When ISPs Think They Know Best

Is your Internet provider protecting you from spam--or censoring your inbox?

Anne Kandra is a contributing editor for PC World. You can send her e-mail at consumerwatch@pcworld.com.

It was supposed to be all in good fun. When Jackie Pettycrew came across a video online that made her laugh out loud, she did what many of us would do: She e-mailed the URL to some friends.

But a few hours later, Pettycrew wasn't laughing. And neither were the half dozen or so of her friends who used AOL. They never saw her e-mail because the online giant's servers bounced her message at the gate. The reason? According to an automatically generated e-mail that AOL sent to Pettycrew, "the URL contained in your email to AOL members has generated a high volume of complaints."

Pettycrew was both mystified and incensed. "[The video file] was a European car ad," she recalls. In the ad, a shiny new auto, in the sights of a pigeon, raises its hood to smack the bird before it can leave a calling card. "There was nothing vulgar or obscene about it. Since when has AOL assumed a censorship role?"

It's a question that more and more e-mail users may be asking as ISPs call in heavy artillery in the war against spammers. AOL has adopted some particularly aggressive tactics, and users like Pettycrew whose messages have been apprehended by server-level filters--as well as recipients like Pettycrew's friends who find that some of their innocent incoming messages are being snagged--are starting to wonder if the online giant has gone from security guard to thought police.

The Public's Will

Not surprisingly, what Pettycrew calls censorship, AOL considers good customer service. According to AOL spokesperson Nicholas Graham, the service's mail filters are based in part on member complaints. "When the number of complaints about a given message or link reaches a particular point relative to the number of times that message is sent," he explains, "the filters kick in and the message is blocked. Clearly, in this case we were responding to a large number of members who felt that [the video] was objectionable." (Note: PC World provides some content for AOL.)

Spam is often offensive, but not everything that's offensive is spam. So when an ISP blocks e-mail based on its content, isn't that tantamount to censorship? "Balderdash!" responds Graham. "If anything, it's the opposite [of censorship]. We're responding to feedback from our customers, not our own positions on the issues. We have to put control in the hands of our members because that's who the service is for."

Clearly, it's good for consumers that ISPs are tightening up their spam filters and trying new techniques--many of them highly effective--to weed out junk mail. But if protection against spam means that you may not get an online coupon you requested or that a racy joke you sent will be squashed, it may be time to consider whether your ISP's filtering methods give you enough control.

There's a very good reason why ISPs are turning to the kind of "let the tribe decide" approach to screening spam that Graham describes: It works. In our roundup last month of stand-alone spam filters, we discovered that the most effective filters incorporated some kind of feedback from users to identify spam. It's one of the techniques employed by Brightmail, a large-scale spam filter used by many major ISPs, including EarthLink, MSN, and Adelphia.

The feedback method is helpful because the definition of spam is squishy--a message that you delete I may think is a worthwhile offer. By polling users, spam guardians can identify the messages that the vast majority of users consider spam.

I spoke with representatives of a number of ISPs, and all of them bragged that "false positives"--legitimate e-mail messages that their filters mistakenly identify as spam--are extremely rare.

AOL's Graham says the company does not maintain specific numbers on false positives. "But," he says, "the 2 billion e-mails that we detect [as spam] on a daily basis are almost all spam, and almost all of it is based on what our members have reported to us as spam. Many people want to claim that 'good' e-mail is deleted by AOL, but they don't understand that good mail to them is many times deemed 'bad mail' by our members in droves."

But what if you and your friends disagree with AOL's vox populi? If Pettycrew's friends would have gotten a laugh out of the pigeon video, should it matter that lots of other AOL users found it offensive? AOL's Graham points out, "There's no such thing as a 'universal joke' that's acceptable to everyone." But in our pursuit of a spam-free inbox, will we find ourselves limited to communication that's acceptable to everyone?

The philosophical questions that underlie spam filtering merit continued debate. But from a practical perspective, you want to ensure that the e-mail you send doesn't get diverted into some digital Dumpster. Here are a couple of tips:

Check your subject line. Obviously, any mention of debt consolidation, enlargement devices, or rich Nigerian relatives is a sitting duck for any filter worth its code. Other red flags: headers that use ALL CAPITAL LETTERS or vulgar language.

Don't send file attachments unless they are absolutely necessary--they're filter magnets. So next time you decide to forward a funny, paste the text or image into the body of the message instead of attaching it as a separate document.

If you don't receive an e-mail message that you know a friend or company sent your way, check with your ISP. If you feel that the company's antispam efforts have crossed the line into censorship, let it know. And if the company's answer isn't acceptable, start looking around for a new ISP that will take your concerns more seriously. There's no reason to pay for spam control that's out of control.