Sorry if I'm being dense - will someone please E-mail me and tell me why outgoing-only (or incoming-address-unavailable) remailers are useful?
I'm not sure that they are. The idea is, that some people might want to philanthropically provide a remailing service without revealing their identity. (or, heck, do it for a profit with anonymous ecash). This is currently pretty much unworkable. If a system that provided a reliable decentralized remailer infrastructure (like I _think_ the one I've proposed does), then it might become more workable, but I'm not certain if it would cross the boundry into something actually practical. But it's an interesting idea.
Also, with respect to getting the addresses of working remailers from a newsgroup - it may not be a good idea to treat any address advertising itself as a remailer as a useful remailer. Remailer 'x' may well be run by a remailer-hater who publishes its traffic openly,
This is a valid point, but it exists in _any_ remailer system or infrastructure. It's safest to assume that some cypherpunks list members who set up remailers are actually NSA agents. If you chain your message through 15 or 20 remailers, as long as 3 or 4 of them are not "evil", you are probably in good shape. But there's certainly a chance that all 20 are evil; I don't see how the "alt.anonremailer.net" concept changes the odds of getting a chain of all evil remailers. And, yes, I agree that the wise person wouldn't indiscriminately use remailers from this newsgroup, but only use those whose keys are signed by someone he trusts. It's obviously up to the user _how_ to use this hypothetical infrastructure, and there are ways that it could be used that wouldn't give you very much security. But I'm confident that if used properly it wouldn't give you any _less_ security then the current system, which is basically people telling each other about new remailers, and manualy adding them to their PGP rings and such.