May We Have Your Attention, Please?

With the workplace ever more full of distractions, researchers are developing tools to keep us on task
Nigel Buchanan

It's official: The average knowledge worker has the attention span of a sparrow. Roughly once every three minutes, typical cubicle dwellers set aside whatever they're doing and start something else—anything else. It could be answering the phone, checking e-mail, responding to an instant message, clicking over to YouTube (GOOG), or posting something amusing on Facebook. Constant interruptions are the Achilles' heel of the information economy in the U.S. These distractions consume as much as 28% of the average U.S. worker's day, including recovery time, and sap productivity to the tune of $650 billion a year, according to Basex, a business research company in New York City.

Soon, however, the same kinds of social networking software and communications technologies that make it deliciously easy to lose concentration may start steering us back to the tasks at hand. Scientists at U.S. research labs are developing tools to help people prioritize the flood of information they face and fend off irrelevant info-bytes. New modes of e-mail and phone messaging can wait patiently for an opportune time to interrupt. One program allows senders to "whisper" something urgent via a pop-up on a screen.

Innovations like these belong to a sub-branch of computer science that's geekily called "attentional user interfaces." The goal, says Scott E. Hudson, a professor in this discipline at Carnegie Mellon University, is finding a way to reap benefits from the data deluge "without having it destroy us on the attention side."

Ours is hardly the first generation to fret about distraction. Humans are essentially interruption-driven because they must be alert to change, says Gary Marcus, author of Kluge: The Haphazard Construction of the Human Mind. "We're not built to stay on task," he contends. But people in past eras never had to cope with so much beeping, blinking, pinging, and ringing. Gloria Mark, a professor of informatics at the University of California at Irvine, monitored thousands of hours of workplace behavior. Her studies documented that most workers switch gears every few minutes, and once they're distracted, it can take nearly half an hour to get back on track. "When you see the hard numbers, it kind of hits home how bad it really is," says Mark.


One solution might be to construct a work world supported by all-seeing, ultra-organized digital assistants. Eric Horvitz, a principal researcher at Microsoft (MSFT), has spent more than a decade creating artificial-intelligence systems that observe humans at work. These software programs, which reside on computers and various handheld gadgets, watch and listen to the user, tracking digital calendars and noting key contacts. And they apply mathematical formulas known as Bayesian probability models to predict the cost and benefit of interrupting someone at work. Having served as a guinea pig, Horvitz considers his latest prototype a trusted "presence." Recently, he received an urgent e-mail from a new intern. His e-mail triage program, called Priorities, ranked the message 100—a perfect score on a timeliness scale of 1-100. That afternoon, an announcement of "free food" down the hall ranked an ignorable 6.

This trusted presence isn't ready for prime time. But Priorities inspired Microsoft to create Outlook Mobile Manager, a product that enables Outlook to recognize urgent e-mails and to do "presence forecasting." That means users of OMM 2.0 can essentially let the software decide whether e-mails should be routed to their computers, phones, or some other device. Future versions of Windows will likely include another feature born in Horvitz's lab. Called Bounded Deferral, the feature holds messages in reserve until the recipient is ready for a "cognitive break." The ideal, says Horvitz, is "software that takes the fragility of human attention into deep consideration."

In contrast to Horvitz's sweeping, soup-to-nuts attention management strategy, IBM (IBM) favors a human-centric à la carte approach. One prototype it's testing is an instant-messaging answering machine known as IMSavvy that allows messages to tap gently at your consciousness. Invented by Gary Hsieh, a graduate student of Scott Hudson at Carnegie Mellon who served as an intern at IBM, the program can sense when you are away or busy by your typing and mouse patterns. It protects your focus by telling would-be interrupters you are not available. But it also suggests ways they can get through to you. Future versions may gauge "interruptibility" by using audio sensors, too. "It's just what your mom said: Don't interrupt when someone is talking,'" Hudson says.

But what about the dicier judgment calls? Picture the moment when the phone has fallen silent, your in-boxes are closed, and you're lost in a creative thought. Even the smartest digital assistant is likely to conclude it's safe to interrupt, but this is dangerous territory. "If you handle the exception cases wrong, your users will stop using your tool," says IBM team leader Jennifer Lai.

She found a solution in a time-honored social convention. If you pop into a co-worker's office when he's on the phone, he may try to wave you away but will listen if you whisper some important news. IMSavvy offers you a "whisper" option, with text that flickers on the recipient's screen, even if he has instructed the system to withhold his messages. "Instead of trying to predict if an interruption is good or bad, we want to give people lightweight tools to help them do the right thing," says Lai.

In her new book, Distracted (Prometheus Books), Maggie Jackson explores the latest discoveries about attention and how we concentrate. It details the rising costs of living in a split-focus world.

    Before it's here, it's on the Bloomberg Terminal.