Gmail's Permanent Failure: Only Humans Can Build Software For Humans

Editor’s note: Guest author Adam Rifkin is a Silicon Valley veteran who organizes a networking group for entrepreneurial engineers called 106 Miles. In this post he argues that Gmail is perhaps not the best vessel for Google’s social ambitions.

Last week was marred for me by a temporary but super-painful Gmail failure, and the software’s behavior points to why a “more social” Gmail would be a PERMANENT FAILURE. It pains me to write this because I actually believe the Gmail team has been the best web application team long-term in the entire company, and they come way closer than anyone else inside Google to understanding how normal people work and think.

So it’s telling that even within the Gmail team, there is a basic, fundamental, deep-seeded inability to put things together in a contextually graceful way that makes sense to actual (non-Googler) users—in other words, to deliver a great user experience.

Let me explain with a personal experience. I have been a devoted Gmail user for six years, receiving up to 600 emails a day, so Google makes me pay for extra storage. When I failed to understand the procedure for the annual renewal, Google shut off my extra storage and summarily bounced all incoming mail with a “PERMANENT FAILURE” header (contrary to spec, because “full inbox” is a transient fail) until I paid a paltry $5 in extra storage fees.

The worst part is that once I realized that Gmail was failing, I quickly paid, but Gmail took a full 21 hours to restore my service, bouncing me off scores of social networks and mailing lists, and denying my inbox’s existence to all of my friends and professional connections on Twitter, LinkedIn, Facebook, and email itself.

The awesome thing about this incident is that it showcases all of Google’s weakest points as a creator of user-facing applications.

1) Google policy was designed to treat their best users the worst. This Gmail punishment reminds me of the way Google’s cold mechanical cops treated me and my friends during the “Orkut jail” months of 2004 (when users were randomly banned), stripping us of our humanity by labeling us prisoners for no good reason… I understand why heavy users are an inconvenience to Gmail’s scalability—how many people get so much email they exceed the quota?—but for any normal business, the biggest users are the best customers.

2) Google screwed me over five bucks! Google takes in more than $2.3 billion in revenue each month, or $75 million every single day, so my five dollars and 21 hours of pain represent less than one millisecond of revenue. In other words, selling extra storage is not a significant money stream for Google, and it’s an inconvenience for the biggest users. Which is a self-fulfilling tragedy: because it’s not a real revenue center, what incentive does Google have to fix the user experience?

3) No user understands the relationship between Google services, and they aren’t explained in a way that makes sense to an actual human being. Why does my Gmail fail depend on a different product called “Google Paid Storage”? Yahoo! Mail, for which I have also paid for premium services, sends me clear, concise, annual big splashy HTML emails like this:

Your annual subscription to Yahoo! Email is due in a month and your credit card has expired. CLICK HERE NOW [big red button] and pay us $19.95, or you will lose your email storage!

By contrast, Google’s idea of ample warning was this masterpiece of mumbling minutiae:

You’re nearing the end of your year subscription of Google Paid Storage, which gives you extra quota in Google services like Picasa Web Albums and Gmail. Since you’ve chosen to disable auto-renewal for your account, you’ll need to renew your account by visiting the account management page if you want to continue using Google Paid Storage at the current level. If you do not act, your current plan will expire on Aug 24, 2010 and you will be downgraded to the Basic Plan at that time.

Really?! Come on! I did what any normal, busy person with a full email inbox would do in this situation: I ignored the message, because it utterly failed to convey to me that I had a relationship with the sender (Google Paid Storage, whodat?) which led to an immediately actionable item (account management page, ker-wha?) for which the consequences would be extremely dire (downgraded to blah blah what’s for lunch?) . . . Okay, so I wasn’t as vigilant as I should have been, but I submit that no human could easily understand from the message above that all email was going to bounce for 24 hours.

4) There was no customer support whatsoever during the incident, just a message on the receipt for my payment telling me to expect a 24 hour turnaround time. And seriously, what is “turnaround time” about? Is Google telling me that the best technical company on the web can turn off my storage automatically, but cannot restore it automatically? Are human sysadmins provisioning Google Paid Storage by hand??? Does not compute! Yes, I’m ranting now, but stay with me. I have a point.

5) They could have just quietly saved my incoming mail during the outage. Every hard bounce on a social network labeled PERMANENT FAILURE removed me from that social network—goodbye Twitter, Facebook, LinkedIn, and all the others. Every hard bounce on a mailing list labeled PERMANENT FAILURE instantly unsubscribed me from said list. Every hard bounce to a human labeled PERMANENT FAILURE resulted in multiple voice mails and text messages from said human asking me, “What the dilly?” I lost several days of productivity due to my inability to communicate cascading failures because, well, my email was down and Google’s servers instructed the senders not to re-send, so I had to re-establish all those connections and apologize to people manually.

