There is a simple solution to most of this.
Most botnets are used to send spam, the zombied machine connects to whichever server is the MX for a domain, and pretends to be another email server relaying a message for one of it's users.
So by default ISP's should restrict connection to SMTP servers so end user machines cannot connect to any SMTP servers apart from the ones owned by the ISP.
Your run of the mill AOL,Tiscali,BT customer uses the email address that came with the ISP, so they'd be fine. The rest probably use web based systems like gmail/hotmail etc.
The more techy savvy of us, who lets face it aren't really the big risk when it comes to ending up on a botnet, would of course have some kind of web interface on the ISP so we can permit other SMTP servers, or open it up for all. Please note web interface, not a call centre in India! (Done that once this week already thanks!)
It's not as if it's a hard thing to spot from an ISP level. They spend so much cash and technology mangling P2P, it wouldn't take 10 minutes to spot zombie behaviour, nobody normal initiates over a thousand SMTP connection in a day for starters. That would be enough to pass on their details to the sales team and send them an internet security package, or at least some advice on protection!