For the past ten years, I’ve seen all sorts of civic tools come and go, and all of them were supposed to be the next great thing in civics. But if you really look at the metrics, like the number of users, number of return visitors, or time spent using the tool, and compare them to the metrics of products designed to help people do things like shop or hail a cab, it’s clear we haven’t even begun to tap the internet’s potential to change the way we make decisions and allocate resources within our communities.
Over time, I have come to believe that our inability to attract mass usage for our products is, in part, due to a willful ignorance of how users are motivated to civic action. To build the technologies that will transform government and governance, we need to do a better job of designing and building tools for real world users.
My argument is based on two realities:
PEOPLE STILL NEED TO MAKE DINNER.
Or, in other words, we need to start treating our users as rational actors. The people we want to use our tools to influence politics and governance are very busy, and there are plenty of ways they can spend their time aside from on our platforms.
THE INTERNET ISN’T MAGIC.
We all wish users would suddenly become passionate about the intricate policy decisions that impact their lives, but they aren’t going to develop that interest just because we invented the internet.
As creators of campaigns and civic engagement tools, we need to be much more honest with ourselves, and each other, about the motivations and interests of “real people.” Just to be clear, no one reading this post count as real people for the purposes of civic engagement— if you are voluntarily reading a Medium post titled “The Three Levers of Civic Engagement,” you are squarely in the civics geek camp and not a typical user. In order to more effectively talk about user expectations and behavior, I’m going to lay out a new framework to use when talking about digital civic engagement and whether real people will (or won’t) interact with those tools.
Right after law school, I took a job as a community organizer. I was responsible for empowering constituents to find shared interests and to help them work together to impact the decisions being made about their communities. My day-to-day involved working with church leadership, holding house meetings, organizing one-on-ones and yes, I also did my share of door knocking and phone banking. It was as a community organizer that I first learned about leadership and power, and how to use action to achieve specific outcomes.
And it was very hard work. The people in my community cared about their neighborhood and the decisions being made by local representatives, but they also had two jobs. They had mothers with health problems. They had roofs with leaks. They needed to pick the kids up from soccer. They had very busy lives. Much of my time was spent convincing people that the work they could do on behalf of their community was worth the time they would sacrifice from other very important life tasks.
My time as an organizer has informed my work as a product manager. When designing tools to help people take action online, I never forget how hard it was for my friends in Milwaukee to find time to work on local issues.
The numbers back up my anecdotal evidence. A 2013 Pew study, Civic Engagement in the Digital Age, found that, in the previous 12 months, only 22% of respondent had attended a political meeting on local, town, or school affairs. 18% had contacted their elected representative about an issue that was important to them. 17% had signed a petition. In the California state primary last Tuesday,only 13.5% of California’s eligible voters voted.
GAME THEORY AND CIVIC ENGAGEMENT TECH
A mathematical equation buried in a 50-year-old academic paper might not seem the most logical place to start a conversation about human motivation and community action. But it’s what got me started.
Last year, I stumbled across a paper written in 1968 by William Riker and Peter Ordeshook. In it, William and Peter (I think of them as first-name-basis friends now) apply game theory and mathematics to elections to explain voting behavior. The paper “The Calculus of Voting” has had a profound impact on my thinking about civic technology.
For this talk, I’m going to adapt this equation to capture the relative importance of various factors influencing a citizen to take a civic action. So basically, I’m going to use this to explain why people will or won’t use your awesome new civic engagement tool.
P is the perceived probability that an action will change the outcome of a civic decision. Or, for example, the likelihood that writing a letter to your Congressional representative will change the outcome of a floor vote.
B is the benefit the citizen as an individual will receive if the civic decision swings in their favor. As an example, if I have a pre-existing condition, the passage of the Affordable Healthcare Act would be a very real and measurable benefit to me. Or, if I had a lot of money invested in the stock market, a change to the tax code that cut capital gains tax would increase my net worth.
