More Spam Tricks

More Spam Tricks Why do spam filters sometimes miss email spam that looks obvious? The question comes up frequently.

Spam filters continue to get better at what they do. Spammers continually come up with new tricks to foil them. Here's a few of my least favorite.

Junk HTML
Spammers sneak junk html in-between letters in words. Usually the extraneous HTML can be removed. This reveals the actual text for evaluation. Sometimes spammers put in additional characters formatted so they actually don't render to the end-user. This can make things really tough.

Message in an Image
This is kind of self-explanatory. The entire message is included in an image. The fact that an image is present in the message can help indicate it is spam. However, this isn't enough on its own as many users include pictures in their signatures. I really wish they wouldn't do that. Tip - including three or more smileys at the bottom of your email is a great way to get moved to the bottom of the support queue.

Hopefully image intensive messages are on the decline. Microsoft has included on option in their Outlook spam blocker to not render images.

Bayesian Filter Tricks
Bayesian filters use statistics to determine how likely a message is spam or not. Offending phrases are only according to the percentage of the email message they make up. When 500+ bogus words are included it throws the averages off. Often the text is copied from other popular works. This makes it grammatically valid, and extremely hard for antispam programs to distinguish from a real letter.

SpamButcher's spam email blocker use fuzzy logic as opposed to Bayesian and is not susceptible to this problem.

Back