Thursday, June 06, 2013

Gaming for Sustainability: How Balaji Prabhakar’s Lottery Beats Traffic Congestion

Imagine someone offered you $1 to leave the house an hour earlier in the morning so as to shift your commute to an off-peak train. Will you do it? If you’re like most people, you will say no. Now imagine you’d be offered a 1-in-100 chance of winning $100. You might find this offer much more enticing. You may even end up gladly altering your commuting habit. This might seem rather obvious when you consider the behavioral-economics insight that the average person is risk-seeking when stakes are small. But how ingenious it is to apply this insight to reduce congestion-related costs (fuel, pollution, time) and, ultimately, congestion itself in urban areas with some of the world’s worst daily traffic jams!

INSINC poster at Singapore's train stations
I love bright ideas that in hindsight seem self-evident and make you wonder why no one thought of them sooner. Such a bright idea is Balaji Prabhakar’s Incentives for Singapore’s Commuters (INSINC) study that aims to reduce traffic congestion in the city-state of Singapore. Balaji, a Stanford Professor of Computer Science and Electrical Engineering (and well-known to my buddies at Cisco as a computer networking expert), is becoming a global traffic guru. I greatly enjoyed meeting him and was fascinated to learn about his work on topics ranging from the nature of traffic congestion to social structures to the psychology of incentives.

I was especially intrigued by the ways in which Balaji applies principles, architectures and algorithms from computer networks to fundamentally redesign Societal Networks. Societal Networks are networks that are vital for a society's functioning. They include transportation networks, waste management systems, energy networks and health-care systems. Balaji’s current major focus is on transportation networks, and he’s making strides in redesigning them to be more scalable and efficient.

In January 2012, Balaji launched INSINC jointly with the National University of Singapore and with support by Singapore’s Land Transport Authority (LTA). Using a cleverly playful scheme, the joint study experiments with reducing overall traffic congestion and with better managing crowdedness on Singapore’s rail system by incentivizing commuters to shift some of the peak travel demand to off-peak periods. INSINC allows commuters to earn credits proportional to the distance they travel on the rail system, with extra credits rewarded for off-peak journeys. But studies have shown that, while incentives can generate positive behavior, guaranteed small rewards don’t generate significant behavior changes and, accordingly, that guaranteed rebates are too low amounts to convince people to alter their commuting habits. Many commuters don’t even notice they’re paying less for their train rides. They will, however, be far more inclined to take “decongestion” trips during off-peak hours and, in the long run, change their commuting habits if they stand a chance to win larger prizes.

Balaji’s INSINC idea draws on this risk-seeking effect from behavioral economics. Participants can exchange their off-peak commute credits for chances to win up to $100 by playing a game on the INSINC website, and the prizes are then credited to their transportation card every month. Moreover, Balaji’s work shows that this risk-seeking effect is amplified in small networks: regularly hearing about other winners leads individuals to overestimate their own chances of success. The scheme’s lure therefore strengthens when participants go online to compare their results with those of friends and colleagues. This already worked particularly well in Bangalore, where Balaji ran a pilot project in collaboration with the Indian software company Infosys Technologies. That scheme, in which winners were advertised through Infosys, doubled the number of off-peak commuters to the company’s main research site, significantly reducing congestion on the Infosys peak-time buses (and I assume lessening the overall time spent by their employees on buses). INSINC aims to create a social network among commuters to produce a similar effect. The success of the Bangalore study has also led to a similar program that Balaji is running at Stanford University.

Visual of updated INSINC game board
In a recent press release, the LTA reports that INSINC is proving to be highly successful. Since its launch, more than 40,000 commuters have signed up for the program and more than $320,000 has been paid out to participants. The number of participants is growing and surged since last June, when the LTA announced that the study would be extended and enhanced. As of November 2012, more credits per off-peak train journeys are given, with the maximum cash prize doubling from $100 to $200. The refreshed INSINC website also has new game boards. The LTA says it hopes the study can eventually cover up to 60,000 commuters.

Businesses and policymakers are becoming increasingly aware that psychology plays a central role in the long journey towards a sustainable future, because a key aspect of sustainability is widespread behavior change. The beauty about Balaji’s random payoff scheme is that it generates behavior change in a way that’s both engaging and cost-effective. It not only maximizes small amounts of money, but also redesigns transportation networks to be “smarter” without the need to reengineer them, drawing on the insight that it takes many people to cause congestion, but a surprisingly small number to reduce it. As Balaji says,
“congestion is an 80-20 type of problem. In other words, if you get rid of 20 percent of the load, then the congestion measures will drop down potentially 80 percent. The question is: Which 20 percent is going to yield the road at the peak time? Somebody may value time more and somebody may value money more, and there’s a trade that’s set up.”
Blending technology, economics, and policy, Balaji hit upon a bright idea that establishes the right tradeoff to create the desired 80-20 effect. If it proves successful at city-scale, this approach may well be applicable to other large societal problems of the 21st century, such as energy consumption, water conservation, health and wellness.

No comments: