Given some of the press releases around, I suppose it can sometimes be a really fine line deciding whether a mail is a press release or spam. I don't envy you. Good article by the way.

Well - why on earth do you continue to use the ISP's email service? Why not just use them as a pipe, and get a more useful email host elsewhere? Some have anti-spam solutions like SpamAssassin installed for you. Get a shell account, and you can usually install it yourself.

For a techy publication, using the ISP's default service seems a bit, well, wimpy, really.

Not only do Bayesian filter out some legitimate mail, the spam mills particularly target the Bayesian algorithms with deliberate misspellings. Mail washer with the option to highlight mail from known open relays and known spam DNSs before downloading from the server keeps my network spam free and helps reduce mal-ware intrusions. There are additional spam reporting options, Spamcop and Blue frog. Spam cop doesn’t seem to do much except bolt the stable door after the horse etc. etc. I've just started trial in the new link to Blue Frog, a system for mass posting of opt out requests to "encourage" spammers to cleanse their address lists of Blue Frog Users. No noticeable change in spam at the server, still 3-5 a day.

I 'm pretty certain that my address got harvested because of the habit many people have piling all the addresses in to the “To” field. Mail program providers could help by enabling the “BCC” field as a default rather than the user having to hunt through options tabs to enable it.

With mail washer, I still make a visual check before downloading. It's not a cure but a damn good pro tem spam laxative.

Mirapoint is "perhaps 80 per cent to 90 per cent effective in identifying and junking spam messages"?

Wot no quantitative study?

Speaking about approaches to anti-spam filtering, I still think that Bayesian is the best way to go. I've had nice luck with Spam Bully http://www.spambully.com which uses Bayesian as well as a few other techniques.