Well, the cost and consequences are completely predicated on actually moving to CMTP. And nothing I said has anything to do with developing new protocols. Spam filtering, at least in the sense that I was thinking about it (is there any other?) is an application level solution. No matter where you put it on the network, its application level - just like a mail server is an application level solution. In fact, SMTP proxies make the network location of these tools largely irrelevant.
There are anti-spam measures that work at a protocol level. Certain ISPs (the smart ones) force their users to use specific SMTP servers. This means that customers of these ISPs are forced to use that SMTP server and only that SMTP server - which means that relaying is impossible from these networks. As long as these ISPs also place limits on how much mail their customers can send and how often they send it, no more spam gets sent from these networks.
Which leads me to my own spam solution. It has nothing to do with CMTP (whatever that is), has nothing to do with charging fees for sending or receiving and has nothing to do with enhanced filtering anywhere on the network.
Providers must be made legally responsible for the actions of users on their networks - or they must take this responsibility. Imagine what would happen if every single ISP and network provider in the United States said tomorrow that SPAM was no longer acceptable on their networks and took affirmative actions such as those that I describe to ensure that none gets sent from their networks? I predict it would go away in less than one year - globally.
Now what are the chances of that happening?