Jarvis, Handle My Email

On opening one of my more-than-a-few email accounts, that little red number by Inbox — or the marching rank-on-rank of boldfaced unreads — sparks a tiny twinge and a little wave of fatigue. Yes, Mom, I know I haven’t read them all. (No, Mom, I’m never going to read them all.) You?

Everyone hits that quiet email crossroad sooner or later — the moment of silent delete, one or two or a thousand messages sent to the big bit-bucket in the sky, unseen.  No sympathy for the anguish of the writer who sweated that artfully crafted never-to-be-read copy, or the angst of the marketer whose stats just took another glacial drip down. Click-delete-silence. Long around January, a few brave souls may revisit their crossroad and try the email version of dieting — “Inbox Zero” — again. Most of us have just surrendered.

With Colleagues Above A Certain Age, I can usually provoke a snort of recognition with a quip about OST and PST files. (Kids: just look away, you shouldn’t have to see this.) We used to have a constraining factor called “disk space” that forced at least the occasional frenzy of deletion. But disk space has floated up into the cloud while costs have dropped like restaurant visits, and that constraint is gone. “Delete” really means “move to the folder called Trash”, because maybe you really will go back and read that in your spare time. Hah.

Net result: more d*** email. That twinge isn’t going away. The Unread coastline is going to keep rising. Are we going to keep retreating up to the cloud, or are we going to find other ways to deal with it?

Technologies and innovations generally aim to solve problems, improve things or enable new things, right? My aim in this post is to make some observations about maybe-not-so-obvious opportunities for innovation in email. 

There’s no shortage of innovation in the field now, to be sure. Deliverability and security have modestly-but-quietly improved with standards like SPF, DKIM, and DMARC. For the most part, we’ve gotten a handle on responsive/mobile challenges — thanks in no small part to companies like Email on Acid and Litmus. Message content is inexorably shifting away glorified mail-merge text to rich, visual and real-time, thanks to better editors like BEE, real-time stalwarts like Movable Ink & LiveClicker and others.

Shiny objects are, of course, naturally fascinating to homo technologicus, so there also are efforts to make email more programmable, interactive and, uh, fun  — AMP for Email, Actionable Emails. Past thoughts on this aside...I don’t see much of a standards-level effort to change the fundamental behavior of email, and I don’t think either Google or Microsoft’s efforts will become standards. But email handling and inbox management? Bueller? Bueller?

I’d originally started down this topic path thinking about how reading has changed. I don’t read every word of every email; you probably don’t either. But reading — certainly reading of email, and probably reading of almost all kinds — had to change, due to sheer volume. That rank of unreads isn’t because of a lack of willpower or skill, it’s for a lack of time and a finite amount of attention that simply can’t be spread across all of that...stuff. Reading isn’t reading any more – it’s scanning, skimming, reading, subvocalization and (I suspect) a fair amount of near-unconscious visual pattern filtering. I don’t think reading is going to change much more; I suspect we’ve adapted our reading skills about as far as they’ll go.

So...is there any hope, or will we all just drown quietly in our inboxes?

The answer to that may be: AI.

I’m not qualified to define AI; you’ve read (or scanned, or skimmed) enough “content” to know the basics about AI by now. For the sake of brevity: given a large enough ‘corpus’ — body of data — programs can deduce and apply patterns that produce human-intelligence-like understanding of similar data. Show a good AI enough kitty pictures and it’ll recognize that your specific cat is a general cat — that sort of thing. Key algorithms have been around for decades, but the field didn’t have enough data (corpuses? corpi?) until relatively recently.

AI is rapidly getting better and better at crunching through big data sets of all kinds — including text and images — to become ‘intelligent’ about that data set. So I find myself wondering if we’d be able to apply that to the “corpus of 1” known as email. 

Could an AI help me with the endless, thankless prioritization job of messages? This may be verging on email fantasy material, but just imagine…:

• A daily “unsubscribe from these?” list of the stuff you never read.
• An hourly “you might have missed” popup of messages from people who always get a response sooner or later.
• An instant “didn’t you mean ___?” for mis-forwards or missing recipients.

Some of this is starting to show up; Gmail’s suggested sentence completion, for example, is pretty darn good. No doubt the algorithmic decisions about which messages go in Promotions, and which in the Inbox, is big AI at work. But they feel like “early days” for a couple of key reasons.

For one thing, Gmail’s “decisions” seem more like rules for everyone than “corpus of 1” AI, derived from my inbox. I don’t want announcements from BEE, Email on Acid or iPost in the Promotions tab; when I do find them there, I read them. Yes, I could drag them to another folder so Gmail would ‘learn’ but if my read-messages data were taken into account, that wouldn’t be necessary.

Assume a large, fantastic AI leap for a moment, though. Say an AI engine successfully trained itself on your email, and derived a fantastically accurate idea how you would like to have messages handled — read this, don’t read that, file the other, put this to sleep for now, delete old notifications & outdated sales, etc. etc. Email nirvana, right?

Well...you tell me! Email is almost comically personal at times. Would you cede some degree of control to an AI, or would you want to have the final say in every decision? Would you “trust” — and that really is the word — that the risk of a miss would be outweighed by the benefits of all that handling? As much of an automator as I am, the thought gives me pause. It’s a big step.

There’s another perspective to AI/email handling as well — the email industry side of the equation. Email marketers don’t universally love the Promotions tab — no doubt an AI-driven decision. Would an even-more-intrusive, 1:1-trained Email Jarvis be a Good Thing for email marketers, or yet another dagger in the heart of their stats? For better or worse, marketing decisions are generally made on the assumption of a human being on the other end of the email string; how would ‘digital twins’ handling at least some of those actions affect marketing strategy, content and the myriad other judgement calls involved in a program?

I do think this is more a “when” than an “if” question. Email is growing, not shrinking, but our ability to handle it effectively is really quite strained. McKinsey put a figure of 2.6 hours spent on email per day by full-time workers; all that time, and we’re still not keeping up; multiply that out and there’s an enormous market opportunity.

Solving that wouldn’t be simple; it is a messy and very, very human problem, with a considerable emotion and habit load. But if we can move toward self-driving EVs doing 70mph, is the Inbox so impossible? 

Thoughts and comments welcome; include your Inbox Unread count for extra points ;-)