Organ Donation, Power Consumption, and Social Choice: Social Media and Behavioural Economics | MIT Center for Civic Media
At the Center for Civic Media and the Berkman Center for Internet and Society, Nathan designs and researches civic technologies for cooperation across diversity. At the Berkman Center, he applies data analysis and design to the topics of peer-based social technologies, civic engagement, journalism, gender diversity, and creative learning.
Nathan's current projects include large scale research on community building online. In the summer of 2015, Nathan will be a PhD intern at the Microsoft Research Social Media Collective. A full project list is at natematias.com.
Nathan regularly liveblogs talks and events. He also publishes data journalism with the Guardian Datablog and PBS IdeaLab. He coordinated the Media Lab Festival of Learning in 2012 and 2013.
Before MIT, Nathan completed an MA in English literature at the University of Cambridge, where he was a Davies Jackson scholar. In earlier years, he was Riddick Scholar and Hugh Cannon Memorial Scholar at the American Institute of Parliamentarians. He won the Ted Nelson award at ACM Hypertext 2005 with a work of tangible scholarly hypermedia. He facilitated #1book140, The Atlantic's Twitter book club from 2012-2014, and was an intern at Microsoft Research Fuse Labs in the summer of 2013.
Organ Donation, Power Consumption, and Social Choice: Social Media and Behavioural Economics
This is the second session on Social Media and Behavioral Economics. Our moderator is David Laibson, Professor of Economics at Harvard University.
- Session One: Data, Experiments, and Social Networks
- Session Two: Organ Donation, power Consumption, and Social Choice
- Session Three: Malleability vs Serendipity in Social Choice
Sendhil Mullainatha starts out by talking about technologies that assume that other people can help us by giving advice. He talks about "the surprising power of neighborly advice," a study by Gilbert, Kllingsworth et al. The study asked people to predict, in advance, how much they might like a date. Even though people were more accurate at predicting the outcome of a date when they received advice from friends, people tended to trust written profiles more than people.
Why do we trust profiles more than each other? Sendhil asks us to judge how good a driver we are, from one to ten, relative to the other people in this room. How many of us put ourselves in the bottom decile? Only one person. People tend to be overconfident about our own judgment. As a result, we tend to undervalue the opinions of others.
Sendhil next talks to us about a "uniqueness bias." To illustrate this, he talks to us about a study about people's attitudes toward policies. Basically, if you're a liberal, you like what people tell you Democrats favour. The details of the policy don't matter. When you ask people what influences them, they say "the details of the proposal" or "my philosophy of government," even if that's not what the data shows.
Others can help, but we tend to undervalue it. Although sites like TripAdvisor are popular, there probably still isn't enough demand.
Sendhil next asks us, how can we outsource self help? He shows us Clocky, the alarm clock on wheels which runs away. This is a case of outsourcing self help to devices. Of course, we outsource self help to people as well. Many of us hire trainers. Just going to a buddy can increase our likelihood to go to the gym. Making a public commitment to lose weight is also another way to outsource self help. In all of these cases, we're on the verge of using technology to implement these things and increase the scale of outsourced self help. For example, it's easy to create an app which helps you find gym partners.
Sendhil shows us a pill bottle. For the vast majority of diseases, we are left with one problem. Although we have had huge drug advances, the drugs aren't working because people aren't taking them. How do we solve this problem? Sendhil shows us GlowCap, a passive-agressive needy friend. If it isn't opened the right number of times a day, it starts beeping and glowing. After time, it sends you a text message, irritating you until you take the pill bottle.
Perhaps we could use social network data to further improve adherence, helping people remind and persuade each other.
Next he talks about the idea of descriptive norms. When people are told that 75% of people in a room have reused their towels, they tend to reuse their towels. Instead of telling people that they ought to do something, we should give people messages saying that lots of people do this. Usually when we try to convince people, we often talk about the magnitude of the problem. Experiments have shown that ads emphasizing that no one is voting are much less effective than ads which set a positive norm (see Gerber and Rogers's paper on Voting).
