I don't like blacklists, but....

they are very resource friendly and stop the spam as its source. But there are 2 things causing me to avoid any blacklist

First, it cannot distinguish if a single IP address is sending *both* spam and ham, eg. a shared web server having a single spammer (or just a broken php mailer script) and lots of legit users. All those emails will be lost. Exceptions (white lists)? After many lost emails and complaints. (=no way for a business customer).

Second, no blacklist is accurate enough. If anybody has ever tried a Bayesian antispam solution, he knows what difference I'm talking about.