3.1. Data for policy focus area 1: digital and data-driven transformations in governance

The vision of Data for Policy originates from extensive inquiries into the philosophy, ideation, formulation, and implementation of new approaches that lead to paradigm shifts, innovation, and efficiency gains in collective and critical decision-making processes (O’Reilly, Reference O’Reilly2011; Margetts and Dunleavy, Reference Margetts and Dunleavy2013; Pentland, Reference Pentland2013; Scholl and Scholl, Reference Scholl and Scholl2014; Engin et al., Reference Engin, van Dijk, Lan, Longley, Treleaven, Batty and Penn2019). This typically covers a vast range of topics—from the foundational aspects of democratic systems and institutions to more practical applications such as citizen service delivery, public infrastructure monitoring, and support for civil servants (Engin et al., Reference Engin, van Dijk, Lan, Longley, Treleaven, Batty and Penn2019). Reimagining governance requires transcending existing human and institutional decision-making processes and policy cycles, and establishing connections between top-down, theory-driven, and value-based thinking and bottom-up, data-driven, and emergent approaches to generate new knowledge for critical decision-making. However, the challenge lies in navigating the existing infrastructure, established communities, and policies while adapting to the new forms of engagement in sociotechnical and virtual-physical environments. Three larger areas particularly emerge from this perspective: (1) public participation and collective intelligence; (2) transformation and reimagination of relationships and organizations; and (3) openness. The stakeholder domain covers a broad spectrum of society, including individuals and organizations across the public, private, and voluntary sectors at local, national, and international scales. This raises novel questions regarding government–private sector–individual citizen interactions, data and information access, and their implications for democracy, citizen rights, and public services. An example contribution to our community that fits this area is the work by Boniface et al. (Reference Boniface, Carmichael, Hall, Pickering, Stalla-Bourdillon and Taylor2022) on the Social Data Foundation model, which combines governance with data trust services to allow citizens, service providers, and researchers to work together to transform systems.

The next three areas focus more specifically on the current methodologies, strategies, and concerns that shape data-driven transformations in governance and policy.

3.2. Data for policy focus area 2: technologies and analytics

Data for Policy research builds on both established and new data sources in all forms, coming from personal, proprietary, administrative and public sources. Similarly, both established theory-driven methods (e.g., qualitative methods and traditional/computational statistics) and the new data-driven methods (e.g., machine learning) are valuable tools for extracting useful and complementary insights from real-time and historical data. The Data for Policy community emphasizes the following technologies and analytics as primarily underpinning the current digital government and governance transformations:

• • Data science and AI are arguably the two of the most powerful technologies of our time with potential to revolutionize the way governments operate, if developed and deployed responsibly. The ability to analyze large amounts of data beyond human comprehension and making any insights available instantly to support critical decision-making processes offers unprecedented potential in public sector operations, as well as governance decisions more broadly. It can inform evidence-based policymaking, enhance service delivery, optimize resource allocation, and address societal challenges more proactively.

• • Behavioral and predictive analytics utilize data patterns and trends to analyze user/citizen behavior and forecast needs, which may inform the optimization of service delivery. This approach has the capacity for personalization and efficacy in public services. This type of capacity clearly also holds potential harms at scale. For example, discussions around China’s Social Credit System provide rich understanding of potential use cases and harms that can emerge from government utilization of such capabilities.

• • LLMs and foundation models received massive public recognition particularly with the OpenAI’s introduction of ChatGPT in late 2022. The technology essentially enabled a new interaction model between any citizen and the cutting-edge AI models. The potential impact of such technologies in governance and policymaking is enormous—ranging from processing and analyzing large volumes of information and drafting policy documents, to exploring different scenarios and handling citizen inquiries in an efficient and personalized manner. They have also sparked high profile public debate with calls from technology leaders to immediately pause giant AI experiments (Future of Life Institute, 2023).

