Recently we replaced the spam engine in iHateSpam Server with a new engine from Cloudmark. This is a marked improvement -- extraordinary increases in speed and accuracy. The vast majority of customers are extremely pleased with this move.
But it does bring to light one of the interesting side notes on the spam frontier. After I announced the deal, we got approached by other antispam companies wanting us to look at licensing their code. To me, this only reinforced the viewpoint that the antispam business is largely becoming commoditized; engines are a dime-a-dozen, you can pick and choose what you want, and slip it in.
But this is sloppy thinking. The problem is there are disparities in antispam engines. No engines are getting 100% of the spam coming, with zero false positives. There are engines that are highly accurate, getting lots of spam but almost no false positives. Then there are fuzzier engines, which get lots of spam but have an increased amount of false positives. In other words, there's room for innovation.
There's also the pressing need for a change in the protocols. SMTP just doesn't cut it. It's spoofable and flexible for spammers. The only hope we have of a total antispam solution is to go toward a new mail architecture--a secular change in the platform.
More another time.