The Technologies of Political Communication and Control

Information and communication technology can, and already is transforming government. These new technologies have substantially changed the political calculus that is used to attract voters to a given political candidate. Relatedly, the policy platforms used to attract those voters, as well as the way citizens are mobilized into volunteers and voters for political campaigns have also evolved dramatically.

Specifically, the use of predictive modeling in political campaigns has shifted the attention toward the ‘persuadable voter’ at an individual level. Once those voters have been identified, information and communication technologies (ICTs) have reduced the costs of communication between candidates and the citizens they aim to mobilize into volunteers and voters. Simultaneously, ICTs increase the information asymmetry between the two groups. With the newfound ability to micro-target messages at the individual level, candidates are able to increasingly campaign on controversial wedge issues with less fear of public backlash. Together, these technological developments have negative consequences for citizens’ democratic participation, and result in increased polarization among the electorate.

To begin, I will review the existing literature on the transformative potential of ICTs in government to outline helpful theories and frameworks for understanding this rapidly changing environment. Special attention will be given to the work of Archon Fung et al, who put forward six models for the internet and politics, of which the “Social Monitoring” and “Constituent Mobilization” models work particularly well for understanding the changes occurring in political campaigning.

From there, I will provide evidence from the 2008, 2016, and 2020 presidential campaigns in the United States of America to demonstrate the transformative potential of these technologies over a short time period. My discussion will analyze the impact these changes are having on partisanship, civic participation, and trust in government, while simultaneously refuting common critiques of their transformative impact. My conclusion will offer several areas of concern that deserve close attention over the next decades as these technologies mature.

At this point, readers may be questioning whether dramatic changes in the nature of political campaigns actually constitute transformative change in government. On this point, I will borrow a line from the book Ground Wars by Rasmus Neilson, which has inspired much of my work on these topics: “How campaigns are waged matters, not only for electoral outcomes but also for what democratic politics is.[1] ” Who participates in our electoral process—and how they participate—has compounding effects on the type of representation and policies that politicians enact once in office, and ultimately who wins and losses in society.

Literature Review

Let’s begin with a few definitions. At their most basic level, ICTs refer to technologies that provide access to information through telecommunication. These can include both hardware and software like the internet, wireless networks, cell phones, and apps such as Facebook, Twitter, and the like.[2] The rapid development of these technologies has given rise to the recent phenomenon of E-government.

Stated simply again, e-government is the use of ICTs to improve government processes. A similar and more commonplace term is e-commerce: the use of ICTs to improve marketing, sales, and business operations. Taking this a step further, we will use e-government to refer to the “delivery of government information and services online through the internet or other digital means.”[3] For the purpose of this essay, we will also assume that political campaigns are encompassed within the definition of “government” for the reasons stated at the end of my introduction.

These terms emerged in the early to mid 1990s along with the World Wide Web, and scholars have debated their transformative potential on society ever since. The core of the debate is centered around whether these changes simply amount to doing the same things with different tools, or whether technology has fundamentally changed balances of power and outcome of policy processes.[4] This essay will directly contribute to this debate, as political campaigns are fundamentally about who controls power and the policies they choose to propose.

Those who argue for the transformative potential of these technologies point to their ability to dramatically reduce the costs of organizing and collective action, as well as their impact on communication. In their landmark book, Here Comes Everybody, Clay Shirky highlights just how much easier it has become to join a civic organization in the last twenty years. They do this by asking hypothetical questions to a pre-Google activist seeking to volunteer for an organization including: “How would you locate the organization? How would you contact it?”[5] When considered at scale, it becomes clear how ICTs have made civic participation far more accessible to citizens. Now consider the same effects on the costs of communication, which can be done easier, cheaper, and faster—as well as across further distances, asynchronously, and surreptitiously.[6]

Harvard academics Archon Fung, Hollie Gilman, and Jennifer Shkabatur have put forward a related framework in their article outlining six new models for understanding the internet and politics. Of the six, the two most relevant models for the transformation of political campaigns are the “Social Monitoring” and “Constituent Mobilization” models. The Social Monitoring model claims that civic organizations deploy digital tools to better understand public opinion and more quickly identify public issues requiring their attention. The Constituent Mobilization model purports that as digital technologies reduce the transaction costs of political messaging, it becomes easier for politicians to communicate to, and mobilize their followers to take civic action.[7]

What these authors have in common is that they argue these developments may lead to greater democratization of political participation. I argue the opposite. This essay focuses instead on how these technological developments can be weaponized to segment and divide the electorate, increase polarization, and reduce transparency. Political actors have always tried to influence citizens, win elections, and gain power—this is nothing new.[8] What is new, however, is the extraordinary power of digital communication tools to help politicians achieve their goals.

In the sections that follow, I will highlight how predictive modeling and computer algorithms have allowed campaigns to build psychological profiles of voters to segment the electorate into groups they deem “persuadable.” Next, I will explain how campaigns then use the microtargeting capabilities of digital communication tools to deliver tailored political messages to individual voters in an effort to influence their behavior. My argument will utilize evidence from other scholars conducting empirical studies on the impacts of these technologies, as well as investigative journalism reports on the use of these tools by campaigns. My discussion will touch on the transformative impact these developments have had on civil society—specifically the increase in polarization and development of echo chambers.

Predictive Modeling & Segmenting the Electorate?

To understand the transformative power of these technologies, we need to put them in the greater historical perspective of political communication in campaigns. To do this, it is helpful to trace the development of these practices while providing foils of how they were done historically to how they are done today. A starting assumption is that political campaigns have always had to maximize efficiency to accomplish their goals (win votes) with limited resources (money, labor, and time). To maximize efficiency, campaigns do not focus their resources on all voters, but rather segment the electorate into groups of voters they think they can persuade to vote for their campaign.

The process of segmentation can be understood as “the art and science of using the available information about the audience, which is to say the work product of all that rating and data-mining, to best advantage.”[9] Segmentation requires data on voter preferences and behavior, and prior to 2008 this data was produced by the field staff who canvass door-to-door for individual campaigns, as well as the vast amount of individual-level data amassed by the national Democratic Party and hosted on the digital national voter file called VoteBuilder”[10] This data included information about how respondents have voted in previous elections—whether they have volunteered or donated to a campaign, issues of importance to them, as well as personal details such as their home addresses, emails, and phone numbers.

Campaigns would then use this data to segment constituents in their districts into different groups based on their demographics and likelihood to vote. This often took the form of grouping people into demographic groups: veterans, women, co-ethnics, and low-income for example. This allowed candidates to direct their campaign’s message to these groups using their peers as the mode of communication. For example, during the final two-week get-out-the-vote (GOTV) phase of campaign field operations, volunteers who were war veterans would be sent to knock on the doors of other veterans; women would canvass women; and minority groups would focus on their own communities. This was an early form of political targeting (distinct from current microtargeting), that allowed campaigns to deliver personalized messages to their audience.

It was understood that political communications were more effective in mobilizing supporters when they were delivered in a message and mode familiar to the recipient. By leveraging the targeting capabilities provided by segmentation, candidates were able to promote the relevant aspects of their policy platform to communities of strategic importance to their success.

These practices have remained consistent in political campaigns since 2008, but what has changed dramatically is their ability to profile voters at an individual level. This is primarily a result of the exponential growth in available consumer data, and the rapidly decreasing costs of processing that data. No longer do campaigns rely on their own response data and VoteBuilder. Today the data ecosystem is a complicated web of layered data from a wide range of sources including consumer records and web activity compiled by firms like Experian, Acxiom, Google, and Facebook.[11] With vastly more data, and cheaper methods of analyzing it, political campaigns have transformed the process of voter segmentation by using predictive statistical modeling.

This transformation has become increasingly granular with each passing year. At present, the campaign industry is worth an annual $6 billion dollars with over 7000 full time professionals with backgrounds that now include data science and advanced statistics.[12] Voter segmentation is now almost exclusively outsourced to this industry of campaign consultants who construct predictive statistical models that profile voters at an eerily individual level. An infamous example of one of these consultancies is Cambridge Analytica, the data consultancy that worked with Donald Trump’s 2016 campaign to produce up to 5000 individual data points on every US voter in what amounted to “psychographic profiles.”[13] These tools allowed political campaigns to predict the behavioral responses to political messages more accurately than ever before, and marked the transition from basic targeting practices to microtargeting.

Political Microtargeting, Disinformation, & Polarization

Microtargeting is intimately tied to the rise of predictive modeling in political campaigns. Microtargeting “involves creating finely honed messages targeted at narrow categories of voters based on sophisticated combinatorial analysis of data” by applying “predictive modeling techniques to voter data.”[14] As previously mentioned, campaigns have always used targeting techniques to maximize resources and mobilize voters. Early iterations of this included fundraising phone calls to wealthy donors, and direct mail to supporters’ homes during GOTV. What has changed dramatically is the power of digital communication tools to manipulate voter behavior.

The goal of every political marketer is to deliver the appropriate message, to their ideal audience, at a moment that is most opportune to induce an action—whether that action is donating, volunteering, or voting. One of the most common ICT tools that campaigns use to do this today is Facebook. By combining their predictive modeling with the power of Facebook’s own artificial intelligence, campaigns aim to “maximize the emotional and psychological impact on their recipients” and tap into the “anxieties, concerns or vulnerabilities [that] a particular political message can trigger.”[15] The types of issues that trigger these anxieties are often some of the most controversial political debates. Imagine a scenario in which you have recently lost your job, and while researching your unemployment insurance options online you receive a targeted Facebook advertisement about the dangers of social welfare policies. This is the modern reality of weaponized information and communication.

