The idea for The Influence Industry started in March 2016, when by a stroke of luck I found myself sitting on a plane from Berlin to London next to the founder of a start-up. At the time, Tactical Tech was undertaking an investigation into the work of data brokers – those who trade in the data of millions of people at the national, regional and international level. We had been attending trade fairs and conducting anonymous interviews to try to better understand the ins and outs of the industry.1 The person sitting next to me on the plane was working on a Powerpoint presentation about selling data. I started a conversation and discovered that he was the proud CEO of a company buying, analysing and reselling mobile phone data. As someone who had clearly spent a lot of time thinking about the pitfalls and challenges of his company, I found him quite open to talking about his business model. In the process, he told me in passing that they had been approached by a major political party in Germany to see how the data profiles they held could be used to target key regions in future elections. This was striking to me for two reasons. First, just how important it was who was buying the data and for what reasons – something we had not really explored before. Second, if this was happening in Germany, a country widely regarded as having strict data-protection laws and a high regard for privacy, then what was happening elsewhere?
Working from this starting point, I found an entire sector built around the acquisition and use of personal data for elections: consultants, strategists and start-ups, as well as the biggest and most established platforms, like Facebook and Google, all offering data services to political campaigns. I knew that personal data and tech platforms had been used systematically and at a large-scale in the Obama election campaigns; at Tactical Tech we had worked in the past with some of the data-scientists who were involved in this. But intuitively this felt different. The people we had known on the Obama campaign were tech-activists and volunteers who took a break from their regular jobs during an election cycle to help out. This was another story; this was an established ecosystem of commercial companies across the political spectrum.
On looking around further, I found that a variety of actors were turning the significant innovations of the commercial digital marketing and advertising industry over the past five years toward a client-base of politicians, political parties and political strategists. New techniques in digital profiling, targeting and persuasion – which were being tuned with ever-increasing accuracy to listen to and persuade consumers to buy more products – were now being applied to a far more significant set of ‘returns on investment’, as the industry-speak goes. These data-driven persuasion techniques were being applied with a promise not of influencing sales but of influencing votes. To these politically-orientated data and tech companies, likes were equivalent to political opinions and consumers were interchangeable with voters. The deeper these companies' and strategists' knowledge appeared to be of the public's fears and hopes, the more they claimed they could turn their insights into political results and thus the more profit they made.
As the co-founder of Tactical Tech, an organisation focused on political participation, I wanted to know how this could impact the outcome of elections and efforts to gain and maintain political power. As a technologist and researcher, I wanted to know exactly what kind of data was being bought and sold and the mechanisms for extracting value from it. As a citizen, I wanted to know how other people would feel about this use of their data if they knew about it. The existing body of academic work I found on the subject helped orient me, most notably the work of Phil Howard, Daniel Kreiss and Colin Bennett. A host of important journalistic articles led me deeper into specific cases, such as the Ted Cruz campaign’s use of ‘geo-fencing’ in the run-up to the 2016 US elections.2 But I found very little material that looked practically at how these techniques worked at the intersection of technology and politics or within the context of the broader data industry. I found even less that looked at these methods and practices outside of the US.
Tactical Tech's research project got funded and off to a start just as the full force of these questions was coming to light, with the Brexit vote starting to be scrutinised and Trump being elected. After bringing in two colleagues – Varoon Bashyakarla, a data-scientist wanting to see the world outside of Silicon Valley, and Gary Wright, a researcher with the same obsession for the twinning of politics and data as myself – we began what has become in equal measure a fascinating and frustrating journey – and one which shows no signs of slowing down. The project has dominated our attention, taking us deeper into the world of data and politics – from an obscure conference in Arlington, Virginia in early 2017 to hear Trump's digital strategist speak first-hand, to a stint in late 2017 collaborating with the Channel 4 investigative team looking into the elections in Kenya. Along the way we picked up new colleagues and support from the broader team at Tactical Tech and our Exposing the Invisible project, most significantly Raquel Renno, a researcher and tech-activist from Latin America with a background in opinion polling; Amber MacIntyre, a PhD student with a shared passion for data and activism; and more recently the word-smithing wizardry of Christy Lange who has joined the project team as it has grown in 2018.
Over the last year, we have been fortunate enough to build associations with some of the leading digital rights groups, lawyers, academics and journalists working in the broader areas of internet politics and data privacy around the world, through more than 15 partnerships in Africa, Asia, Europe, Latin America and North America who have been essential allies in carrying out the country case studies. We have also been able to collaborate with leading international actors, including Ravi Naik, one of the most prominent human rights and data protections lawyers in the UK, now working on the Cambridge Analytica case; Frederike Kaltheuner and Lucy Purdon at Privacy International; and the Oxford Internet Institute’s Computational Propaganda Project, where I am now a Visiting Industry Associate. Working with this range of partners is an ongoing process, with more work to come over the next year.
Thanks to some of these collaborations, we were aware that a significant story about Cambridge Analytica and Facebook was going to break in March 2018. We had no idea, however, just how much it would capture the world's attention. The unfolding media narrative has focused not only on the politics but also the business of collecting and using personal data. For many people these facts were new, for others they were a reminder of just how much insight some parts of the technology industry have into our everyday lives and how much effort goes into interpreting and monetising it – from small companies that make apps to some of the world's biggest platforms. In this sense, it is an important ‘teachable moment’ for the wider public and an opportunity to understand more broadly how this enormously profitable industry turns users' free services into valuable assets.
