I disagree. Suppose, for example, that we designed a new mail protocol
(call it X.399 if you like) and convinced everyone to implement it on their
systems. If it worked at least as well as SMTP and additionally kept spam
out, I'll bet it would be a short time before the only folks using SMTP
were the spammers.
seems farfetched. it's hard to imagine how a mail protocol could thwart
spam except to require authentication for all mail. if you can manage to
deploy an authentication system at all (and the infrastructure that is
required to make it work), it is probably easier to retro-fit SMTP to do
that than to deploy a completely new protocol.
the real difficulty is getting the authentication infrastructure in place.
and BTW, this problem is not limited to email.
Keith