The ability to micro target political communications has contributed to the increased willingness of candidates to deliver controversial messages on wedge issues that would otherwise be extremely divisive in more public settings, and a general trend toward “single issue politics that leads to increased partisanship among voters.”[16] Many argue that candidates have always exploited prejudice, bias, and anxieties in their messaging (think xenophobia related to immigration), and that these developments are not novel phenomena. However, “the ability to fine-tune communications to the individual level is novel,” and such subtle personalization makes it exponentially harder to identify, track, and guard against manipulation.[17] The result has been a massive increase in disinformation and “fake news.”

In this context, Fung et al’s Social Monitoring and Constituent Mobilization models become much more insidious. Rather than using these tools to identify citizen concerns and unite them through civic action, campaigns use data to carve out segments of the population that they believe they can manipulate with coercive messaging. Candidates are now able to maintain a relatively broad national policy platform, while microtargeting contradictory messages to the narrow interest groups they aim to persuade. Voters rarely encounter political information that challenges their existing beliefs, and increasingly engage with online material that reaffirms their long-held biases. These virtual spaces of like minded political opinion where extremism flourishes are known as “echo chambers,” and have been shown to consist of individuals with far higher levels of political partisanship.[18] ?

Indeed, empirical studies have shown that microtargeting increases political polarization far more than it increases voter turnout. Research by Mathieu Lavigne from my alma mater, McGill University, found that microtargeting causes voters to be 8% less likely to change their voting preferences, and is an “efficient means for political parties to secure their partisan support” by heightening political divisions.[19] There is significant evidence that Trump’s 2020 campaign repeated their 2016 playbook by using emotionally charged digital ads to retain the votes of rural, white, males—but also used these tools to suppress voter turnout among traditionally democratic voters.

Recent investigative journalism by the Miami Herald found that Trump’s digital team sought to reduce Black and Hispanic voter turnout in Miami-Dade County through a strategy they coined “deterrence.”[20] This strategy took two stages: first the campaign used predictive modeling to identify more than 116,000 Black and Hispanic voters that they deemed vulnerable to online manipulation.

Next, they used the micro targeting capabilities of the Facebook Ads platform to spread disinformation among these groups. Racially-charged, anti-Biden messages were shown to thousands in the Black community, while ads comparing democratic policies to Cuban and Venezuelan Socialism were targeted to Hispanics—many of whom in Miami fled from those countries. As a result, nearly 60,000 voters targeted with the “deterrence” strategy did not vote, which was a drop of six percentages points from the previous election.[21] Emerging evidence points to the use of similar strategies in multiple other swing-states.

The use of ICTs to spread disinformation has continued after the election. Trump repeatedly used his Twitter account to falsely challenge the legitimacy of the election results, eventually leading the company to ban him from using their platform. Although Twitter comments were not microtargeted, the disinformation about voter fraud still spread rapidly within existing echo chambers and were shown to reduce participation in the following Georgia Senate runoff election.[22]

Conclusion

The threats posed by the use of ICTs in political campaigns are dangerous and urgent. Some argue that their use has had life-and-death consequences, and place blame on their spread of disinformation as a major factor in the deadly January 6th attack on the US Capitol. Others argue that this just amounts to old political practices through new technologies and claim that there is a lack of empirical evidence to support the effectiveness of these tools. To those critics, I argue that the transformation has only just begun. These technologies have only emerged in the past 20 years and are increasing in their sophistication and effectiveness every month.

In an era of exponential technological development, the emergence of other new technologies will have a compounding effect on the use of predictive modeling and microtargeting. The Internet of Things (IoT) promises to dramatically increase the amount of available consumer data in the next century; blockchains are changing how data is stored, processed, and shared; and artificial intelligence and machine learning are improving the accuracy of the algorithms that govern these tools. These three technologies have the power to both combat and contribute to issues of disinformation, polarization, and political participation. It is urgent that we study and regulate them to our advantage before an all-out arms race begins over the use of information warfare.

This essay has highlighted the use of ICTs by political campaigns—specifically predictive modeling and microtargeting—to segment the electorate into groups of persuadable voters and target them with controversial messages on wedge issues. My analysis has shown how these campaign strategies can have negative effects on polarization and voter participation. In the era of COVID-19, our society is spending more time online than ever before. If citizens increasingly consume disinformation within their own echo chambers, the long-term consequences on political discourse could be detrimental.

