The exponential growth in the production and use of data, intensified by the COVID-19 crisis, has created vast amounts of new data. These data have incredible potential for development: from helping to improve decision making, to mapping the spread of diseases, to providing marginalized communities with access to online services.
At the same time, these data can be misused to cause harm to individuals or groups, either willfully or through weak or improper data governance and management practices. Enabling trusted and effective use of data for development requires the adoption of rules and harmonized standards for the formatting, archiving and dissemination of data. Similarly, ensuring the value from data is equitably distributed requires clear rules on who has access to and control over data infrastructure, and even what data are collected.
This is the role of data governance. It is about designing and implementing the norms, rules, and infrastructure that will foster trust by shaping the way people, organizations, and governments can legitimately collect, use, and share data to create sustainable value that is realized equitably.
All stakeholders have a role in governing data effectively
the process is critical in promoting trust and legitimacy, and therefore incentivizing participation in the data economy.This participatory approach is the basis of what we call multistakeholder data governance. It is designed to culminate in shared principles and rules that foster engagement and buy-in. But the fairness, inclusivity, transparency and effectiveness of
Key stakeholders include governments, the private sector, civil society organizations (CSOs), academia, and individual producers and users of data. These actors may have competing incentives, play different or overlapping roles, and may have more or less power to influence the decision-making process. For example, governments take the lead on policy making and regulation, implementation and enforcement of rules, and fostering multilateral coordination and cooperation. CSOs, academia, and the private sector can contribute technical expertise and their perspective as end users to enrich the quality of policies, laws, and regulations. The private sector and nongovernment organizations (NGOs) can also develop principles (such as the FAIR); technological solutions; and standards (such as the W3C guidelines for data category vocabularies), including through standard-setting organizations, and consortiums. Another good example is the Global Partnership for Sustainable Development Data’s Data Values Project.
CSOs and the private sector can also support better data governance practices through advocacy and by building the capacity and literacy of data users. Two good examples are the programs offered by the GovLab Academy; and Data Pop Alliance’s “collaborative laboratory,” which conducts training sessions and workshops and provides forums to share knowledge.
Multistakeholder data governance approaches should be purpose-driven and enable meaningful engagement
course on the basics of Artificial Intelligence (AI) with the aim of training one percent of the country’s population. After scaling the program to over 250 private sector participants, the Ministry of Foreign Affairs and the Tax Authority used the course to train their staff. Here, buy-in grew gradually to involve naturally interested stakeholders from the public and private sectors, led by academia, to fulfil the policy objective of making Finland competitive in AI. Starting with a specific sector can also help create momentum, while putting in place foundations that can be scaled. Citizen Science, an NGO whose members include fishermen, indigenous peoples, and students, monitors fish migration and water quality in the Amazon basin to guide policy making in an industry critical to the livelihood of more than 30 million people.For sector-specific rules, it may not be necessary or appropriate to open consultation to stakeholders beyond the relevant industry and interested parties. Moreover, initiatives do not necessarily need to start with government. For instance, in Finland, the University of Helsinki and private sector partners developed a
However, if the data policy issue has government-wide and society-wide impact, such as in the use of facial recognition technologies in the public sector, the process needs to have a broader coalition of stakeholders, including underrepresented groups. For example, after calling for a moratorium on facial recognition and sparking a public debate on the ethical and legal implications of biometric technology, the Ada Lovelace Institute established the Citizens’ Biometrics Council in 2020. The Council’s recommendations to determine what forms and uses of biometric technologies are acceptable were informed by a wide variety of views.
Within the chosen forum, the participation of non-government stakeholders must be meaningful. This requires giving them real input into the decision-making process. Mere consultation of these groups by government is not sufficient to foster trust and legitimacy and may even backfire. This does not mean participation processes must be formal: bottom-up contributions to nonbinding decision-making processes can be meaningful if they are integrated and feedback loops are created.
Effective multistakeholder data governance is challenging to implement in practice
A lack of transparency and openness of the proceedings, or barriers to participation, such as prohibitive membership fees, will impede participation and reduce trust in the process. These challenges are particularly felt by participants from low- and middle-income countries (LICs and LMICs), whose financial resources and technical capacity are usually not on par with those of higher-income countries. These challenges affect both the participatory nature of the process itself and the inclusiveness and quality of the outcome. Even where a level playing field exists, the effectiveness of the process can be limited if decision makers do not incorporate input from other stakeholders.
Notwithstanding the challenges, multistakeholder data governance is an essential component of the “trust framework” that strengthens the social contract for data. In practice, this will require supporting the development of diverse forums—formal or informal, digital or analog—to foster engagement on key data governance policies, rules, and standards, and the allocation of funds and technical assistance by governments and nongovernmental actors to support the effective participation of LMICs and underrepresented groups. At the global level, this will also require international cooperation to support consensus-building, while creating an environment for people from LICs and LMICs to meaningfully contribute as “standard setters” rather than as mere “standard takers.”
As we experiment with different models for such approaches, it’s important to share experiences and lessons learned. Watch out for the space here as we continue the dialogue on this topic. See below the first of a series of dialogues discussing the opportunities and challenges of multistakeholder governance:
To download the full report, click here.
To learn more about the World Development Report 2021, please visit this website.
 This represents Finland’s aspiration to occupy a niche as world leader in practical applications of AI and remain competitive in the global AL industry, according to Economy Minister Mika Lintilä. See Politico, Finland’s Grand AI Experiment, https://www.politico.eu/article/finland-one-percent-ai-artificial-intelligence-courses-learning-training/.
 For details, see Spotlight 8.2: Promoting Citizen Science in the Amazon Basin: https://elibrary.worldbank.org/doi/10.1596/978-1-4648-1600-0_Spotlight8_2