Antispam -- lively discussion in these blogs

Family squabble time here. Alex Scoble disagrees with me, and Richi Jennings mischaracterizes what I said, which I will however gladly forgive, since his employer is David Ferris, who besides being the authority on all things email is also one of my oldest friends. Contrary to Richi's cavalier misparaphrase, I don't think outsourced spam filtering "blows". I just think one of Alex Scoble's main arguments, regarding bandwidth, is nonsense. He seems to be arguing that $150,000 per month (less quantity discounts) for outsourced antispam services at an enterprise with 5000 employees is cost-justified by saving 32Kb/sec in bandwidth. Huh??????? That's not to say spam-control outsourcing and/or full email outsourcing are all bad, or indeed bad at all. Alex's generic "Why bother with this stuff at all?" argument has a lot of merit, as at many enterprises do straight TCO-based lines of reasoning. Actually, David and I spent a lot of time musing, some years back, about just why full email outsourcing hadn't taken greater hold. Our tentative conclusions were that Microsoft Exchange wasn't scalable enough to allow major outsourcing economies of scale, and web interfaces to Exchange weren't rich enough at that time. Meanwhile, the leading non-Exchange outsourcer Critical Path was a bunch of thieves. (At least they were indicted; I've forgotten whether they were actually convicted and sent to jail ...) I look forward with interest to learning what Ferris Research's current views on the subject of outsourcing are. As for the other area of disagreement -- false positives -- yes, the false positive rate has to be waaaay under 1%. In particular, it has to be so low that the messages lost to false positives simply aren't important in the grand scheme of things, because for most users, if they have to scan their spam folder, most of the benefits of antispam technology are lost. Truly sophisticated email users might be exceptions to this rule -- e.g., if your mail is filtered to dozens of individual folders, having most of the spam concentrate in one saves some time. Or if you rush eagerly to Outlook each time a new mail ping is heard, you might be happy to have the spam be segregated to a non-ping area. But for the ordinary user, who reads email in a batch and not particularly granular way, worrying about false positives is almost as bad as having no spam filter at all. Of course, if a user is particularly porn-phobic, none of the preceding argument applies; it's only applies to those enterprises that view spam primarily as an economic, time-wasting problem.