A bit less than four years ago, Facebook decided to get into the email business. I wrote a blog entry at the time, warning it a bit about what it was getting into.

Facebook announced the closure of its email service earlier this year. Facebook emails will automatically be forwarded to whatever email address Facebook users have listed as their primary one.

I warned it about the technical complexities of email, and the pitfalls that required email veterans on the team to avoid repeating. I really thought the biggest problem it faced might be technical. I figured that with its brand, it certainly had ‘market permission’ to enter the email space.

But we never got a chance, really, to find out how good Facebook mail was, because almost nobody used it. I didn't see that coming, because I thought that there was potential value in integrating Facebook messaging with email. I should have known better, though, because I made a similar mistake back around 1982.

In 1982, I was developing and maintaining email clients for a couple of timesharing systems of the day, when I discovered that two future friends were developing a bulletin board system and a calendaring system for the same environment. We decided that what was really needed was to integrate all three into a single user interface that streamlined everyone's communication.

We called the system BAGS, after our last names - the Borenstein Anderson Garlan System. It was modestly successful, and was maintained for many years after I moved on. But people didn't use it as a single user interface. Some used it for both email and bulletin boards, but separately, as if the fact that they were all one program was something they needed to work around. Like Facebook, we found that users just weren't drawn to the kind of ‘universal interface’ that draws computer scientists like moths to a flame.

It turns out there are good reasons why people have always had multiple communication mechanisms. The characteristics of a communication technology, coupled with the community rules, standards, and customs that develop around that technology, inevitably result in a mechanism that’s better for some things than others.

If you need to send me a message, what's the difference between email and instant messaging? It's not just a matter of whether you're using a laptop or a phone, because either can be used either way. But when you're using a laptop, you're likely to be in a more relaxed or serious environment, so it's natural to compose an email, which is likely to be longer, more nuanced, funnier, or otherwise more complicated than seems right for an instant message. On the other hand, if you're running across an airport, dashing off an instant message will be rather more appealing. And if you're like me, you'll sometimes dash off an instant message to yourself, reminding you about a more complex email you need to send.

Facebook was one of the pioneers of social networking, which as a communication medium is radically different than email. People use it to communicate with whole groups of friends or relatives at once, and they think of themselves, generally, as operating in a semi-public forum. Email feels (rightly or wrongly) more closely controlled and limited in distribution. Combining two media that differ in important aspects is a recipe for confusion, and users intuitively resist it.

The email world and the Facebook world often leak into each other, but that doesn't mean users want them to merge. The best email programs have user interfaces that are highly evolved to what users expect from an email medium - features that make it well suited to complex threads of discussion, but less well suited to ad hoc group discussions with your friends' friends. Merging the two doesn't necessarily make things simpler - the features of one can actually get in the way of the other.

The bottom line is simple: email is very, very important to a lot of people, and they are wary of anything that might weaken its usefulness. If Facebook had set up its email service to be entirely independent of the social networking system, it might have been able to attract users, and then gradually introduce carefully selected features that connect the two in useful ways. Perhaps that's how it’s planning to approach its acquisition of WhatsApp; if it’s sufficiently cautious in how it integrates the two services, it might well succeed.

So, while we are saying goodbye to Facebook mail, perhaps it’s not forever. There's still plenty of room for innovation in email, in social networking and in the spaces in between. But it’ll take a more open, incremental and modest design to succeed.


Welcome, Facebook, to my home of the last thirty years -- the wild, wonderful, wacky, wheels-within-wheels world of modern email!

It sounds like you plan to be here for the long haul, so I hope you brought everything you need:  good programmers  and deep pockets shouldn't be a problem for you, but you also need people who understand the many important email standards (including the new and emerging ones for domain-based email signatures and non-western character sets for email addresses and domains), the complex interplay between spammers and spam-fighters, and the remarkable variety of ways that email composed on your system will appear on the hundreds of other platforms in the world that might receive it.

From the outside, email seems pretty simple -- there's a To, a From, and a few other relevant fields, right?  But almost every aspect of email harbors a "gotcha" -- some fundamental, some a legacy of email's evolution, but all critical if you want to "play nice" and have your email interoperate well with everyone else's.

Unfortunately, most of these pitfalls are known primarily through Internet folklore; when old email hands get together at an IETF meeting, it's likely that someone will pass out before anyone runs out of horror stories.  If you haven't hired any experienced IETF email gurus, I urge you to do so, or to start sending your newer-to-email programmers to IETF meetings, or -- best of all -- both.  I hope I'm wrong, but the absence of any Facebook attendees (or anyone from their recent Zenbe acquisition) at the last two IETF meetings makes me fear you may not know what a tar pit you are stepping into.

Here at Mimecast, it is amazing how many violations of the standards we have to cope with every day.  When possible we try to work with the offending vendor to fix the problems, but sometimes we have no choice but to simply do our best to cope with their mistakes.  For example, imagine that you get a mail message that consists (irrelevancies removed) of:

In the absence of the key MIME header fields "Content-type" and "Content-transfer-encoding" there is really no correct way to interpret this message. Both the body and the subject line violate the 8 bit restrictions of SMTP (assuming the 8 bith extension hasn't been used), the subject violates header encoding rules from MIME, and the body itself should, technically, be interpreted as US-ASCII, which is just obviously wrong in this case.  Worse still, the line "" breaks the rules for continuation headers and, arguably, makes the subject part of the body!

For any specific example, a human being can usually figure out how to fix a message, but the variety of mistakes is unbounded and apparently impossible to anticipate.   A recent Internet Draft has begun to try to pool community knowledge about these mistakes; I hope your guys have read it carefully!

So, I welcome the facebook team to the never-ending struggle that is a smoothly-operating email service.  I look forward to the new ideas you are bringing to the table, and I hope that some of them will prove valuable enough to bring enhancements to the whole community (another reason to come to the IETF meetings).  I even hope that you succeed in revolutionizing the email world, because that would be fun to watch.  But I don't think you have much chance of doing that if you can't first deal with the complexity we all confront today.