Friendly Fire in the War on Spam – Part 1: Filters

Part 1 | Part 2 | Part 3

Not only is spam incredibly frustrating, it continues to hamper legitimate business in new ways. Sending out honest email newsletters is becoming increasingly difficult for a number of reasons. Most of these are the direct result of entanglements with tools and protocols that are in place to protect users from spam.

A big problem with spam prevention today is that most anti-spam software can report "false positives" where legitimate email is labeled and treated like spam. The spam filtering software looks over an email message if it fits the profile of spam. This can be done at the server level, an IT department's Exchange server, or at the client level, a copy of Outlook or Eudora on a desktop. The problem is that determining what is spam and what isn't spam is a very delicate job. Unfortunately most anti-spam software just isn't up to it. You run the risk of either tossing legitimate email in the trash without seeing it or only eliminating a small fraction of incoming spam. Granted that cutting out half of the spam you receive is a positive step but it's hardly an ideal solution especially when you loose real email in the process.

Dealing with incoming spam:
In my experience, most anti-spam software causes as much hassle as it prevents. This is especially true for people like me who depend on email for new business leads. Those of us in this situation simply cannot afford a false positive because that costs us business. For a long time my solution was to run a spam filter on my email client at a low level. Set this way it would flag obvious spam and let the rest through to my inbox for me to deal with. I still had a ton of spam coming into my inbox but it wasn't quite as bad. Even with a low grade filter I had to skim through my spam folder once per day to look at hundreds of emails to avoid throwing out something important.

The problem with these types of filters is that the software is making a guess. Emails that have blank subjects or subjects like "hello" or "hey" look a lot like spam. Heaven help you if you actually happen to be in the mortgage business or work for Pfizer.

Better than filters:
Since software isn't up to the task of determining what is spam and what isn't spam what can we do? Cloudmark has developed a unique and quite brilliant solution to the problem. Rather than rely on software to make judgment calls on what is or isn't spam, Cloudmark leverages the opinions of millions of email users. Here's the process in a nutshell: The vast majority of spam is a basic message that gets sent to a large number of people. Everyone with Cloudmark's software has the option to mark an email as spam. When a message gets marked, Cloudmark reads this as a "vote" to label a message as spam. When a message gets enough "votes" Cloudmark's software gets updated and removes future occurrences of that message from everyone's mailbox. The genius of this is that Cloudmark only marks something as spam if several people have marked it. If you get a new business lead from an individual, there is no way that email can be flagged as spam by Cloudmark because it is a unique message. No one else has received the email other than you, so no one else could have marked it as spam. No more false positives!