What they're talking about is blocking outgoing traffic that attempts to connect to port 25 on any server outside the ISP's network. That would effectively shut down the zombies because all mail servers listen on port 25 for incoming mail.
My problem with this is that it would be a major PITA for anyone who operates his own sever (for mailing lists or whatever). Instead of blocking port 25, I'd rather have my ISP use the Open Relay Database and other blacklists to block traffic coming from known zombie or spambot servers. The reverse DNS lookup is also effective against spam with forged headers- in my company this technique alone shuts out well over half the junk.
this is the problem, when you infect millions of desktop PC's an RBL is no longer going to work. Given the fact most high speed nets are DHCP you could find yourself blocked because the guy before you was a bot. If anything white list would work if not for the fact maintenance would be damn near impossible..
Bingo. All of the spam that I get to my spam-bucket comes in to my ISP's POP server and I pick it up using fetchmail. Any spam sent to my mail server gets blocked, because I use the blacklists. I don't know why the ISPs won't.