Next up is Michael Sachse, Vice President of Regulatory Affairs and General Counsel, Opower. How can we use energy data to promote the behaviour that we want people to take? He asks us how many checked our email or looked at our bank statement or energy usage? Very few of us look at our energy usage.
On average, people spend 6 minutes a year thinking about energy data. That's not going to get you very far with productive thought. "I spend more time each year trying to match my socks than the average American spends thinking about energy usage." Why? Energy data is in the magic quadrant of boring and confusing. Worse, people don't know what a Kilowatt-hour is, and even if they did, they wouldn't know how to make meaningful changes. Should we turn the lights off or turn off the temperature? Sachse shows us an energy bill, "a disaster of line items that are a reflection of a very complex regulatory system."
Why is this important? There's a dirty secret in clean energy. Everything they do relies on customers paying. Renewables want us to pay more because it's clean. Smart Grid wants us to "pay for our stuff, it will help with a lot of other stuff." Utilities say, "we are so glad that you are paying for some more stuff." This approach hasn't worked very well. He tells us a story about the introduction of smart meters in California in 2009. Meters coincided with an especially hot summer, and people blamed the meters for their unusually high energy bills.
How does Opower fix this? (see an OPower slide deck here) They contextualise information in a way that is motivating for them. Firstly, they give people a sense of what their peers do. Secondly, people don't want data. Engineers often believe that people will respond to better quality data. People want insight. They also want something at the sweet spot between suggestion and direction.
It's not obvious how to connect people within the same utility service region-- that's not normally how we choose friends. So OPower created an app that helps customers connect with each other.
How effective is all of this social data? Opower has been saving around 2% across the US. In a year, they can save an entire terawatt-hour, enough to power a midsize American city for an entire year. What shape does this change take? It's mostly large numbers of people taking small actions. They're saving the insights from Opower, printing them and putting on their fridge, and showing them to their friends. This is just the beginning.
Sarah Feinberg, the director of corporate communications at Facebook, who led their organ donation program. She was formerly chief of staff to Rahm Emanuel when Rahm was chief of staff to Obama.
At Facebook, they think about the organ crisis as a social crisis. Doctors can do the work, the infrastructure exists. The only reason people die is that there aren't enough organs. People like organ donation, but not enough people become organ donors. Why is that? Typically, people make the decision to be an organ donor, it's at the department of motor vehicles. It's not a moment when they're not thinking about organ donation. Secondly, people don't usually think about organ donation unless a friend of family member has received a donation. It's not a pleasant thing to consider or talk about. In principle, if people talk about it more, and if it's more comfortable, then people will be more likely to become organ donors.
If you register to become an organ donor and share it on Facebook, it creates a NewsFeed item that's shared with your friends. That feed item might be shared with a subset of your friends. Sarah is excited about the traffic this received.
Next up is Andrew Cameron, Doctor of Liver Transplants at Johns Hopkins University, talks about the behavioral economics of the organ donation project. There is a crisis in US organ donation. Well over 100,000 people are waiting for an organ transplant. The number of people needing transplants has increased by a factor of six since 1998, while the number of donations has hardly changed at all. "I need you to focus on this because 117,000 people need your help."
Why aren't there enough organs to go around? Let's talk about deceased donation, Andrew says. In the US, 2.5 million people die each year. You have to die in a very certain way to donate your organs. You have to die in the hospital. That yields around 10-15k donors per year. Half of those people are lost due to failure to obtain consent.
Over 100 million people have registered, around 40% of the US population. That looks good, until we realise that 95% of the country supports donation. Why don't people do the right thing?
Compared to public health campaigns, the DMV registration campaign has been very effective. But for those who say no, why would they say no? People are concerned that if they sign up to organ donation programs, they won't be properly treated in a hospital that wants their organs.
Andrew next offers public thanks to Sheryl Sandberg, Sarah Feinberg, and Mark Zuckerberg, who came to him and convinced him to participate.
