It's irrelevant because it doesn't matter which of the two scenarios
above prevails; all of these schemes are pre-failed because they
all presume that something is true which is emphatically false: network
endpoints are secure. All of these proposals thus amount to attempts
to layer a trust mechanism on top of an already-known badly-broken base.
I think you're on the wrong mailing list. You should be on the "how do we un-pwn a few million Windows PCs" list, if you think that that's an essential prerequisite to doing anything technical about spam.
mathew