• • Digital twins, ledger systems, platforms, cloud technologies, and so forth provide the essential infrastructure for data storage, analysis and sharing (Bennett et al., Reference Bennett, Birkin, Ding, Duncan and Engin2023), safe and secure transactions (Singh and Michels, Reference Singh and Michels2018), collaboration environments (e.g., online platforms and service nodes), and scalable computing resources—hence enabling seamless and efficient operations in the digital realm.

• • Edge analytics and federated learning are technologies that process data closer to the source, helping improve the performance and preserving privacy and security while harnessing the collective intelligence of distributed systems. For example, smart city or public health infrastructures can be empowered to maximize real-time insight generation from a diversity of data sources to localize and personalize services.

• • User interaction and experience interfaces facilitate citizen engagement and access to services, voice opinions and participate in public decision-making processes. Mobile devices, location-based technologies, and biometrics provide new forms of human interactions with the latest technologies, government services, and physical environment. Autonomous systems with attributed agency in human social interactions, such as ChatGPT, caretaker or delivery robots, and virtual reality technologies also underpin new types of human–machine ecosystems.

• • GovTech, RegTech, LegalTech, CivicTech, and so forth encompass a range of technologies tailored for specific government, regulatory, legal, and civic contexts. These technologies support digital government and governance transformations by streamlining administrative processes, ensuring regulatory compliance, facilitating legal services, and promoting citizen engagement and participation in governance.

These technologies collectively provide previously unimaginable capacity to transform governance operations and policymaking, as well as new kinds of ethical and legal challenges (e.g., the risks of a surveillance society). Furthermore, expertise on these technologies and their applications in government/policy domains are scattered across different disciplines and sectors (Hon et al., Reference Hon, Millard, Singh, Walden and Crowcroft2015). The Data for Policy community aims to bring together this diversity to lay foundations for a transdisciplinary outlook in this space, unifying different methods of knowledge creation to address the most pressing issues of our current era, through the so-called “digital revolution.” An example contribution to our community that fits this area is by Simonofski et al. (Reference Simonofski, Tombal, De Terwangne, Willem, Frenay and Janssen2022) on the use of advanced analytics (data mining, big data analysis, or AI) to detect fraud and the complexity of balancing these with legal requirements.

3.3. Data for policy focus area 3: policy and literacy for data

Beyond the opportunities in government, governance and policy, and the complications within technology and analytics environments, the third main area of interest for data for policy research deals with the policy, governance, and management issues involved in development and implementation of data-driven solutions. The governance models and frameworks for data and associated technologies are developed across the globe with variances according to local context and value judgments, as well as public literacy and acceptance. Europe’s GDPR has set a global gold standard in data practices (Voigt and von dem Bussche, Reference Voigt and von dem Bussche2017) despite its limitations (Gal and Aviv, Reference Gal and Aviv2020). Laws and regulations related to technologies such as AI and Blockchain are increasingly established through both reinterpretation of existing laws and introduction of new laws and regulations (e.g., EU AI Act—2021). Data intermediaries in the form of trusts or collaboratives, or concepts such as data banking or “trust-less” peer-to-peer data infrastructures (e.g., distributed ledger) are important areas in development as potential solutions to a range of problems associated with sensitive/personal data sharing and public–private collaborations (Boniface et al., Reference Boniface, Carmichael, Hall, Pickering, Stalla-Bourdillon and Taylor2022; Wong et al., Reference Wong, Henderson and Ball2022). From a management perspective, data supply chains, ownership and provenance, sharing and linkage, and curation and expiration are of primary concern. Closely overlapping with this, effective data and technology governance requires high-quality meta-data to be made openly available with common standards to ensure interoperability at all scales. An example that fits this area is the work by Wong et al. (Reference Wong, Henderson and Ball2022) on a data protection-focused data commons to encourage policymakers to reconsider balances of power between data subjects, data controllers, and data protection stakeholders.

