Back in June, aparently, the FTC said that a do-not-email list (likethe do-not-call list) would not work, and would generate more spambecause spammers would use it as a source of new email addresses. Though it's a bit late now, I have to wonder about the latterpoint. Why not simply map each address into its MD5 checksumbefore storing it?
So email@example.com would become "a0b6e8fd2367f5999b6b4e7e1ce9e2d2"which is useless for sending email. However, spammers could use any of many available toolsto check for "hits" on their email lists, so it's still perfectlyusable for filtering out email addresses. Of course it would alsotell spammers that they have a 'real' email address on their list, butonly if they already had it -- so I don't think that would be givingthem much information at all.
I still think the list would be useless because spammers would simplyignore it. But it wouldn't generate new spam, and it would driveup the cost of spamming by making the threat of legal action a bit morepossible.