Harnessing the Power of Feedback Loops
By Thomas Goetz June 19, 2011 | 9:45 am | Wired July 2011
The premise of a feedback loop is simple: Provide people with information about their actions in real time, then give them a chance to change those actions, pushing them toward better behaviors.
Photo: Kevin Van Aelst
In 2003, officials in Garden Grove, California, a community of 170,000 people wedged amid the suburban sprawl of Orange County, set out to confront a problem that afflicts most every town in America: drivers speeding through school zones.
Local authorities had tried many tactics to get people to slow down. They replaced old speed limit signs with bright new ones to remind drivers of the 25-mile-an-hour limit during school hours. Police began ticketing speeding motorists during drop-off and pickup times. But these efforts had only limited success, and speeding cars continued to hit bicyclists and pedestrians in the school zones with depressing regularity.
The Feedback Loop
by Thomas Goetz (52.5 MB .mp3)
Subscribe: Wired Features PodcastSo city engineers decided to take another approach. In five Garden Grove school zones, they put up what are known as dynamic speed displays, or driver feedback signs: a speed limit posting coupled with a radar sensor attached to a huge digital readout announcing “Your Speed.”
The signs were curious in a few ways. For one thing, they didn’t tell drivers anything they didn’t already know—there is, after all, a speedometer in every car. If a motorist wanted to know their speed, a glance at the dashboard would do it. For another thing, the signs used radar, which decades earlier had appeared on American roads as a talisman technology, reserved for police officers only. Now Garden Grove had scattered radar sensors along the side of the road like traffic cones. And the Your Speed signs came with no punitive follow-up—no police officer standing by ready to write a ticket. This defied decades of law-enforcement dogma, which held that most people obey speed limits only if they face some clear negative consequence for exceeding them.
In other words, officials in Garden Grove were betting that giving speeders redundant information with no consequence would somehow compel them to do something few of us are inclined to do: slow down.
The results fascinated and delighted the city officials. In the vicinity of the schools where the dynamic displays were installed, drivers slowed an average of 14 percent. Not only that, at three schools the average speed dipped below the posted speed limit. Since this experiment, Garden Grove has installed 10 more driver feedback signs. “Frankly, it’s hard to get people to slow down,” says Dan Candelaria, Garden Grove’s traffic engineer. “But these encourage people to do the right thing.”
In the years since the Garden Grove project began, radar technology has dropped steadily in price and Your Speed signs have proliferated on American roadways. Yet despite their ubiquity, the signs haven’t faded into the landscape like so many other motorist warnings. Instead, they’ve proven to be consistently effective at getting drivers to slow down—reducing speeds, on average, by about 10 percent, an effect that lasts for several miles down the road. Indeed, traffic engineers and safety experts consider them to be more effective at changing driving habits than a cop with a radar gun. Despite their redundancy, despite their lack of repercussions, the signs have accomplished what seemed impossible: They get us to let up on the gas.
The signs leverage what’s called a feedback loop, a profoundly effective tool for changing behavior. The basic premise is simple. Provide people with information about their actions in real time (or something close to it), then give them an opportunity to change those actions, pushing them toward better behaviors. Action, information, reaction. It’s the operating principle behind a home thermostat, which fires the furnace to maintain a specific temperature, or the consumption display in a Toyota Prius, which tends to turn drivers into so-called hypermilers trying to wring every last mile from the gas tank. But the simplicity of feedback loops is deceptive. They are in fact powerful tools that can help people change bad behavior patterns, even those that seem intractable. Just as important, they can be used to encourage good habits, turning progress itself into a reward. In other words, feedback loops change human behavior. And thanks to an explosion of new technology, the opportunity to put them into action in nearly every part of our lives is quickly becoming a reality.
A feedback loop involves four distinct stages. First comes the data: A behavior must be measured, captured, and stored. This is the evidence stage. Second, the information must be relayed to the individual, not in the raw-data form in which it was captured but in a context that makes it emotionally resonant. This is the relevance stage. But even compelling information is useless if we don’t know what to make of it, so we need a third stage: consequence. The information must illuminate one or more paths ahead. And finally, the fourth stage: action. There must be a clear moment when the individual can recalibrate a behavior, make a choice, and act. Then that action is measured, and the feedback loop can run once more, every action stimulating new behaviors that inch us closer to our goals.
This basic framework has been shaped and refined by thinkers and researchers for ages. In the 18th century, engineers developed regulators and governors to modulate steam engines and other mechanical systems, an early application of feedback loops that later became codified into control theory, the engineering discipline behind everything from aerospace to robotics. The mathematician Norbert Wiener expanded on this work in the 1940s, devising the field of cybernetics, which analyzed how feedback loops operate in machinery and electronics and explored how those principles might be broadened to human systems.
Over the past 40 years, feedback loops have been thoroughly researched and validated in psychology, epidemiology, military strategy, environmental studies, engineering, and economics.
