Deluged by unsolicited commercial, or spam, e-mail messages, Internet service provider (ISP) America Online Inc. (AOL) is trying a new technology for cracking down on one common spammer tool: forged sender addresses, which spammers and virus writers use to bypass blacklists and trick unsuspecting recipients.
AOL is conducting a trial of a new e-mail protocol called Sender Permitted From, or SPF, across its entire user base of 33 million subscribers. The company hopes that SPF will eliminate e-mail forgeries by enabling organizations to specify which servers are allowed to send mail on behalf of their Internet domain, according to AOL spokesman Nicholas Graham .
SPF stops e-mail address spoofing by modifying the Domain Name System (DNS) to declare which servers can send mail from a particular Internet domain. AOL is using SPF to publish the IP (Internet Protocol) addresses of the servers it uses to send outgoing e-mail. DNS is the system that translates numeric IP addresses into readable Internet domain names.
Once widely deployed, SPF records can be referenced by Mail Transfer Agents (MTAs) stationed throughout the Internet when routing e-mail messages from a particular domain to determine whether an e-mail message's source is legitimate or "spoofed," according to Graham.
AOL briefly tested the protocol two weeks ago, before shutting it off to make technical changes based on feedback from other ISPs, according to Graham, who declined to describe the changes.
The program is still experimental and for the time being AOL will not use SPF to filter mail from other Internet domains, Graham said. "(SPF) is just getting off the ground. AOL is interested in putting the proposal out there and getting feedback from stakeholders," he said. Those stakeholders include other major ISPs such as Microsoft Corp.'s MSN, Yahoo Inc. and Earthlink Inc., as well as other major domain owners processing bulk e-mail, Graham said.
The trial is a major test of SPF, which is one of a number of new technologies designed to thwart spammers, according to John Levine, co-chairman of the Anti-Spam Research Group.
SPF patches a hole in SMTP (Simple Mail Transfer Protocol), which is currently used to route e-mail messages from one e-mail inbox to another. Developed in the early 1980s, SMTP was designed to provide a reliable and efficient way to relay messages between host systems using different computer hardware and operating systems.
In recent years, spammers and viruses such as Sobig-F and the recent Beagle/Bagel worm have exploited SMTP's flexibility, easily transposing the actual source of messages with legitimate e-mail addresses from lists that are traded online or harvested from infected computers' hard drives.
The long term benefit of SPF is that, when the technology is widely deployed, e-mail providers will be able to associate reputations with Internet domains rather than with IP addresses, which are harder to track, according to Eric Raymond, president of the Open Source Initiative, who gave a presentation on SPF during January's Spam Conference 2004 at the Massachusetts Institute of Technology in Cambridge.
SPF itself will not stop spam, but it will help other antispam technologies like spam traps, by enabling spam to be tracked back to specific domains and forcing spammers to move to new domains more frequently, Raymond said. The combination of technologies can be likened to a "drug cocktail" that, taken together, may stop spam, he said.
However, the protocol still has problems, including incompatibility with some e-mail forwarding services and Web sites that use mail forwarding features, Levine said. For example, online greeting card services and news Web sites use forwarding to allow readers to send e-mail cards and articles to friends, Levine said.
SPF also causes performance problems under certain circumstances and has features that spammers could exploit to slow down and derail the system, he said. "I would be surprised if SPF survived in its current form, but something like it might survive," Levine said.
Levine is more optimistic about a technology called "domain keys," which is being championed by Yahoo and uses public key encryption technology at the domain level to verify an e-mail message sender.
AOL realizes SPF's problems and is soliciting feedback from other users on it, Graham said. "We want to remind folks that we're in the beta process. These are things that are in consideration as we make refinements and enhancements (to SPF)," Graham said.
AOL's current test of SPF is scheduled to run for the foreseeable future, pending feedback from ISPs, organizations receiving AOL e-mail in bulk and ordinary Internet users. However, AOL will wait for consensus within the Internet community before making any final moves regarding SPF. "It's premature to start looking forward. This is intended to be nothing less than a collaborative, cooperative process," he said.