He's talking about something called the PBL (policy blocklist). The idea is it contains IP ranges that aren't "supposed" to send mail, like consumer ADSL ranges. People who want to send mail from home directly are supposed to do so via their ISPs SMTP servers, which may be configured to relay but only from IPs the ISP controls.
I am familiar with the SpamHaus PBL. It was my understanding that anyone could easily remove their own IP address from the list if required? Presumably that is what happens when Rackspace customers want to send legitimate mail. So why would it cause any problems for spammers?
One downside with ISP outbound relaying is SPF; if you want to use it you need to figure out how mail exits the ISP and keep the SPF records up to date; and even if you do, suddenly all the other customers of that ISP can fake mail from you while passing SPF.
Another downside is you lose logs of and insight in the mailq and the recipient smtp server responses.
