(namedroppers removed as my comment is more of a meta-comment on running mailing lists and I'm not subscribed to namedroppers itself) Olafur Gudmundsson <ogud@ogud.com> writes: > Randy is currently wasting valuable time in manually scanning 100+ spams > a day that are sent to namedroppers and other IETF mailing lists he runs > and we all should thank him for the good citizen service he provides! > Every meesage that is reposted from the bounced list contains a header > explaining that posting address is not a subscribed address. This is an obnoxious amount of work. I can thank him for the service he's providing in trying to keep the list free of spam and also think that this is way more work than someone should be expected to do. :) While I really dislike the technology when applied to personal mailboxes, this sounds like a place where a confirmation system would work well. If the list receives a message from a non-subscriber, send back a message saying so and asking them to respond and include a confirmation code of some kind, similar to mailing list subscription confirmations. If they do so, release the message into the mailing list and also whitelist their address (in case they're participating in an ongoing discussion). My experience with a technique like this is that it eliminates 99% of the spam still, is reasonably intuitive for at least a technical audience, and eliminates the need for anyone to wade through all the spam to look for the gems, a task that I would not wish on anyone. It also has the side advantage of being unassailably impartial. We all already spend far too much human effort dealing with spam. Centralizing that human effort onto one person optimizes it somewhat but still wastes valuable time that could be better used for some productive purpose. It's rather like periodically cleaning the bathroom, and if there were some way we could get computers to do that for us, I think we should jump at the chance, even if the computer doesn't do *quite* as good of a job. :) This solution does require some additional setup on the server side: The list software has to be able to do those sorts of confirmations, has to maintain a server-side queue of messages that are pending confirmation, and has to implement the whitelist system. It would likely require a small amount of work to implement over an existing mailing list manager, and I'm certainly not suggesting that Randy have to do that implementation work. But since this situation comes up very frequently for IETF mailing lists, perhaps someone could volunteer to implement this feature for whatever mailing list management software the IETF mailng list system is using? That would at least help the problem for people hosting their lists with the IETF. I wonder if Mailman already has this feature. If not, that might be a good place to start in adding it, since Mailman is very actively developed and seems to be passing the venerable Majordomo as the most widely deployed mailing list management system. -- Russ Allbery (rra@stanford.edu) <http://www.eyrie.org/~eagle/>