Why We Need Industry-Independent Research on Tech & SocietyJanuary, 2020
When research is funded by companies that people no longer trust, we have a problem.
Is too much red meat bad for your health? The fate of a trillion dollar industry depends on the answer.
Last year, researchers published controversial new findings that encouraged people to continue eating as much red meat as they like. They didn’t mention that the lead author was funded by the beef industry. The journal has now issued a correction, and today’s Washington Post’s headline reads, “You’ll never guess who helped to fund that controversial ‘keep eating red meat’ study.”
Even if the science was good, it’s now much harder to believe the authors’ claims.
Have we reached this point with the tech industry? How can researchers advance the public interest when a majority of Americans (and other societies too) don’t believe that tech companies can be trusted to do the right thing most of the time?
It’s now three years since multiple fiascos in 2016 convinced Western democracies that the internet might not be the uncomplicated miracle promoted by tech firms and the US government. Hopes are now wearing thin¹ toward corporate promises to study and manage their own social impact. In 2019, researchers famous for their independent voices were pushed out of firms that employed them. In privacy research, scientists organized campaigns to prevent surveillance companies from sponsoring a major research conference.
While some fields couldn’t exist without close industry ties, it’s time to prioritize industry-independent research on technology and society. In this post, I describe the unique value of independent research to the public. I also describe how the Citizens and Technology Lab plans to maintain our independence from the companies we study.
This post, which represents my opinions only, is based on work with the Cyberlaw Clinic on Conflict of Interest policies for the Citizens and Technology Lab (CAT Lab, formerly CivilServant). I expect some version these policies to be fully in place by mid-2020, as we complete our transition into Cornell.
How Funding Shapes What Your Research Can Achieve
How do you want your research to create change? Your answer will shape your approach to funding ethics.
At CAT Lab, we work for a world where digital power is guided by evidence and accountable to the public. As a citizen science organization, we believe we can do high quality, public-interest research by working directly with the public rather than begging companies for permission and resources (see examples of our work).
Our theory of change is adjacent to journalism, consumer protection labs, and environmental citizen science. By testing tech policies/products and testing citizen ideas for change, we develop credible knowledge that can be used by anyone. As a lab that grew out of audits of platform policies about online harassment, we are committed to asking public-interest questions even when companies find them uncomfortable. In my PhD at MIT’s Center for Civic Media, I demonstrated that industry-independent research can change people’s lives and influence policy in powerful ways.
Differently from CAT Lab, many researchers attempt to influence industry by combining collaboration with education. In computer science, industry success is one of the field’s goals and revenue sources. In this model, managers inside companies support researchers to advance manager agendas/careers, burnish their firm’s reputation, access future employees, get new ideas, build allies, and directly/indirectly influence the direction of research. Researchers do good work and gain prestige by being close to companies, their money, and their data. When students graduate and work for those companies, university prestige and revenue also grow.
Taking industry money can also increase a researcher’s impact. Companies often listen more to researchers who draw the greatest consulting fees, grant amounts, and alumni donations. Ultimately, collaboration with industry works best when researchers need a close relationship with firms and lots of money to do good work–the closer you are to power, the more effective you can be.
we can do high quality public-interest research by working directly with the public rather than begging companies for permission
After a century of scandals over corporate influence on research in medicine, public health, nutrition, and climate change, many institutions accept industry money while trying to protect researchers from substantive influence. These mechanisms include undirected gifts, research fellowships/grants, industry labs, consortia like the Partnership on AI, corporate foundations like Google.org, and many others. But remember: even when companies don’t directly tell researchers what to study or alter findings, they set the agenda with topic-specific grants and decisions about which applications to fund.
Good work is expensive, so researchers and institutions take funds from industry while trying to stay independent-minded. Examples include convenors of people and ideas like Data & Society, the MIT Media Lab, the New America Foundation, Princeton’s Center for IT Policy, etc. To work well, these organizations need to be (a) useful enough for companies to invest, (b) critical enough to be trusted by the public, and (c) have institutional protections for intellectual freedom.
Managing Integrity & Credibility
At CAT Lab, public trust is one of our most precious resources. Like other research organizations, we work to manage the integrity and credibility of our research.
Integrity refers (among other things) to our independence in the questions we ask and how we ask them. By avoiding funding from companies that we might study, we protect ourselves from funding-related influence, both conscious and unconscious. Instead, we prioritize relationships and collaboration with the communities we hope our research will serve.
Credibility is even harder to manage and protect than integrity– it refers to the perceived independence of our research. Many people have learned from experience to mistrust the tech industry. Numerous scandals, including Cambridge Analytica and the Facebook emotion contagion study, have also given the public reason to mistrust researchers.
Until recently, industry funds enhanced reputations and helped researchers do even better work. Now, industry funding is hurting research credibility on some topics, and that trend may continue. Last year, I watched Wikipedians reject a collaboration funded partly by Facebook grants. Academics have publicly attacked the credibility of economists working with Uber because their industry-funded research depends on secret datasets that can’t be independently verified.
