Applying the metaphor of war to the clash between spammers and anti-spammers is fitting. Receivers such as Internet service providers or businesses introduce a measure to minimize the amount of spam entering their networks, and spammers quickly find a way to circumvent this measure. That pattern has recurred for the past decade: Regardless of the technical measures receivers adopt, spammers readily adapt.
ISPs began changing filtering and receiving policies frequently and surreptitiously. Senders of legitimate, permissioned e-mails are not informed of these changes. Consequently, messages often are returned to the sender, identified incorrectly as spam and routed to the junk e-mail folder, or even deleted with no notification to the sender.
The most recent approach has been the emergence of Bayesian filters, heralded as the best hope for removing inbound spam. These filters also have a learning capability that permits improved ability over time to distinguish between legitimate messages and spam. But, again, spammers developed countermeasures to defeat Bayesian filters such as injection of randomized text into a message or character substitution.
After it became apparent that spammers had subverted Bayesian filters, a conference at MIT in January sought to escalate the spam wars by finding ways to close the loopholes in Bayesian filters. Though the conference drew some of the brightest minds working on the spam problem, it was evident that technical solutions have failed. Solutions put forth at the conference were at best Band-Aids patching the gaping holes in the filters discovered by spammers.
Legislation, too, has failed to abate the flood of spam. The CAN-SPAM Act took effect more than a year ago, but the problem is worse than ever. One provider of managed anti-spam services, Messagelabs, saw spam jump from 40 percent of e-mail in 2003 to almost 75 percent in 2004. Postini, another provider, observed that 22 percent of the e-mails it processed in January 2004 were legitimate but that by December the figure was 12 percent. Spam levels universally are expected to rise in 2005.
Legitimate marketers should be concerned about this trend for two reasons. First, the likelihood of their messages being perceived as spam is increased, which could result in messages being blocked by an ISP or deleted by the recipient unread. Second, legitimate marketing messages may be lost in the onslaught of spam.
CAN-SPAM lets the Federal Trade Commission, state attorneys general and ISPs file lawsuits for violations of the act. All three have taken advantage of the statutory authority. Some states have enacted further legislation pertaining to fraudulent or deceptive e-mail. In a recent case in Virginia, a spammer was sentenced to a nine-year prison term. Yet spam levels continue to rise.
The war between receivers and spammers has made relationships between ISPs and legitimate marketers difficult. A guilty-until-proven-innocent attitude by ISPs is common. The problem is worsened when ISPs do not provide transparent mechanisms for resolving deliverability issues. Also, false positives become an issue when ISPs incorrectly identify messages as spam. A study released by Return Path in March 2004 reported that as much as 37 percent of legitimate, permissioned e-mail was being rejected.
Technical and legislative measures have failed to reduce spam, yet these measures are harming businesses. Filters used by ISPs frequently reject e-mail from legitimate businesses using e-mail in a legitimate manner in accordance with CAN-SPAM and other laws. Even when using double-opt-in lists, businesses find their messages bounced or routed to the spam folder.
Proven marketing language and techniques have been abandoned to bypass ISP filters. Messages with images are suspect. Certain words – “free,” for example – must be avoided. Even opt-out instructions may trigger filters. Instead of composing a message based on sales techniques known to be effective, marketers must be more concerned about how their message content will be interpreted by ISP filters. This reduces the appeal of marketing messages, thus lowering response rates and revenue. Also, senders have had to hire staff specializing in ISP relations and the resolution of deliverability issues.
It is time to devise new approaches. One strong candidate has emerged: Let the market be the arbiter of what is acceptable and unacceptable e-mail.
Brian McWilliams is the author of “Spam Kings,” a look inside the world of spammers. He was the only speaker at the MIT conference to defy the conventional wisdom. McWilliams disagreed with the prevalent belief that if filters become sufficiently effective, then spammers will stop sending spam. “Furtive shoppers like spam!” he said.
McWilliams cited a December 2004 Forrester Data report that said 41 percent of U.S. consumers surveyed had made a purchase as the result of spam they received. He learned that consumers seek out and read messages routed to the spam folder. McWilliams concluded that spam will become like circulars in Sunday newspapers: Consumers will still peruse – and make purchases as a result of – segregated content.
Many consumers are interested in receiving unsolicited commercial e-mail and making purchases from those messages. A large market for unsolicited commercial e-mail exists. But rather than preventing businesses from reaching this market of interested consumers, receivers should permit controlled access to it.
The lack of market friction is the main reason spam is such a large problem. Barriers to entry are minimal, and the market is flooded with spam as a result. No means exist to distinguish legitimate marketing messages from spam. The solution is to create controlled friction in the market by establishing partnerships between marketers and receiving ISPs.
All participants in any well-functioning supply chain should be compensated for providing service. Therefore, ISPs should get a share of the dollars spent on sending marketing messages because they bear part of the costs of delivering those messages. A revenue stream for ISPs would be created that can be used to offset the costs of delivering legitimate e-mail. Marketers benefit because they can be assured their messages will be delivered. And adding friction to the market will act as a barrier to entry for illicit spammers.
Only by creating a true market will there be hope for dealing with spam while still giving legitimate marketers the chance to realize the true ROI from this important channel. Neither technology nor legislation can claim victory over spam, but can spam withstand the inexorable advance of market forces?