Andrew Newton knows an absence of consensus when he sees it, and lately its been all over the MTA Authorization Records in DNS, or MARID, working group, which has been attempting to formulate a standard for SMTP authentication for months.
After some weeks in the “Last Call” stage of the process, it became clear to everyone that most of the participants objected to the portions of the proposed specification that had been proposed by Microsoft along with a claim of intellectual property rights and a royalty-free patent license. There were lots of reasons—some good, some bad—but all that really mattered was that the working group didnt like it.
So working group chairs Newton and Marshall Rose have proposed a compromise that uses parts of the spec that were not as controversial. Instead of relying on the Microsoft PRA PRA (Purportedly Responsible Address) algorithm, it allows the administrator to specify different “scopes” that correspond to the different message elements that have been proposed for checking. You want to check MAIL FROM? Go ahead. You want to check PTR? Have it your way. Want to check HELO? Party on.
The compromise is a variation on an older proposal by SPF author Meng Wong called Unified SPF. This slideshow is pretty much all that exists of documentation. Unified SPF was the hot proposal when Microsoft happened along to combine Caller ID to make Sender ID.
In a way its an abrogation of responsibility by the working group because it doesnt specify which scopes to use. For a standard like this to be successful, eventually everyone would have to agree on which scopes are important. On the other hand, you could say that this proposal provides a common framework for administrators to check whatever they want, so the market will decide what scopes are important.
No doubt, if the proposal passes the working group and is adopted widely, a lot of real-world data will be generated about the effectiveness of different scopes and different combinations of scopes. Other useful proposals, such as the SUBMITTER extension to SMTP, have come out of this exercise. And when a crypto solution like Yahoo!s DomainKeys comes along, it too can probably be expressed through this system.
And even Microsoft can walk out of the process and implement their proposal. Their scheme, the object of most of the controversy, is just another scope. If they want to check it, they can. Several other participants expressed support for PRA as well, although perhaps some of that was motivated by wanting to have a single strong standard rather than PRA specifically.
The only one of these schemes with any kind of real-world testing is SPF (which checks MAIL FROM), and even that one hasnt really been out all that long. Both spammers and the good guys have a lot of experimentation to do.
In the wake of the Sender ID debacle, we cant lose a sense of urgency to do something systematic to begin hardening the Internet e-mail system against abuse. All e-mail security solutions begin with authentication.
Security Center Editor Larry Seltzer has worked in and written about the computer industry since 1983.
Be sure to add our eWEEK.com security news feed to your RSS newsreader or My Yahoo page:
More from Larry Seltzer