What was the response to the Facebook campaign? Over the first week, there was a tenfold increase of organ donation. Some states, like Georgia, had 5x increases in organ donations over several weeks. However, within two weeks, many states were back to baseline levels of organ donation.
Andrew compares the effectiveness of the DMV to Facebook. The latter offers an immediate opportunity as part of a discussion amongst friends. Every time someone makes the choice, it re-initiates the conversation within that friendship group. The initiative treated people as people, pointed out social norms, and also offered people a bump in their personal reputation.
"We didn't cure the organ donation process," Andrew points out. Firstly, the project didn't utilise defaults. Facebook wasn't in a position to opt-in every Facebook user to organ donation. They also didn't mandate that all users make a choice. Andrew looks at Sarah and says how many people's lives he thinks could have been saved. Next, the study didn't use incentives. It might be a good idea to tie organ donation to a dollar off Chipotle.
What's next? Facebook will be rolling this out worldwide. Next, they plan to make the system work on mobile phones. They also want the DMV to make it possible for people who register for organ donation to share with their friends on Facebook.
18 people will die today from lack of transplantable organs, Andrew tells us. Facebook has taken an important step, and we need to do more to solve this.
David Laibson, our moderator, starts out by "throwing some cold water" on these claims. Sendhil, the economist, argues that we often don't listen to peer advice, and that we need to pay more attention to it. One alternative is that peer advice is often toxic. It influences, pulls, and drives us in ways that hurt us and are socially damaging. Laibson talks to us about a study in 1955 by Solomon Asch showing ways that our tendency to conform to peers can lead us to a wrong conclusion.
Next, Laibson responds to Michael's talk about OPower. Laibson recently conducted a study on retirement savings that showed that employees were less likely to sign up for a savings plan if they found out that large numbers of their peers were signed up (pdf). Peer influence can have positive effects, but we don't yet know what leads to those positive peer effects, and what leads to more perverse effects.
Laibson points out that the Facebook organ donation program didn't have long-term effects. Three months later, organ donation registrations were back to the baseline. Might the bump had come from publicity around this study? The Facebook organ donation wasn't able to create a national conversation that could be measured in terms of Facebook sign-ups. Laibson especially liked Andrew's suggestion of mandated options.
Max, from the Harvard Business School, refers to the Johnson/Goldstein study on defaults. Why aren't we able to create change in government to create better defaults? Sarah Feinberg points out that many people believe it's a personal choice. It would be hard to find members of Congress who wouldn't see it that way. Some European countries offer a baseline opt-out. Facebook has been talking with countries with opt-out programs, and those countries don't want Facebook to be involved. If Facebook runs the program, these governments say, then people might know that they can opt out and decide to do so.
Andrew responds that the best country at organ donation is Spain, which has an opt-out program. After talking to Spanish surgeons, he points out the benefits of a public health education program. After going from hospital to hospital around the country, the entire country was ready to set up a default program. And in Spain, with the highest organ donation rate, still checks with families before organs are donated. Andrew argues that doctors will never take out organs if family members object. If they did that, it would be an irrecoverable fracture to the public trust, something which no one would ever risk.
Michael Sachse, from Opower, points out the challenge that people often won't answer surveys honestly. Maybe fewer than 95% of Americans want to be organ donors, but they are uncomfortable saying that they're against it. Is this really an issue of helping people achieve what they already want to do? Andrew backtracks a bit, citing surveys which state that only 70% of Americans, not 95% are willing to donate their organs.
Sendhil is more optimistic about a policy solution. A federal default may not happen, but the idea of active choice is much more feasible-- requiring everyone to make a choice of some kind. The main barrier to implement this is just money-- it takes lobbying money. State level lobbying isn't very expensive; just putting forward model bills.
The laws in many states are completely archaic, says Sarah. New York is the worst state. They require people to print, sign, and notarize a form to become an organ donor. Governor Cuomo is interested in changing that, but how high is it on his priority list? It's also incredibly difficult to get data from individual states.