3.4. Data for policy focus area 4: ethics, equity, and trustworthiness

With the rapid adoption of LLMs in particular, several high-profile experts and public figures cautioned in a May 2023 statement that mitigating AI’s extinction-level risk should be a global priority, on par with societal-scale risks such as pandemics and nuclear war (Centre for AI Safety, 2023). Conversely, many others, such as Bender et al. (Reference Bender, Gebru, McMillan-Major and Shmitchell2021), highlight more immediate concerns, including the concentration of power, the perpetuation of systemic oppression, and damage to both the information ecosystem and the natural environment. Opacity and complexity of data-driven insights, together with growing artificial agency in social and policy contexts, are arguably the main distinguishing characteristics of the ongoing data science and AI revolution. In comparison to other general purpose technologies and disruptive innovations which drastically altered preexisting economic and social structures (such as the steam engine, electricity, computers, or the internet), disruption caused by data-driven technologies, particularly AI, goes beyond process change to impact human intelligence, knowledge generation methods, and cognitive capabilities. It is therefore inevitable that the public and policymakers are increasingly concerned about potential unintended consequences, malicious actors in the system, and possible loss of human control in this space. Increasingly, the concern is that the new data-driven technologies essentially replicate and exacerbate both known and unknown problems of human decision-making processes, such as systemic inequalities, amplification of power imbalance, the widespread exploitation of human labor, and the entrenching of power imbalance (Pasquale, Reference Pasquale2015; O’Neill, Reference O’Neill2016; Russell, Reference Russell2019; Zuboff, Reference Zuboff2019). It is also becoming a computational problem to address such topics, which traditionally fall into the remit of social science and humanities. Algorithmic fairness has emerged as a particularly interesting subfield of machine learning, striving to computationally formulate, analyze and embed “fair” behavior in data-driven learning processes (Barocas et al., Reference Barocas, Hardt and Narayanan2019; Wachter et al., Reference Wachter, Mittelstadt and Russell2021). Other emerging areas include algorithm transparency, explainability and interpretability (Weller, Reference Weller2017), trustworthiness (Spiegelhalter, Reference Spiegelhalter2020), interoperability (Brown and Korff, Reference Brown and Korff2022), accountability (Binns, Reference Binns2018), and contestability (Lyons et al., Reference Lyons, Velloso and Miller2021). An example contribution to our community in this area is the article by Biddle et al. (Reference Biddle, Edwards, Gray, Hiscox, McEachern and Sollis2022) that aim to help policymakers and other stakeholders understand how data trust and attitudes on data privacy have changed throughout the Covid-19 pandemic, and which demographic groups have lower levels of trust.

Beyond the above sector-agnostic classification of the Data for Policy research landscape, we have established the following two areas of primary interest.