The good professors William Riker and Peter Ordeshook referred to D as Civic Duty. But I think it’s something more than that. D represents feelings of goodwill, and the sense of being part of something bigger than yourself. D is the pleasure people get from wearing an “I Voted” sticker and the pride they feel by being good citizen. It is the sum total of enjoyment a person gets from being part of a communities’ civic life.
C is the time, effort and financial cost of taking a civic action.
In the 2012 general election, 129 million people completed a ballot. The likelihood that your vote was going to decide the outcome of that election was so close to zero, it makes no difference. Essentially P = 0. Zero times times B equals zero. So the equation becomes 0 + D > C. As applied to the 2012 election, people didn’t vote because they thought they were going to be the deciding vote between Governor Romney and President Obama. They voted because their D, or their sense of civic duty, outweighed the C, or the cost of finding a polling place, making sure they were registered to vote, making sure they had the right ID and taking the time to go and vote.
However, at the local level, your chance of influencing a decision creeps upward. Let’s say your local city council is considering funding a dog park across the street from your house, and you own a dog. Going to a city council meeting and making a passionate argument might sway a vote. The P value is high. However, the cost of this particular civic action is also pretty high. You need to figure out that the city council is thinking about funding a dog park, identify when the city council is meeting, take time to travel to the meeting, and then (eek) public speaking! In this case, the higher P value along with the benefit to you as an individual dog owner might help you overcome the inhibitions presented by the cost of the civic action.
One of the mistakes we make as a community interested in building civic engagement tools is that we assume the old world rules no longer apply because internet = magic. But the internet isn’t magic.
Users have and will continue to act in a mostly rational way online, just as they did offline. Our users are people with lots of demands on their time, and the likelihood that their civic action will change the outcome of a civic decision is close to zero. It’s going to be hard to get them to pay attention to us.
And so, as technologists who are building the next generation of civic tools, we are left with these three levers to influence user behavior. All of our work falls into these three categories — probability, cost and duty.
LOWER THE COST
The cost lever is about ensuring we connect users with the information they need to be involved.
Many of us build tools that reduce the cost of civic engagement. As a company whose mission is all about organizing information and making it accessible, this is Google’s sweet spot. Here in the US, in partnership with the Voting Information Project, my team makes information about elections available to users. We want users to be able to find where to vote, who they can vote for, and how they can vote. And because we know not all people use Google, we make this information to other developers through the Civic Information API so they can surface this information on their platforms and lower the cost of voting for their users.
We also provide this kind of information to voters abroad through election portals. For example, we built a hub for the Indian election earlier this year which featured, among other things, the number of felonies with which each candidate was charged. This is a high demand piece of information that Indian voters find useful.
INCREASE THE DUTY
The duty lever is all about increasing a user’s sense of civic duty and it’s the kind of thing campaigns do very well. The Obama 2012 campaign was particularly good at using images and video to create these positive feelings of belonging.
This picture was tweeted by @BarackObama on election night 2012 with the caption “Four more years.” It was retweeted over 775,000 times, and held the record for the most retweeted tweet ever until March 2014. (See here for more on Obama director of social media Laura Olin and the most viral photo of the campaign.) In part, Facebook, Twitter and Tumblr are effective engagement tools because they make use of this lever. All of the social media platforms are used by campaigns to increase a user’s sense of civic duty and thus their likelihood of taking action.
We’ve also seen successful “social pressure” experiments like the one run by the Democratic National Committee which sent a letter to registered voters telling voters that “our records indicate that you voted in the 2008 election” and thanking them for their “good citizenship.” This mailer increased voter turnout by 2.5 percentage point.
INCREASE THE PROBABILITY
The probability lever, at first glance, looks like we can’t move it except through very local civic actions. Right now, at the federal level, the likelihood of a civic action, like calling congress or writing a letter to congress, changing the outcome of a vote is basically zero.
However, the internet allows us to measure online actions in a way that we never could with offline actions. Community organizers have always known that impact is made when groups of people take action together. If we can help our users find each other online and take action in small groups, then maybe we can start to shift “P”
Here’s another way to think about it. In campaigns, there are individuals called kingmakers or bundlers who each pledge to raise x many millions of dollars for a campaign. But when you look at the way fundraising is structured, it turns out that a donor who raised a million dollars didn’t actually bring one million unique dollars in the door. That donor went out and found 10 friends who pledged to raise 100,000. And those friends found 10 friends who pledged to raise 10,000— it’s all a giant pyramid game.
We can do the same thing with civic action. If a user can get other users to take impactful action in coordination with others, they scale their efforts and increase the likelihood that they have impact. If these actions take place online, we can bundle a user actions in the same way that we bundle donations to help shift the probability that their work will change the civic outcomes.
USERS ARE RATIONAL ACTORS
Too often, I’ve seen tools designed and built for the way that we wish users would behave. Yes, it would be nice if suddenly the whole world got passionately interested in the nuance of policy decision— but that’s not going to happen just because we invented the internet. Our users still have dinner to make.
To the funders, I challenge you to think critically about the consumer facing civic projects you’re funding. Are you funding products for rational users?
To the technologists in this space, we need to spend more time thinking about the ways we can reduce the cost of civic action and increase the probability that an action will impact an outcome. This means continued investment in standardizing and scaling acquisition of information for the data layer that supports our tools. We have the technology to create personalized, localized civic engagement tools. But we don’t have the organized data to power those products. To build a tool that lets me know when the city council is considering an issue about my neighborhood that will impact my life, I need an integrated, accurate, real-time dataset of laws, city council agendas, city council meeting minutes, city council votes, city council members and city council jurisdictions. It also means looking beyond the federal layer of government for potential civic tools. We have a better chance at truly transforming governance at the local level than we ever had at changing things at the federal level.
And to the academics, we need research that is focused on the right ways to move these three levers. What motivates our friends and neighbors to action? What is the information they need to participate in the civic process? What actions are the most impactful, and how do we demonstrate this to our users?
We’ve seen the internet transform all sorts of industries— from news and media to banking and transportation. But we haven’t yet seen this kind of disruption in our civic vertical. I’m certain the internet will fundamentally change the way we make decisions and allocate resources in our communities. However, if we want internet scale usage of the next generation of civic tools, we need to better understand real world people and how they will act in the civic space.
SOME END NOTES
The above piece is adapted from a talk I gave at the Personal Democracy Forum. Unfortunately, because of time constraints, I was not able to go into any of the complexities and potential weaknesses of this framework. In future posts, I hope to spend more time exploring what this could mean for civic tech product design. However, until then, I leave you with two additional important notes.
- In individual level studies, citizen perception of an election’s closeness have little effect on their decisions of whether or not to vote. (Ferejohn, J.A., & Fiorina, M.P. (1975). Closeness counts only in horseshoes and dancing. American Political Science Review) Some have interpreted this to mean that P is not an important factor in the calculation we make when deciding whether or not to vote. However, others have argued that this proves that we are all REALLY BAD AT MATH. ( Darmofal D. (2010) Reexamining the Calculus of Voting: A Social Cognitive Perspective on the Turnout Decision. Political Psychology) Or, in other words, that users consistently overestimate the impact our civic actions will have. In the talk, I did not necessarily clarify the difference between perceived probability and actual probability of impact.
- Although I have defined B in terms of pure self-interest, some of our users care about others and take actions that are not motivated by self-interest. An individual’s altruism, or how much she cares about others’ welfare and the benefits to all of society of one outcome over another may also play a factor in the calculus. (For more Fowler, J. (2006) Altruism and Turnout. The Journal of Politics.)
Tech President’s lovely Sam Roudman interviewed me shortly after I gave the talk. A transcript of our chat can be found here.