
understanding. I'm CCing Henry Corrigan-Gibbs, one of the Dissent
designers, who will hopefully correct my mistakes. :-)
On 03/07/13 17:11, Wasabee wrote:
> is it really feasible to get good latency/bandwith with such
> system? it seems users need to transmit packets at the same time;
> so it seems the latency and bandwidth is a bottleneck because
> everyone must wait for the users with highest latency and lowest
> bandwidth? Or is there a scheduling mechanism involved (which would
> "eat up" "usable" bandwidth)?
Dissent has a scheduling mechanism that allows the members of each
DC-net to transmit different amounts of data in each round. I assume
each member must send and receive as many bits as all the members want
to transmit in total, plus the scheduling overhead, but that could
still be a big efficiency gain compared with each member having a
fixed-length transmission slot.
> Also, how much trust is put on servers compared to Tor? At first
> sight, it seems that the compromise of one server will compromise
> all clients connected to this server since servers knows all their
> shared secret.
Every client shares a secret with every server, and a client's
anonymity is only broken if all the servers collude. So there only
needs to be one honest server, and the client doesn't need to know
which one it is.
It would be interesting to imagine a network where the servers were
run by mutually distrusting parties, and every client was satisfied
that there was at least one trustworthy server, but different clients
trusted different servers. Then clients would be able to communicate
anonymously with each other even if they couldn't agree on any server
they could all trust.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.10 (GNU/Linux)
-----END PGP SIGNATURE-----
More information about the cryptography