If my ISP were to ditch everything that even smelled faintly of spam then I'm pretty sure that a fair bit of mail that I do want to read would go in the bin too and emailing me would be highly unreliable.
Complete server-side filtering cannot work unless the ISP has individual whitelists and blacklists for every client and I don't know of one that does.
Based on personal experience, I can absolutely irrefutably say that lot you've just spouted is a load of codswallop.
On the system I and numerous others in the world are using, if correctly configured, you get zero false positives and very few misses.
I've been running my present system for over 3 years now, and have indeed had zero false positives, and at most I get two or three spams a week make it through. I run email services for others on the same platform too... so we're talking a substantial number of messages being parsed.
You do not need individual blacklists or whitelists.... indeed blacklist/whitelist/graylist type systems only make things worse, not better because they require manual maintenance.
Commercial systems such as Symantec Brightmail and Postini also don't require messing around with blacklist/whitelists and have zero false positives and minimal misses.