Spam - Content Still Matters

In recent years, systems intended to block junk e-mail have gotten pretty good. This shouldn't be surprising considering how obvious most spam is.

"Click Here"
"Male Enhancement"
"Special Offer"
"Dear Subscriber"

In SpamButcher and other top-tier applications, there are probably about 1000 more data points for each of the above examples. Most spammers still use standard marketing-speak when pitching their wares.

Beyond that, there seems to be specific language spammers use to suggest that what just arrived in your inbox isn't in fact spam e-mail. Well, perhaps by some definition, you are a subscriber. Maybe it just sounds better than, "Dear poor soul who had their e-mail address sold to us by Uncle Joe's Hobbies before they went out of business."

The language of spam is also the language of spam filters. Don't tell anyone, but with a decent statistical processing engine and a sufficient dataset, putting together a mostly-effective spam fighting engine is only about half as hard as rocket science. Bayesian, fuzzy logic and other systems all have their pros and cons.

Periodically, super-scrambled spam shows up and seems to break all the rules. It's almost as if the author analyzed common filtering rules to ensure their e-mail doesn't set of any alarms. This may be due to at least one popular filtering system being entirely open source. This is not intended to be critical of open-source software; it does however sometimes have unintended complications.

So, if this super-scrambled spam slips by efforts to block spam e-mail , why is most spam still so easy to detect?

For one, many users don't use any substantial anti-spam technology. For them, even obvious spam still lands in their inboxes. As long as this market exists, spammers will still have people to target without worrying about their e-mails being deleted.

Secondly, much of the super-scrambled spam borders on being unintelligible. I often wonder what kind of fool purchases goods from a spammer. That prospect seems all the more foolish when spammer can't spell or even complete sentences.

Back