Spam Fighting Tools

This post goes hand-in-hand with a previous post where I list some of my favorite Internet related tools. In the following paragraphs I briefly describe some of the spam fighting tools I use.

Protect Your E-Mail Address

Spam is a problem as you've undoubtedly known for some time now. Part of the problem is that many websites force users to provide a valid e-mail address when registering and then turn around and sell your e-mail to spammers. Or even worse, someone at the company grabs a list of all of the e-mails and sells them privately. I'm pretty confident that this sort of thing happens because I've seen e-mail addresses that I've only given out once show up in spam. This is where SpamGourmet comes in. With this service, you can link a valid e-mail address to any number of made-up e-mail addresses. You never have to give out your real e-mail address and junk e-mail, beyond a certain count you specify, sent to the made-up e-mail addresses is automatically deleted before it ever gets to you. In this way you can go through whatever steps are needed to register for a service and then never worry about spam from that service again. This service is free.

Anti-Spam Tools

To deal with the spam that you're already receiving, I recommend taking a look at Cloudmark Desktop which filters e-mail as they come into your e-mail client whether you're using POP3 or IMAP. This is the only spam tool I've found that doesn't need to be trained. Its power lies in that it uses the input of hundreds of thousands of users that collectively decide what is spam and what isn't. This system works better than Bayesian filters as humans are far better at spotting spam than any algorithm. This service used to be free, but now has a yearly charge that was around $40 the last time I looked.

The second spam tool I've use is called PocketKnife Peek. This is a free Outlook add-on that lets you “peek” into an e-mail that you suspect is spam or perhaps contains a virus. The benefit of this tool is that e-mails with malicious code in them or ones that contain web bugs won't function, but you'll be able to see the text of the message so that you can determine if it is legitimate.

If you use CloudMark desktop, you won't need this next tool: SpamBayes, a Bayesian-based system for classifying what is and isn't spam. Bayesian filtering is quite common and you'll come across a lot of solutions that use it (for example Spam Assassin used on Linux servers). The core of this technology is that it learns from the e-mail that each user receives and develops rules specific to that user. It also adapts the rules as the spam changes which makes it difficult for spammers to get past the filters. The biggest drawback is that it needs to be trained by you. Also, very short spam e-mails seem to get through more regularly than longer ones. This program is also free.

Greylisting is Flawed

Greylisting is a technique used by some e-mail providers to reduce spam. The process behind it is quite simple: when an e-mail is received, notify the sending mail server that the e-mail needs to be resent (i.e. temporarily rejected), and once the second copy has been received let it pass through to the user's inbox. The theory is that spammers won't bother to resend the e-mail a second time while legitimate senders will. This differs from whitelisting which allows an e-mail to always get through and blacklisting which always prevents an e-mail from getting through. White and blacklisting are typically manually done i.e. you have to provide each e-mail and mark it appropriately. Greylisting, at least in theory, promises to eliminate the manual efforts.

All fine and dandy, right? Not so fast. The flaw with greylisting is that it assumes that legitimate e-mail senders will send a second copy when prompted to do so. The problem is that within just one week of using greylisting (which my e-mail provider recently activated) I've discovered that e-mails from Google are not getting to me. In particular, news alerts that I configured and opted-in for are disappearing.

I've concluded this by conducting a multi-day experiment. When greylisting is on, no alerts. When greylisting is off, alerts resumed. I flipped the switch on and off a few times over several days and the results were consistent every time. Why isn't Google playing along? I don't know. Perhaps it's because greylisting in effect doubles the processing and bandwidth required since every e-mail needs to be transmitted twice. There's probably a way for me to whitelist a bunch of senders, but that seems cumbersome and puts the onus on me to figure out who should be whitelisted. Thanks, but no.

If you're interested in learning more, you can check out Greylisting.org. The site's content isn't as deep as I'd like it to be, but it does have links to other articles that are as good a starting point as any.

In closing, I'd recommend trying out one of the free solutions and seeing if they do a good enough job. Part of this will depend on how many e-mail accounts you have and how much time you're willing to manually filter e-mails that pass the various filtering tests. If you have more money than time, I'd go straight to CloudMark Desktop as it has the highest rate of detection with the lowest rate of false positive compared to anything else I've tried over the years.