Much of this inconvenience to me could have been averted with vacation messages . . . or soft bounces . . . or better yet, by silently saving all of my email messages for me while we waited for the new storage to be provisioned. Is it really so hard to understand why a human would want a humane solution to a coldly mechanical problem?

Instead Google deliberately chose the technical path (hard bounces) that would be most convenient for them in the short term, while causing me the maximum embarrassment, trouble, anger, frustration, and distress. These emotions of rage and humiliation are likely unfamiliar to the machine that builds Google’s applications.

Essentially my conclusion from this incident is that Google is simply incapable of creating user-facing applications, for the simple reason that only humans can build software for humans. Like Soylent Green, the best social software is made of people. Consider how ex-Googler Bret Taylor found “great user experience” religion when he embraced Facebook:

Taylor noted that he had been “brainwashed by Silicon Valley” before he saw and understood the power of Facebook Photos (he was likely working at Google at the time). He had been thinking like an engineer about the best way to organize photos on the web. But he quickly realized that “the best possible organization of photos is around people,” Taylor said.

“There are ten other industries waiting to have this type of disruption,” Taylor said noting the travel industry, e-commerce, and music as a few of them. Earlier, Zuckerberg agreed. Because of the social element, “every single vertical will be transformed.“

Software is not just a matter of technical capabilities. A tremendous amount of hard work, skill, and taste goes into making software products efficient and comfortable to use:

Understanding how users want to navigate around the application, which tasks to show as buttons versus which to hide in menus, which features should be left out completely, and so on . . . those seemingly minor decisions are often the difference between good software and great software, and the reason great product managers and interaction designers are always in demand.

Even the tone of a product’s error messages—compare “Flickr is having a massage” or the Twitter fail whale, with an eye-glazing Java stack-trace or cryptic Android error message—can change a person’s feeling about the software she or he is using. Feeling is what user experience is all about, and why it is so hard to get right.

What truly scares me is the feeling that Gmail.com is evolving into an increasingly comfortable, warm chamber with a never-ending feeding tube that encourages we, its human inhabitants, to consume and click without surcease. Gmail could very well be the computer-generated dream world that sucks the life out of us, millisecond by millisecond.

What started off as simple, fast, threaded, practical web-based email now resembles a kitchen sink with a hundred settings, dozens of lab extensions, labels and filters and stars (oh my!), calendars, maps, task manager, contact manager, document manager, RSS reader, in-email video player, web clips, instant messaging, video chat, and “Buzz”, all bundled. This past week alone saw Gmail smoosh in more new features, from Skype-like inbox Voice (success or disaster?) to a myriad of magical pixies that learn Inbox Priorities from consumption and clicking behavior.

Friends of mine who follow Google closely believe their next attempt at social software will involve adding far more complexity and functionality to Gmail, making the job of delivering a great user experience that much more difficult. Recent history has demonstrated that Google cannot resist the urge to cram new features into Gmail with alarming velocity. Where does it go from here?

I have nightmares about round-the-clock Googlers on lockdown like at Facebook, busily grafting feature after feature into the mix . . .

  1. Why not add WallProfile InfoPhotos, and Home, and really make Gmail ever more like Facebook?
  2. Why not add Games? (Ok, seriously, no. Can I get a “hell no“?
  3. Why not add things not even Facebook has, from Skype-like Voice, to Inbox Priorities?
  4. Why not add Slide’s Top Friends and ability to Throw SheepAardvark’s Ask-Any-Questions, and Angstro’s news alerts? Heck, let’s dream big: Google says they’re continuing to look at “ways to continue and extend Wave technology in other Google products“, and where better to reincarnate Pulp Fiction than in our Gmail. Anyone? Bueller?
  5. Why stop with copies and acquisitions? Embrace and extend, too! Xanadu’s pleasure dome will have us consuming our Facebook and Twitter statuses, comments, likes, links, photos, and videos through that Gmail Inbox intravenous drip…

Here’s the question I dread asking out loud: does Google think Gmail is The One Ring To Bind Them All? Are we supposed to trust Google with all of these deeply personal communications when we can’t even count on Gmail not to hard bounce a human who hits his storage quota? Maybe PERMANENT FAILURE was a declaration, not an error.

Meanwhile, if Facebook lets Gmail inventor Paul Buchheit turnFacebook Inbox into a product as fast, simple, and useful as the original Gmail, I could see myself migrating to Facebook as my primary email. But as of Labor Day 2010 approaches, that’s just a fantasy, and my fate is still tied to Gmail’s destiny.