-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 > http://wiki.freenetproject.org/FreenetZeroPointEight
To choose nodes for premix routing while avoiding the Sybil attack, have you considered using a network flow algorithm like Raph Levien's trust metric for Advogato? http://www.advogato.org/trust-metric.html http://www.levien.com/free/tmetric-HOWTO.html The reported topology would be converted into a single-source single-sink graph as in the Advogato metric, with your own node as the source. The amount of flow through a node would determine the probability of selecting it for premix routing. Also, what's the advantage of using cells rather than exposing the topology within a certain number of hops? Using a hop count effectively puts each node at the centre of its own cell. The difference could be important if we assume that nearby nodes are (to some extent) more likely to be trustworthy - if I'm at the edge of a 6-hop-diameter cell, the most distant node I need to trust is 6 hops away, whereas with a hop count that covers the same number of nodes, the most distant node I need to trust is 3 hops away. Any thoughts? Michael -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.2.2 (GNU/Linux) iD8DBQFEUNuxyua14OQlJ3sRAozDAJ9tPEhzwrEZQx+3pXTtuIwnAajXTwCg2cyw /E1tIhS/A8W0yGBF9JUhEPs= =oN7G -----END PGP SIGNATURE-----
