The Forum Network is a space for experts and thought leaders—from around the world and all parts of society— to discuss and develop solutions now and for the future. Aiming to foster the fruitful exchange of expertise and perspectives across fields to help shape Better Policies for Better Lives, opinions expressed do not necessarily represent the views of the OECD.
Thanks to the introduction of new regulations in the United States and Europe, researchers are set to access more data from social media than ever before. Setting aside the monumental task of sorting out access rules, how well-equipped are researchers to use these data?
The short answer is poorly. Processing and analysing massive numbers of posts, often involving multimedia, requires engineering resources that few research centres currently have. And that is before we even get to studying how social media activity interacts with the broader information environment across print, radio, television and websites. Data access alone is not enough. To make sense of a complex and polluted information environment, the research community needs shared infrastructure. Fortunately, there are models for scaling up scientific research that would meet this challenge.
What kind of information environment can foster democratic societies and encourage active citizen participation? Sadly, the evidence needed to guide policy-making and social action in this domain is sorely lacking.
Before we get to those models, let’s remember what’s at stake. The need for better research models transcends academic concerns. Democracies around the world are grappling with how to safeguard democratic values against online abuse, the proliferation of illiberal and xenophobic narratives, malign interference, and a host of other challenges in our rapidly evolving information environment. Some are even backsliding into authoritarianism as information pollution, alongside increasing controls over content in some countries are eroding public trust in institutions. What kind of information environment can foster democratic societies and encourage active citizen participation?
Sadly, the evidence needed to guide policymaking and social action in this domain is sorely lacking. We have many anecdotes and case studies but little systematic evidence about how the information environment works as a system, or what the longer-term effects of threats like disinformation are; even less is known about the impact of interventions to counter such threats. While research is trickling in, it tends to be limited to North America and Europe; mostly focuses on the easiest platforms to study; is more qualitative and anecdotal than quantitative; and so far has limited policy relevance.
So what’s the solution?
One reason research isn’t farther ahead is that how social media works—and thus how they shape the overall information environment—is constantly changing. Research needs to move fast to keep up.
Getting more data is an important first step to doing so. The operational reporting and data access provisions in the European Union’s Digital Services Act are a critical advance. The impact of such laws will be strengthened if other democracies adopt similar approaches, ensuring companies cannot work around new rules—a process the OECD is well-placed to help on.
More on the Forum Network: Democracy: What’s at stake? by Elsa Pilichowski, Director for Public Governance, OECD
Our modern democracies, while not without flaws, are worth fighting for: they guarantee us many freedoms that we now take for granted. The OECD is at the forefront of the effort to advance concrete actions to address the pressing challenges facing democracies today.
But speeding up research also requires getting radically more efficient at how we build knowledge in the first place. Our current research production model is piecemeal and wildly inefficient. Some companies build high-quality interfaces for select information, but most do not. Instead of accessing representative samples or baselines on a given information ecosystem, researchers start anew each time.
Each research centre develops their own data collection and processing pipelines, which are rarely maintained over time or designed with an eye to reusability across multiple studies. And the engineering work behind these studies is mostly done by graduate students or post-docs with little training or experience. Instead of doing research in their trained fields, future social scientists are building bespoke data science pipelines.
Funders mostly support work on a per-project basis, making it hard for scholars to set up labs. And those who do are not funded to maintain robust engineering teams. While the US National Science Foundation has funded at least one new initiative to track the full online behaviour of thousands of people, that initiative covers only volunteers from the United States and is thus just a fraction of what is needed.
This production model puts the research community perpetually behind the curve in studying key questions, such as the impact of deplatforming adversarial state media and the role of misinformation in how people respond to public health guidelines. If the research community is going to help policymakers understand whether interventions mandated through the likes of European Commission’s Code of Practice on Disinformation are working, major infrastructure investments are necessary.
This is nothing new. As societies, we invest in shared scientific infrastructure all the time, from electron microscopes for materials research and massive surveys for measuring employment to satellites in polar orbit for understanding climate and weather.
The European Centre for Nuclear Research (CERN) has brought scientists from around the world to collaborate and accelerate scientific discovery by working with shared infrastructure to unlock the mysteries of the physical universe. Such large-scale collaboration is once again needed (...), this time to unlock the mysteries of the information environment.
The history of particle physics is especially instructive. For more than 70 years, the European Centre for Nuclear Research (CERN) has brought scientists from around the world to collaborate and accelerate scientific discovery by working with shared infrastructure to unlock the mysteries of the physical universe. Given the need to accelerate particles at extremely high velocities, the CERN model enables physicists to use shared machines that could not be built with the resources of any single academic institution.
Such large-scale collaboration is once again needed to connect scholars, policymakers and practitioners internationally and to accelerate research, this time to unlock the mysteries of the information environment. What is needed is a multi-national facility to answer these difficult questions: the Institute for Research on the Information Environment (IRIE).
This model would help scholars quickly test their ideas in many settings, without having to learn each anew. IRIE could create standardised datasets of labelled posts, which would speed the development of tools to make sense of the conversation, or build tools to help more researchers process image and videos at scale to understand the spread and impact of memes. And by enabling impact evaluations of the dozens of publicly-announced changes in platform practices every year, it would build the mosaic of evidence we need to understand the likely response to new policies before they are implemented. In short, IRIE would connect all the disparate parts together that must converge for democratic societies to test and articulate a plan for stewardship of the information environment.
If the pandemic has demonstrated one thing, it’s that our heavily polluted information environment makes public decision-making exceptionally hard. If conspiracy theories and disinformation hindered public health measures during an immediate life-or-death crisis, what hope do democracies have in responding to more complex, politicised challenges like climate change? Identifying that there is a problem is only the first step. Understanding what to do about it is much harder.
Current approaches to research on the information environment are too slow and inefficient the task, but there are better models. If we work together to speed up observation, measurement and analysis, scientists can do their part to enable better policymaking for the information environment. Democracies need an Institute for Research on the Information Environment.