Microsoft Chairman Bill Gates on Tuesday announced a proposed open standard to deter e-mail spoofing, a k a caller ID for e-mail. Microsofts proposal pitched to the security experts at the RSA Conference falls right in line with the emerging industry consensus that changes must be made to the e-mail infrastructure in order to make serious progress in the battle against spam.
The actual "Caller ID for E-Mail" specification calls for a system that has more in common with competing proposals than it has separating them. Like the SPF (Sender Policy Framework) specification, the basic mechanism involves requiring domains that send messages to add records to their DNS that will allow recipients to determine the addresses of the authorized sending servers. Recipients can then see if the sender of a message is really authorized to do so.
SMTP authentications systems like Caller ID wont stop spam in and of themselves, but they will make it possible for reputation systems from vendors such as Brightmail Inc., as well as simpler blacklists and whitelists, to be more effective. With a far smaller amount of spam surviving such challenges, filtering would also become more effective.
Authentication would also put up a huge barrier to e-mail worms like MyDoom, Bagle and Sobig that have comprised the majority of malware attacks in recent years. These worms all use SMTP engines built into the attack code itself. With Caller ID, none of them would authenticate properly in their current form, and its not clear that virus authors would have an effective way to send authenticated mail on arbitrary systems they infected.
There are differences between Caller ID and Sender Policy Framework, but they arent fundamental ones. The format of the records is different, and the method of comparing message headers to those records also differs. There are advantages and disadvantages to both.
At the same time, these differences wont change the basic fact that to implement either standard, owners of e-mail domains will have to upgrade their SMTP mail server (often known as a Message Transfer Agent or MTA) to one that supports the standard.
Microsoft should be able to provide this capability in their Exchange Server by means of a plug-in, although its not clear which versions of Exchange they would support. Like SPF, Caller ID would require changes in the Message Transfer Agent to deal with forwarded mail accounts (accounts from which mail is simply moved on to another account, rather than being collected by the users mail client). And like SPF, Caller ID would authenticate just the domain of the sender, not the user within that domain.
With so much in common, why doesnt Microsoft just adopt SPF? According to George Webb, Group Manager of Microsofts Antispam Technology and Strategy Team, the company decided that the differences are important.