I get spam. You get spam. We all get spam. And not the meat kind. Nobody likes it, but depending on who you ask about 80-90% of all the emails sent are spam. And of course, you have better things to do that sort out which emails are ones that are important to you and which are asking you whether you want a penis enlargement for the umpteenth time. So this is where SpamSieve comes in.

Now OS X’s Mail has a built in spam filter, which can do a pretty good job, but you do have to watch it. For some people it works, for some it doesn’t. I’m lucky enough to be the latter. It got to the point where almost all emails were being put into my spam box, even ones I’d told mail to specifically not include. I was starting to feel like nobody loved me. So, fed up of trudging through manually sorting my email I decided to give SpamSieve a whirl.

SpamSieve works using a technique called Bayesian analysis. Without going into the workings in too much detail, this method analyses how likely certain words are to appear in spam messages, making it highly accurate, and more importantly, it learns as you receive more email.

Installing SpamSieve is an easy task. You open the application, close your email and simply choose one of the install menu items. SpamSieve supports most of the popular email clients such as Mail, Eudora, Mailsmith and Entourage. Once installed you create a new mailbox called “Spam” and set up a single rule to help direct your email through SpamSieve. The next job is to teach SpamSieve what you class as spam and what you class as important mail. This is a simple job of selecting a mix of spam and good mail and marking them as such using a menu item (the developer recommends a mix of 2/3 spam in order to work properly).

So just how accurate is it? Well, the simple answer is very accurate. In the first week or two you do need to check your spam box just to make sure that nothing is getting put their by accident. But after that you can just sit back and relax and let SpamSieve do the rest. SpamSieve also lets you see some very nice statistics to measure how well it is performing. Here is a selection of my stats:

Filtered Mail
199 Good Messages
824 Spam Messages (81%)
29 Spam Messages Per Day

SpamSieve Accuracy
12 False Positives
2 False Negatives (14%)
98.6% Correct

Showing Statistics Since
08/06/2006 4:35 pm

The most important of these figures is the False Positives figure. These are good emails that are marked as spam (false negatives being spam marked as good). In all fairness all of the false positives were ones that are understandably confused (forum auto replies, news letters etc) and simply marking them as good fixes this problem in the future.

SpamSieve also features colour labelling of spam depending on how obviously it is spam and also a black/white list system, allowing you to set up which emails are spam and which aren’t, though SpamSieve quite happily does this for you, basing it of information such as your address book and your past recipients. It also nicely integrates with the Growl framework, giving you a notification when real mail is received.

Overall SpamSieve is a great application. It sets out to do a job and does it near perfectly. For anyone who receives a lot of spam it truly is a must have application and the low price of $25 is more than worth the time you’ll save.