I just realized that I haven't been making something clear. The ephemeral endpoints (to use Lance Cottrell's phrase) will be identified when they're used. The purpose of concealing from all but a trusted group (remailer operators) their address (at first) is to slow that identification as much as possible. If it takes someone (NSA or whatever) repeatedly sending messages through to discover each new address, this will take longer (and make it harder for traffic analysis, especially by other parties) than if the output end's address were immediately made public knowledge. This can be helped by the remailer(s) sending to the output address only accepting such mail from another remailer. -Allen