The revelations about Cambridge Analytica and Facebook have galvanised public and media attention around two open-ended questions that have also driven our research: What are the implications of applying the logic and technologies of the commercial data industry to politics? And what does this mean for our democracies now and in the future? Now that the dust is beginning to settle on this story, it is a good moment to try to answer these questions, to look systematically at the practices of using data in political campaigns, and to get a more in-depth sense of what is still a relatively opaque set of practices. And to view the Cambridge Analytica and Facebook story in the context of a broader ecosystem: an industry that is extensively using, buying and selling data in elections and political campaigns and exporting and trading these services across borders. For this, we also need to answer the questions: What is business-as-usual for modern campaigns? And how can we differentiate this from potentially unethical and manipulative practices? More fundamentally, where are our ethical lines and what boundaries do we need to draw around notions of consent, fairness and agency when it comes to the use of personal data and digital influence in the political process? Finally and most essentially, we need to grasp the implications of using these methods in different contexts and examine how they play out depending on the cultural, socio-political and technological environments in which they are used.
Over the next 12 weeks we will be publishing a collection of articles, dossiers and papers, which will be collated into a report called ‘The Influence Industry: The Global Business of Using Your Data in Elections’, released in the second half of 2018. This documentation and analysis will begin to address some of the questions highlighted above, to surface some of the most pressing concerns and to contribute to broader discussions about what should be done. We hope you join us on this journey as it leads through an ever-widening circle of enquiry into the relationship between data and politics and the current working practices of what we refer to as The Influence Industry.
The content of our report is organised around three core elements related to the use of data and digital influence in political campaigns: the practices, the actors and the contexts. These are broken down as follows:
Practices: How data is used to influence you politically
Exploring more than ten primary methods related to the use of political data and digital influence in campaigns, we examine how they work and look at examples of how they are being applied within political campaigns around the world. Specifically, we look at:
Data as a Political Asset: this section focuses on the valuable troves of existing data on potential voters, and how they are exchanged between political candidates, acquired from national repositories or sold or exposed to those who want to leverage them.
Data as Political Intelligence: this section focuses on how data is accumulated and interpreted by political campaigns to learn about voters' political preferences and to inform campaign strategies and priorities, including creating voter profiles and testing campaign messaging.
Data as Political Influence: this section examines how data is analysed and used to target and reach potential voters, with the aim of influencing or manipulating their views or votes.
Actors: Who is selling data-driven political influence and how do they operate?
Investigating the key digital and political consultants, tech companies and platforms buying, selling and using data in political campaigns. We study the broader ecosystem of the political data trade. We map over 50 actors, looking closely at a number of the largest players. These are divided into:
Digital consultants and strategists who work across multiple elections and borders. This includes a host of companies working across the political spectrum, with Cambridge Analytica being just one of them.
Start-ups, data brokers and tech companies that specialise in bespoke and off-the-shelf tools and services for gathering, analysing and utilising data in political campaigns. These range from small companies making voter apps for European political parties to some of the now-standard political-tracking technology companies operating worldwide.
Technology platforms that sell digital advertising and services for using data in political campaigns. This includes a suite of political influencing tools created by Facebook, as well as some of the less explored, advertising-based big technology companies, such as Google.
Contexts: How is data-driven political campaigning used in different elections around the world?
Working with a cross-disciplinary team of academics, journalists, lawyers, technologists and privacy activists, we have investigated how data and digital influence is used in political campaigns in more than 12 different countries. Each of these country studies is carried out in collaboration with local partners who have a first-hand understanding of the cultural, socio-political and historical context of each country and a long-term interest in technology and the data industry. The studies include Argentina, Brazil, Canada, Chile, France, India, Italy, Kenya, UK and the US, among others.
Conclusions: What does this all mean and why should we care?
Based on our research, we will publish a collection of features that share our own analyses and reflections. We focus on surfacing some of the core findings that emerged from our own research and sharing some of our observations. In the coming months you can expect to find analysis and opinion pieces on:
- ‘The rise of the influence industry: is it really new?’
- ‘Does data-driven political campaigning actually work?’
- ‘What happened to the democratising force of the internet?’
- ‘What is the real platform for political influence in the global south?’
- ‘Maintaining power: how are data and digital influence techniques used in-between elections?’
- ‘What are the options for monitoring and controlling digital influence in political campaigning?’
- ‘Money, power and right-wing experiments in digital influencing.’
Read our first installment of Political Data Practices ‘Persuasion by Personality: The Use of Psychometric Profiling in Elections' by Varoon Bashyakarla.
Stephanie Hankey is the co-founder and Executive Director of Tactical Tech and is currently a Visiting Industry Associate at the Oxford Internet Institute at the University of Oxford. Her work combines her technology, design and activism background. She currently writes, consults and teaches on 'the politics of data' and ethics in technology design.
2 https://www.washingtonpost.com/politics/cruz-campaign-credits-psychological-data-and-analytics-for-its-rising-success/2015/12/13/4cb0baf8-9dc5-11e5-bce4-708fe33e3288_story.html?utm_term=.845e9bc7b4cf ↩
Published 20 April 2018