Tips, tricks, tools, and techniques for Microsoft Office

Last week's slew of spam made me realize the importance of an e-mail filter. Spam Bully ($29.95) is one of a number of products that works with both Outlook Express and Outlook, and combines a white/black list of senders with a Bayesian filter that make your inbox load lighter. Ultimately, however, my experience was disappointing. I tested with Outlook Express 6.0.

Messages thought to be spam are automatically removed from your inbox and sent to a Spam folder. Spam Bully uses a two-pronged attack to direct junk mail into that Spam folder. The first approach uses lists. The white list permits messages from people on the list (or from all addresses in a domain in your list). If you respond to an e-mail, the sender's address is automatically added to the white list. A black list does just the opposite, automatically marking as spam messages from addresses in this list. You can add an address directly to either the white or black list.

The second prong of the attack uses a Bayesian filter that learns patterns over time. At installation it has already learned from 35,000 spam messages, the company says, but as you teach it, it conforms to your e-mail patterns.

How does this teaching work? Simple: highlight messages in your inbox that are spam and click on the "Spam" button (added to a toolbar during installation) and the message is moved to the Spam folder. Likewise, legitimate mail that's been mistakenly sent to your Spam folder can be moved back to your inbox by clicking on the "Not Spam" button. (The Tech Support discussion boards hint at a version 2 that will handle multiple mailboxes, but for now it only works with your inbox.) Of course, the more e-mail you send and receive and the more times you click on the "Spam/Not Spam" buttons, the smarter the filter should become.

The trouble is, Spam Bully’s performance is uneven. It was smart enough to learn that I didn't want to refinance my home and wasn’t interested in any of the dozens of ads for explicit sexual products or services (you know the ones I mean). Unfortunately, during last week's deluge, I ended up having to set up an Outlook Express rule to handle the latest spam crop -- looking for a phrase in messages triggered by the Sobig virus. Despite indicating to Spam Bully that several of these messages were spam, the program just didn't get it (so to speak). Likewise, despite selecting many messages and clicking the "Spam" button, the program never caught on that several e-mails in foreign languages should be removed from my inbox.

Worse, it was often overly aggressive. A message as simple as "Want to go shopping today?" was considered spam, and no amount of learning seemed to make things better.

In addition to the names on your black list, the program can be set to bounce from known spammers, trying to fool the sender by generating a message that looks as though your e-mail address is invalid (when, in fact, it isn't). Whether mass mailers are fooled is anyone's guess.

The program gets into murkier waters when it tries to send challenges to those who have sent you e-mail. If Spam Bully cannot figure out if a sender is legitimate, it can be set to send out a "challenge" message, to which the original sender must reply in order for your message to get through. (Even if the sender enters the correct challenge word, another option lets you either automatically or individually add them to your white list). I could only get the feature to work intermittently, simply because innocent messages (such as shopping invitations) were sent to the Spam folder first, thus bypassing the challenge logic. Sometimes the correctly entered challenge word released the original message, but at other times it did not.

Spam Bully offers a chart showing how many messages it has trapped on a daily, weekly, monthly, or yearly basis, and the Outlook version can block attachments.

The company's Web page (http://www.spambully.com/register.php) offers a two-week free test. Using Spam Bully for the past two weeks was better than nothing, but the frustrating lack of control (it doesn't support a blacklist of words or phrases) and its inability to recognize some obvious patterns means I'll keep looking for a better filtering solution.

-- James E. Powell

^ Top of page ^

If you like this newsletter, share the free edition with your friends and co-workers. Just point them to:

^ Top of page ^