Last year, MAAWG published a white paper titled Trust in Email Begins with Authentication [PDF], which explains that authentication (DKIM) is “[a] safe means of identifying a participant-such as an author or an operator of an email service” while reputation is a “means of assessing their trustworthiness.”

Reputation systems based on IP addresses, including Return Path’s Sender Score, are used by many ISPs and anti-spam vendors to determine which mail to accept, which to reject, and which to subject to additional filtering before making a delivery decision. There, the identifier is the IP address.

The reason this sort of reputation works for delivery decisions is that it’s an attempt to predict whether the sender of a message can be trusted to send mail that the recipients want—or, more accurately, whether the IP address of a message can be trusted to send mail that the recipients won’t complain about. We also mix in the concept of safety, largely in the form of how likely it is that the IP address is sending phishing scams or similar bad stuff.

In part 1 of this series, we described how the DKIM “d=” identifier brings us closer to knowing who sent a message, because it can be tied to the company or person who registered that domain name.

Reputation or certification based on the DKIM d= identifier will have the same goal—and will be more effective, because it will be tied to the signing entity rather than a single IP address. When ADSP is applied, that signing entity could be the author domain (see part 2). If not, it’s still a useful method for determining whether to trust the message. Any d= domain who regularly signs trusted messages becomes trustworthy, and vice versa.

Plus, d= reputation is portable—the owner of the d= domain can use that same identifier on multiple IP addresses, even bringing it to a different ESP (as we described in part 2), without having to start over from scratch or to “warm up” IPs.

While not absolutely perfect, reputation and certification based on d= will be far more accurate, effective, and convenient than when it’s based solely on the IP address. But, does a trustworthy d= domain indicate a truthful message? Stay tuned for part 4.

(This article was originally published by Return Path)