On Wed, 1 Oct 1997, Anonymous wrote:
Bill Stewart wrote:
Latency is essential to security, though high volume reduces the latency that's needed to get a given level of security.
Latency may be a means to get security in the current remailer design, but it is a means to the end and not the end itself. This doesn't mean we shouldn't use latency to get security, but it is undesirable, like using cinderblocks for construction. The reason I point this out is that it is important to separate design choices to achieve a goal from the goal itself.
If we had a remailer network in which each customer had a constant bandwidth connection to one or more remailers, you could have zero latency mail. (Actually, this would be nice to use with those Comsec phones.)
Or use something like anonymous broadcast. Send (fixed sized, constant rate) packets continuously around a fixed set of remailers. Messages may or may not be inserted into these packets, and with encryption it is impossible to tell which node inserted a message (and only until the packet completed the circuit could you tell - inside the machine originating the broadcast). Messages would be resent through the set of remailers, and could be PK encrypted to one of the other nodes. With the broadcast being anonymous, the reciever within the set cannot know who it came from (and theoretically because it is encrypted, no one could know who it is to except the sender, and you can do something to create anonymous target packets). You would still need some mixing for low usage times (or wait for N messages and circulate a control "forward now" message, so that 10 messages come in sequentially and are held, and then at an even interval, 10 messages go out from random nodes.