Illustration: Ulla Puggaard
The potential of the feedback loop to affect behavior was explored in the 1960s, most notably in the work of Albert Bandura, a Stanford University psychologist and pioneer in the study of behavior change and motivation. Drawing on several education experiments involving children, Bandura observed that giving individuals a clear goal and a means to evaluate their progress toward that goal greatly increased the likelihood that they would achieve it. He later expanded this notion into the concept of self-efficacy, which holds that the more we believe we can meet a goal, the more likely we will do so. In the 40 years since Bandura’s early work, feedback loops have been thoroughly researched and validated in psychology, epidemiology, military strategy, environmental studies, engineering, and economics. (In typical academic fashion, each discipline tends to reinvent the methodology and rephrase the terminology, but the basic framework remains the same.) Feedback loops are a common tool in athletic training plans, executive coaching strategies, and a multitude of other self-improvement programs (though some are more true to the science than others).
Despite the volume of research and a proven capacity to affect human behavior, we don’t often use feedback loops in everyday life. Blame this on two factors: Until now, the necessary catalyst—personalized data—has been an expensive commodity. Health spas, athletic training centers, and self-improvement workshops all traffic in fastidiously culled data at premium rates. Outside of those rare realms, the cornerstone information has been just too expensive to come by. As a technologist might put it, personalized data hasn’t really scaled.
Second, collecting data on the cheap is cumbersome. Although the basic idea of self-tracking has been available to anyone willing to put in the effort, few people stick with the routine of toting around a notebook, writing down every Hostess cupcake they consume or every flight of stairs they climb. It’s just too much bother. The technologist would say that capturing that data involves too much friction. As a result, feedback loops are niche tools, for the most part, rewarding for those with the money, willpower, or geeky inclination to obsessively track their own behavior, but impractical for the rest of us.
Illustration: Leo Jung
That’s quickly changing because of one essential technology: sensors. Adding sensors to the feedback equation helps solve problems of friction and scale. They automate the capture of behavioral data, digitizing it so it can be readily crunched and transformed as necessary. And they allow passive measurement, eliminating the need for tedious active monitoring.
In the past two or three years, the plunging price of sensors has begun to foster a feedback-loop revolution. Just as Your Speed signs have been adopted worldwide because the cost of radar technology keeps dropping, other feedback loops are popping up everywhere because sensors keep getting cheaper and better at monitoring behavior and capturing data in all sorts of environments. These new, less expensive devices include accelerometers (which measure motion), GPS sensors (which track location), and inductance sensors (which measure electric current). Accelerometers have dropped to less than $1 each—down from as much as $20 a decade ago—which means they can now be built into tennis shoes, MP3 players, and even toothbrushes. Radio-frequency ID chips are being added to prescription pill bottles, student ID cards, and casino chips. And inductance sensors that were once deployed only in heavy industry are now cheap and tiny enough to be connected to residential breaker boxes, letting consumers track their home’s entire energy diet.
Of course, technology has been tracking what people do for years. Call-center agents have been monitored closely since the 1990s, and the nation’s tractor-trailer fleets have long been equipped with GPS and other location sensors—not just to allow drivers to follow their routes but so that companies can track their cargo and the drivers. But those are top-down, Big Brother techniques. The true power of feedback loops is not to control people but to give them control. It’s like the difference between a speed trap and a speed feedback sign—one is a game of gotcha, the other is a gentle reminder of the rules of the road. The ideal feedback loop gives us an emotional connection to a rational goal.
And today, their promise couldn’t be greater. The intransigence of human behavior has emerged as the root of most of the world’s biggest challenges. Witness the rise in obesity, the persistence of smoking, the soaring number of people who have one or more chronic diseases. Consider our problems with carbon emissions, where managing personal energy consumption could be the difference between a climate under control and one beyond help. And feedback loops aren’t just about solving problems. They could create opportunities. Feedback loops can improve how companies motivate and empower their employees, allowing workers to monitor their own productivity and set their own schedules. They could lead to lower consumption of precious resources and more productive use of what we do consume. They could allow people to set and achieve better-defined, more ambitious goals and curb destructive behaviors, replacing them with positive actions. Used in organizations or communities, they can help groups work together to take on more daunting challenges. In short, the feedback loop is an age-old strategy revitalized by state-of-the-art technology. As such, it is perhaps the most promising tool for behavioral change to have come along in decades.
How a Feedback Loop Works
A modified traffic sign can have a profound effect on drivers’ behavior. Here’s what happens.
The radar-equipped sign flashes a car’s current speed.
First comes the data—quantifying a behavior and presenting that data back to the individual so they know where they stand. After all, you can’t change what you don’t measure. 2 Relevance
The sign also displays the legal speed limit—most people don’t want to be seen as bad drivers.
Data is just digits unless it hits home. Through information design, social context, or some other proxy for meaning, the right incentive will transform rational information into an emotional imperative. 3 Consequences
People are reminded of the downside of speeding, including traffic tickets and the risk of accidents.
