Pursuing Reliable Email: How can we leverage the user?

Posted in The Gnovis Blog

When I was a fledgling Internet user, reveling in my new 2400 baud rate modem, I decided to play a trick on my friend Josh. I set my computer clock forward an hour, and then reconfigured my email account, replacing my name and email address with his. I clicked okay to save the changes, and much like that now famous episode of The Office, I began to send Josh emails from himself.Email is one of the most ubiquitous uses for the Internet, but it is also one of the most problematic. The problems for users tend to exist in three areas: SPAM, authentication, and privacy. Or in other words, how do we make sure that the email we get is 1) what we want, 2) from who it says it is, and 3) has not been tampered with while in transit. We are familiar with SPAM (estimated as 85% of email volume) and hear about its associated costs. Privacy issues range from simple tampering scenarios, where a server might change message content, to larger international security and surveillance issues that are not limited to email. Email authenticity is most frequently discussed as it relates to phising, but would also cover the frustrations my friend Josh experienced.While these three areas are highly interrelated, the focus is typically on the elimination of SPAM, after all, that is what users are complaining about. The solutions are also typically server oriented. SMTP-AUTH, SPF, and CSV are just a few of the notable attempts among dozens of ideas. Most of the proposals out there, however, are contingent on wide and/or universal adoption, and as such have had little impact. So why aren’t we adopting these solutions? By design, each of these solutions block email somewhere along its path. This means email administrators must deal with the terrifying risk that legitimate email might fall through the cracks. This is far from trivial. If you were to choose between loosing one legitimate email and weeding through 100 junk messages, which would you choose? What about 1000? This is the threat of a “false positive”, an email that was filtered when it shouldn’t have been. The impact can be disastrous. If false positives demonstrate anything, it is that the reliability of email delivery is more important to us than its content. As such, any technology intended to reduce SPAM can only do so provided that it does not impact our sense of guaranteed delivery.The fundamental problem is one of judgment. SPAM filters can only remove what they judge to be the obvious offenders, exposing the rest for the user to read or delete. Engineers have spent a lot of time trying to improve the algorithms that leverage email content in the hopes of distinguishing between obvious offenders and potential offenders. It is up to us as users to distinguish between real email and potential offenders.It is my opinion that the user has been under-utilized in the fight against SPAM. We should be leveraging users in addition to the emails they receive. Many web based email services (MSN, Yahoo, Google) have begun to mark email from individuals in your address book in order to do just this. But what else about our users can help?Here is one proposal: Let’s leverage the user’s identity. The most important email we get comes from actual people. They have names, addresses, and other information that is verifiable. When you give Amazon your credit card number, you do so knowing that Amazon exists. Visitors to Amazon can check Amazon’s SSL certificate, and shop comfortably knowing that Verisign (a certifying authority) has verified that the website is what it claims to be. A similar type of certification is available for email. Digital certificates have been created to handle email issues related to authorization. The assurance that the email has arrived securely is just icing on the cake. And what about SPAM? It isn’t too large of a leap to imagine an email client that would prioritize inbox items based on the presence of a digital certificate. There is a catch. Digital certificates cost money. But what if Hotmail offered to foot the bill? It would ultimately be a business decision, but there is an opportunity for one of the major free email services to raise the bar on email. Hotmail could provide users with the option to certify their email account and enable digital signatures free of charge. Mail services are funded by ad revenue, and providing this option could dramatically increase membership (“Hotmail: Signed and Secure!”). How long would it take before other services started doing the same?There are, of course, a number of different approaches to the problems described above. This is just one. Email is complicated both technologically and socially, but by adding the user identities back into the equation we might be able to better assess the integrity of the email we receive. How viable is this solution?