The major ad-tech platforms like Facebook, Twitter, and Google have all taken different approaches to regulating the content on their platforms. The CEOs of all three companies are currently testifying before the US Congress as I write this on March 25th, 2020. Several bills in committee are designed to regulate the use of these ICTs in political campaigning. Biden has appointed multiple Big Tech critics to important posts at the Federal Elections Committee (FEC) and Federal Communications Commission (FCC). Clearly change is underway, and I encourage anyone who found this essay concerning to reach out to your local government or myself to learn how you can advocate for voter protection.

?

References

[1] Rasmus Kleis Nielsen, Ground Wars, Personalized Communication in Political Campaigns (Princeton: Princeton University Press, 2012) 7-8.?

[2] K. Ratheeswari “Information Communication Technology in Education” 3:45 Journal of Applied and Advanced Research (May 2018) 45 at 45.

[3] Darrell M West, “E-Government and the Transformation of Service Delivery and Citizen Attitudes” 64:1 Public Administration Review (January 2004) 15 at 16.

[4] Dan Berliner. “Information Technology and Public Policy” Lecture, Class at The London School of Ecnomics and Political Science, London, UK, March 8th, 2021.

[5] Clay Shirkly, Here Comes Everybody: the Power of Organizing without Organizations (New York: Penguin Books, 2009) 151.

[6] Supra Note 4.

[7] Archon Fung, Hollie Russon Gilman, Jennifer Shkabatur, “Six Models for the Internet + Politics” 15:1 International Studies Review (March 2013) 30 at 40-44. ?

[8] Tom Dobber, Data & Democracy: Political Microtargeting: A Threat to Electoral Integrity? (Amsterdam: Universiteit van Amsterdam, 2020) 8.?

[9] Clifford Bob, Strategy in information and influence campaigns: How policy advocates, social movements, insurgent groups, corporations, governments and others get what they want. New York, NY: Routledge, 2011. Page 50 of 344.

[10] Supra Note 1 at 138.

[11] Minami Funakoshi, Elizabeth Culliford, and Wen Foo, “How political campaigns use your data” Reuters, October 12, 2020, https://graphics.reuters.com/USA-ELECTION/DATA-VISUAL/yxmvjjgojvr/ ?

[12] Michael Serazio, “The New Media Designs of Political Consultants: Campaign Production in a Fragmented Era” 64:4 Journal of Communication (August 2014) 743 at 745

[13] Joe Westby, “‘The Great Hack’: Cambridge Analytica is just the tip of the iceberg,” Amnesty International, July 24, 2019, https://www.amnesty.org/en/latest/news/2019/07/the-great-hack-facebook-cambridge-analytica/

[14] William A Gorton, “Manipulating Citizens: How Political Campaigns’ Use of Behavioral Social Science Harms Democracy” 38:1 New Political Science (February 2016) 61 at 62.

[15] Normann Witzleb and Moira Paterson, “Micro-targeting in Political Campaigns: Political Promise and Democratic Risk” To be published in: Uta Kohl and Jacob Eisler (eds), Data-Driven Personalisation in Markets, Politics and Law (CUP, 2021 Forthcoming) 1 at 4

[16] Solon Barocas, “The price of precision: voter microtargeting and its potential harms to the democratic process” PLEAD 2012 Computer Science (November 2012) 31 at 31

[17] Normann Witzleb and Moira Paterson, “Micro-targeting in Political Campaigns: Political Promise and Democratic Risk” To be published in: Uta Kohl and Jacob Eisler (eds), Data-Driven Personalisation in Markets, Politics and Law (CUP, 2021 Forthcoming) 1 at 4

[18] Fabian Baumann, Philipp Lorenz-Spreen, Igor M. Sokolov, and Michele Starnini “Modeling echo chambers and polarization dynamics in social networks” 124:4 Physical Review Letters (January 2020) 1 at 1.

[19] Mathieu Lavigne, “Strengthening ties: The influence of microtareting on partisan attitudes and the vote” 27:2 Party Politics (April 2020) 1 at 8.

[20] Sarah Blaskey, Nicholas Nehamas, C Isaiah Smalls II, Christina Saint Louis, Ana Claudia Chacin, David Smiley, Shirsho Dasgupta, and Yadira Lopez, “How the Trump campaign used big data to deter Miami-Dade’s Black community from voting” Miami Herald, October 22, 2020. https://www.miamiherald.com/news/politics-government/article246429000.html

[21] Ibid

[22] Laura King, “ ‘Mountains of misinformation’ on voter fraud dim GOP Senate prospects, Georgia official warns” LA Times, December 6th, 2020. https://www.latimes.com/politics/story/2020-12-06/misinformation-voter-fraud-senate-georgia