When CAT Lab audits a platform or test an idea for change, we don’t want public distrust of tech firms to cause doubts in the integrity of our research. We also need to earn people’s trust before they consent to participate in citizen science by sharing their data or allowing us to intervene in their lives and communities.
academics have publicly attacked the credibility of economists working with Uber because their industry-funded research depends on secret datasets that can’t be independently verified.
Historically, citizen science has been powerful in sectors where the public mistrusts companies and government (food, the environment, medicine, biodiversity, car safety, and consumer protection).
Public mistrust motivates people to take photos, measure water quality, count birds, and contribute to knowledge in the public interest. To succeed, citizen science organizations need to cultivate and maintain this integrity and credibility over the long term.
Practical Ways to Manage Industry-Independent Integrity and Credibility
CAT Lab takes the following steps to manage conflicts of interest in our research:
- We do not take money from any for-profit corporations in the technology industry, nor do we take money from any platform on which we study behavior. This includes any offers of travel, accommodation, or other perks.
- We do not accept contracts with tech companies that give them any right to pre-publication review of our papers (that is, entitlement to read a copy of our findings before we release them). We may choose to share our findings with a company before release if and when we feel it is in the public interest. Julia Angwin calls this “contestational peer review.”
- We do not accept any contract that contains a right to pre-publication veto.
- We do not sign non-disclosure agreements (NDAs) with platforms that will limit our rights to publish our research. We will only sign an NDA regarding the protectable business assets of a company, such as its intellectual property or business information, if required by the company to study a particular phenomenon, but will not do so where it impacts our ability to fully publish our research.
We are also working on policies and organizational processes including:
- Disclosure: staff members will disclose financial conflicts of interest and NDAs with studied entities (read my disclosure here)
- Refusal: staff members will refuse stock, gifts, consulting fees, and NDAs
- Recusal: if a staff member does have a conflict of interest, they must recuse themselves from decision-making related to those companies or their competitors
We are also working on a policy for students. Public-interest tech jobs are rare. Many graduate programs assume that students will take summer internships. A good policy will support students to learn and explore while also protecting the integrity and credibility of our research. Disclosure and recusal will be central to that policy.
Is Industry-Independent Public-Interest Research Sustainable?
Can society actually afford industry-independent research and accountability on digital power? I don’t think we can afford a world without it, and I’m not alone.
To name a few, Consumer Reports (who is one of CAT Lab’s partners) is funded by philanthropy and membership dues from millions of members. Ranking Digital Rights, the Tow Center, The Markup, and The Dangerous Speech Project are all strongly committed to this kind of independence. I’m sure I have missed others.
Across sectors, industry-independent research is a basic tool for safeguarding the public and advancing knowledge. It’s a recognized necessity in consumer protection, medicine, the environment, and public safety. As digital power becomes a basic part of human society, researchers, advocates, and funders need to work together on industry-independent institutions that can be uncompromising in our pursuit of the public interest.
While this post represents my own views, I’m very thankful to Kendra Albert, Jon Penney, Zeynep Tufekci, Julia Kamin, Susan Benesch, Ivan Sigal, Julia Angwin, Neil Lewis Jr., Amar Ashar, and Christo Wilson for conversations that have shaped my thinking.
¹ Update, February 25, 2020. Since this post was written, Facebook made available to researchers a dataset of website URLs shared on Facebook. The academic organizers at Social Science One cite legal negotiations with Facebook as a main challenge in completing the project.
- Carpenter, D. (2014). Reputation and power: organizational image and pharmaceutical regulation at the FDA (Vol. 137). Princeton University Press.
- Davis, D. L. (2002). When smoke ran like water: Tales of environmental deception and the battle against pollution (p. 45). New York: Basic Books.
- DeTar, C. (2016). On Selling Out. How the conditions of technological design perpetuate the status quo. https://tirl.org/writing/cfd-on-selling-out.pdf
- Ehrenberg, R. G. (2009). Tuition rising. Harvard University Press.
- Ko, A. Mou, M. Matias, J.N. (2016) The Obligation to Experiment
- Matias, J.N. (2016) A toxic web: what the Victorians can teach us about online abuse. The Guardian
- Ornstein, Charles (2017). From Twitter to Treatment Guidelines, Industry Influence Permeates Medicine. NPR
- Silber, N. I. (1983). Test and protest: The influence of Consumers Union. New York: Holmes & Meier.
- Soares, M. J., Müller, M. J., Boeing, H., Maffeis, C., Misra, A., Muscogiuri, G., … & Zhu, S. (2019). Conflict of interest in nutrition research: an editorial perspective.
- Farrell, J. (2016). Corporate funding and ideological polarization about climate change. Proceedings of the National Academy of Sciences, 113(1), 92-97.
- Weiss, C. (1979) The Many Meanings of Research Utilization. Public administration review.
- Group, C. P. S., Irani, L., Salehi, N., Pal, J., Monroy-Hernández, A., Churchill, E., & Narayan, S. (2019, November). Patron or Poison? Industry Funding of HCI Research. Panel at CSCW 2019