An alternative to using a specific email (which can be banned) is to send to a mailing list or other un-moderated public forum, mailing lists are better since they are push, rather than pull. As long as you public-key encrypt, you can spam it via as many outlets as you want.
An alternative to hard-coded addresses would be to use a formula, like Conficker did, where Conficker went wrong IMO was to not register the domains ahead of time and to not often collide with real domains, so that all domains fitting the pattern could be reasonably banned at the registrar and dns level.
One approach could be to scatter update info around the web and have a collaborative spider, where every infection gets added to a p2p network and where parts to spider are handed off, and sites with update info will have cryptographically signed messages, so that their authenticity can be verified and misbehaving/malicious (or would that be well-meaning?) nodes would be ignored. Of course, proxies could scan for messages, etc, or become part of the network an wait untill they get a URL, then ban that, but it's a bit more flexible...
----------------------------------------------------------
Don't forget our IRC: irc://irc.irchighway.net/#slackers
[
kuza55.blogspot.com]