TNPC'ers, do I have an anti-spam solution for you. Our good friend (and collaborator on utility development), Mike Craven, has devised a most excellent Microsoft Outlook add-in called Junk-Out.

The software comes in two flavors: Basic ($20) and Pro ($30). If you are reading this, however, your level of computer knowledge is such that you should be buying Junk-Out Pro because it is the one product out there that gives you every control over every option you might want (and some you probably didn't think of).

Junk-Out supports Outlook 2000, 2002/XP, and 2003. It installs a toolbar, a menu, and a settings tab in Outlook Tools/Options, plus it runs handy animated icons in your system tray to tell you what it is doing.

The core of Junk-Out is "Bayesian" filtering technology. Thomas Bayes, an 18th Century British mathematician, was a pioneer in analyzing statistical probabilities. Systems that use that analytical approach to this day are called Bayesian. Probability mathematics has become the state-of-the art in spam filters because all the prior approaches have been undercut by spammers.

• "Blacklisting" -- software that would identify and block email addresses or domains used by spammers -- failed because spammers changed addresses faster than the software could update.
• "Whitelisting" -- the reverse approach based on limiting email to people known to you -- failed because of "phishing"-- sending emails disguised to appear to be from legitimate sources.
• "Content checkers" based on looking for dirty words failed because spammers became adept at euphemisms.

So now the concept is to scan content and apply a probability model to the words. For example, the words "member" or "enlarge" in themselves might not raise statistically significant warning flags. But the two of them together in a message would signal trashy content. Moreover, the model is statistical, not substantive, and it is adaptive. If you happen to want member-enhancement products, you can train the filter that such emails are OK by you. Similarly, if you hate, say, baseball, you can train the filter to rule out that material.

Spammers have tried to get around Bayesian screening by creating messages from random words, which is why you are seeing those gibberish emails in your inbox these days. They also have started sending out bogus messages written as if they were replies to something you sent. Junk-Out, however, still effectively detected them (pretty much perfectly with gibberish; the occasional faux replies does get through.)

It took about two weeks to train Junk-Out to operate the way I want, which is to block as much potential junk as possible while still minimizing the number of emails falsely identified as spam, My logic on this is that, since Junk-Out banishes email to a Junk Mail folder in Outlook, my productivity is enhanced if I don't have to be bothered checking that folder regularly to see if I missed an important message. Other people might want to go a different way, and Junk-Out will "learn" accordingly.

Junk-Out, as do many other email screeners, combines the Bayesian approach with whitelisting and blacklisting capabilities. But it goes beyond other programs in its range of capabilities.

For one thing, since Outlook's own rules and spam filters often conflict, Junk-Out is programmed to delay running a few seconds after incoming email is received. That gives Outlook rules time to execute and avoids conflicts.

Instead of the typical one-time seeding of its "friends" list from your Outlook contacts, Junk-Out makes regular sweeps of contacts to keep your list updated.

It monitors your outgoing emails to build up additional names of friends and does so in a more sophisticated manner than competing products. Plus it has nice housekeeping features for moving emails, and will support multiple email inboxes.

The Pro version gives you the ability to fine tune all of the above, gives you automatic updates, and even lets you adjust the Bayesian formula and the words the filter checks. This is where Junk-Out markedly differs (even in the Basic version) from the commercial products such as Norton AntiSpam, which are intentionally designed with few user options.

To illustrate the customization allowed in Pro, consider the friends list. By default if you reply to two emails from the same address within 90, Junk-Out identifies that address as a "frequent correspondent," which is added to the whitelist temporarily. (You would want this kind of feature if, for instance, you have a tech support issue, and want keep emails from and to the technician from being filtered.) After five emails, Junk-Out transfers the address over to the permanent whitelist. Those criteria are fixed in the Basic version, but you can change any of the parameters in Pro.

For me, one of the killer features in Pro is its "Message Details" tool, which gives you a plain text view of email contents and headers without opening it and potentially triggering a virus, worm, or trojan. I use it all the time to peer inside iffy emails to see if that message with the subject "your order" really is about something I ordered or is a scam.

Another powerful feature of the Pro edition is a manual filtering mechanism. If as a Boston Red Sox fan I want to ban any email from the New York Yankees, I can set up a rule to do that. You can set up these rules based on the message body, header, subject, sender, or content. So if the spammers find some way around all other safeguards, or if you have some particular defense you want to erect, you can do it.

While we all wish that spam would just go away, that isn't going to happen any time soon. Junk-Out is the best tool I have found to help you regain control of your inbox.

(c) 2004 Al Gordon.

In addition to his computer interests, Al Gordon is a principal in the Boston-area strategic consulting firm, Mary Fifield Associates, www.maryfifieldassociates.com

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

You can reach Al Gordon at:

al@tnpcnewsletter.com