I am not a mail specialist, but would a simple solution not be to add an
MD5 based checker in the mail ID. Before proceeding a mail the receiving
party would call the "mailchecker" associated to the sending domain name to
check the ID consistency (if there is a checker in the ID), what would also
provide a real time receipt, ie a value added service pushing ISPs to
implement the feature. I suppose it would also easily permit to determine
if a quoted printed mail is a fake.
This could be implemented transparently and progressively. This would add a
relation but would decrease the processing and the storing of the spam: I
suppose everyone would be winning. Servers without a "namechecker" update
six months from now would be deemed suspicious.
Sorry if this is a dumb proposition?
jfc