Well I basically agree with you. I liked it better when they just used SpamAssasin to filter the mail. I did switch my mail from them to an account which does use SpamAssasin.

I don't know exactly what methods they use to determine the originating point of the email. Quite possibly they use some routing information, or the "reply-to" and not the "From" field. They must have some scheme to determine if the reply-to address exists, or if it corresponds with the routing.

They may have developed a system to address the issues which you raise. I'm sure if they are trying to avoid their spam problem, by increasing load on the backbone which they are hooked to, the managers of the backbone will let them know.

It seems that a few strategies are evolving, and being tested in real-time, to fight the spam problem.