Even compelling information is useless unless it ties into some larger goal or purpose. People must have a sense of what to do with the information and any opportunities they will have to act on it. 4 Action
Drivers slow an average of 10 percent—usually for several miles.
The individual has to engage with all of the above and act—thus closing the loop and allowing that new action to be measured. In 2006, Shwetak Patel, then a graduate student in computer science at Georgia Tech, was working on a problem: How could technology help provide remote care for the elderly? The obvious approach would be to install cameras and motion detectors throughout a home, so that observers could see when somebody fell or became sick. Patel found those methods unsophisticated and impractical. “Installing cameras or motion sensors everywhere is unreasonably expensive,” he says. “It might work in theory, but it just won’t happen in practice. So I wondered what would give us the same information and be reasonably priced and easy to deploy. I found those really interesting constraints.”
The answer, Patel realized, is that every home emits something called voltage noise. Think of it as a steady hum in the electrical wires that varies depending on what systems are drawing power. If there were some way to disaggregate this noise, it might be possible to deliver much the same information as cameras and motion sensors. Lights going on and off, for instance, would mean that someone had moved from room to room. If a blender were left on, that might signal that someone had fallen—or had forgotten about the blender, perhaps indicating dementia. If we could hear electricity usage, Patel thought, we could know what was happening inside the house.
A nifty idea, but how to make it happen? The problem wasn’t measuring the voltage noise; that’s easily tracked with a few sensors. The challenge was translating the cacophony of electromagnetic interference into the symphony of signals given off by specific appliances and devices and lights. Finding that pattern amid the noise became the focus of Patel’s PhD work, and in a few years he had both his degree and his answer: a stack of algorithms that could discern a blender from a light switch from a television set and so on. All this data could be captured not by sensors in every electrical outlet throughout the house but through a single device plugged into a single outlet.
This, Patel soon realized, went way beyond elder care. His approach could inform ordinary consumers, in real time, about where the energy they paid for every month was going. “We kind of stumbled across this stuff,” Patel says. “But we realized that, combined with data on the house’s overall draw on power”—which can be measured through a second sensor easily installed at the circuit box—”we were getting really great information about resource consumption in the home. And that could be more than interesting information. It could encourage behavior change.”
By 2008, Patel had started a new job in the computer science and engineering departments at the University of Washington, and his idea had been turned into the startup Zensi. At Washington, he focused on devising similar techniques to monitor home consumption of water and gas. The solutions were even more elegant, perhaps, than the one for monitoring electricity. A transducer affixed to an outdoor spigot can detect changes in water pressure that correspond to the resident’s water usage. That data can then be disaggregated to distinguish a leaky toilet from an over-indulgent bather. And a microphone sensor on a gas meter listens to changes in the regulator to determine how much gas is consumed.
Last year, consumer electronics company Belkin acquired Zensi and made energy conservation a centerpiece of its corporate strategy, with feedback loops as the guiding principle. Belkin has begun modestly, with a device called the Conserve Insight. It’s an outlet adapter that gives consumers a close read of the power used by one select appliance: Plug it into a wall socket and then plug an appliance or gadget into it and a small display shows how much energy the device is consuming, in both watts and dollars. It’s a window onto how energy is actually used, but it’s only a proof-of-concept prototype of the more ambitious product, based on Patel’s PhD work, that Belkin will begin beta-testing in Chicago later this year with an eye toward commercial release in 2013. The company calls it Zorro.
At first glance, the Zorro is just another so-called smart meter, not that different from the boxes that many power companies have been installing in consumers’ homes, with a vague promise that the meters will educate citizens and provide better data to the utility. To the surprise of the utility companies, though, these smart meters have been greeted with hostility in some communities. A small but vocal number of customers object to being monitored, while others worry that the radiation from RFID transmitters is unhealthy (though this has been measured at infinitesimal levels).
Politics aside, in pure feedback terms smart meters fail on at least two levels. For one, the information goes to the utility first, rather than directly to the consumer. For another, most smart meters aren’t very smart; they typically measure overall household consumption, not how much power is being consumed by which specific device or appliance. In other words, they are a broken feedback loop.
Belkin’s device avoids these pitfalls by giving the data directly to consumers and delivering it promptly and continuously. “Real-time feedback is key to conservation,” says Kevin Ashton, Zensi’s former CEO who took over Belkin’s Conserve division after the acquisition. “There’s a visceral impact when you see for yourself how much your toaster is costing you.”
The Zorro is just the first of several Belkin products that Ashton believes will put feedback loops into effect throughout the home. Ashton worked on RFID chips at MIT in the late 1990s and lays claim to coining the phrase “Internet of Things,” meaning a world of interconnected, sensor-laden devices and objects. He predicts that home sensors will one day inform choices in all aspects of our lives. “We’re consuming so many things without thinking about them—energy, plastic, paper, calories. I can envision a ubiquitous sensor network, a platform for real-time feedback that will enhance the comfort, security, and control of our lives.”
As a starting point for a consumer products company, that’s not half bad.
Hide full comment.