3.5. Data for policy focus area 5: algorithmic governance

Governance operations are becoming increasingly dependent on algorithm assistance, from democratic processes, everyday operations and public service delivery to critical life decisions at important junctures, such as credit applications, hiring processes and criminal sentencing. Algorithmic governance (AG) is an emerging field of research which systematically explores the design of social and economic processes through the use of algorithmic systems (Aneesh, Reference Aneesh2009; O’Reilly, Reference O’Reilly, Goldstein and Dyson2013; Danaher et al., Reference Danaher, Hogan, Noone, Kennedy, Behan, De Paor, Felzmann, Haklay, Khoo, Morison, Murphy, O’Brolchain, Schafer and Shankar2017; Athey, Reference Athey2017; Varian, Reference Varian2018; Yeung, Reference Yeung2018; Katzenbach and Ulbricht, Reference Katzenbach and Ulbricht2019; Gritsenko and Wood, Reference Gritsenko and Wood2020). While techno-utopian perspectives advocate for algorithmic capacity to overcome the systemic imperfections of politics (e.g., bias) and faulty forms of knowledge (O’Reilly, Reference O’Reilly, Goldstein and Dyson2013; Bhushan, Reference Bhushan2014; Kitchin, Reference Kitchin2014), techno-dystopian perspectives stress the algorithmic potential to amplify existing forms of power and control by unaccountable actors (Pasquale, Reference Pasquale2015; O’Neill, Reference O’Neill2016; Zuboff, Reference Zuboff2019). AG is also often confused with a growing body of research around the control of algorithmic behavior, which is better termed as governance of algorithms (GA) and better suited for exploration as part of the “Policy & Literacy for Data” focus area in our framework. Although AG and GA are closely related, we distinguish AG by its emphasis on the algorithmic role and agency in critical decision-making tasks, along with existing networks, hierarchies, and bureaucracies of governance. The current global debate on AG (and also GA) is largely dominated by legal and policy perspectives with limited understanding of the technology development and the landscape of engineering solutions offered in this space. Also, although there are substantial technical research communities emerging in areas very relevant to AG—such as algorithmic fairness, transparency, interpretability, accountability and adaptability—these efforts do not directly address the core governance problems and are not formulated as such. They are also predominantly reactive in nature, addressing emerging “issues” from the growing use of algorithms in society, and are limited to solutions offered in specific problem contexts and theoretical configurations (Kroll, Reference Kroll2015; Gebru et al., Reference Gebru, Krause, Wang and Fei-Fei2017; Tsvetkova et al., Reference Tsvetkova, Yasseri, Meyer, Pickering, Engen, Walland, Lüders, Følstad and Bravos2017; Narayanan, Reference Narayanan2018; Chollet, Reference Chollet2019; Russell, Reference Russell2019; Weller, Reference Weller, Samek, Montavon, Vedaldi, Hansen and Müller2019; Friedler et al., Reference Friedler, Scheidegger and Venkatasubramanian2021). There is, therefore, an urgent need for the unity of intellectual frameworks beyond disciplinary perspectives and research practices to offer holistic thought leadership in this space.

3.6. Data for policy focus area 6: global challenges and dynamic threats

The Covid-19 pandemic was yet another reminder of a crucial message: the most pressing challenges of our time transcend national boundaries, necessitating global collaboration that goes beyond the interests of any single nation. In addressing these challenges, data and associated technologies have played a central role. They have provided real-time evidence to policymakers, localized advice to individuals, facilitated the rapid development of vaccines (Latif et al., Reference Latif, Usman, Manzoor, Iqbal, Qadir, Tyson, Castro, Razi, Boulos, Weller and Crowcroft2020; Oliver et al., Reference Oliver, Lepri, Sterly, Lambiotte, Deletaille, De Nadai, Letouzé, Ali Salah, Benjamins, Cattuto, Colizza, de Cordes, Fraiberger, Koebe, Lehmann, Murillo, Pentland, Pham, Pivetta, Saramäki, Scarpino, Tizzoni, Verhulst and Vinck2020; Benjamins et al., Reference Benjamins, Vos and Verhulst2022), and underpinned economic and social responses against the pandemic’s disruption. Looking beyond the pandemic, data-driven innovation, especially with the aid of AI and when deployed responsibly, holds significant potential in tackling other major challenges. These encompass ensuring the sustainability of human existence on the planet, addressing inequalities and discrimination, promoting environmental sustainability, building resilience against global shocks, safeguarding population health from pandemics, countering security threats, and navigating hostile environments. The Data & Policy community recognizes the gravity of these challenges and actively contributes to addressing them. For instance, the Data & Policy community, through initiatives like the Big Data for Migration Alliance’s Data & Policy special collection, seeks to facilitate responsible data innovation and collaboration. This collection aims to inform humanitarian assistance and policy responses by considering the implications of data-driven approaches for the complex dynamics of migration. Moreover, the Data for Policy community actively engages in international collaboration, acknowledging the importance of cross-border partnerships and knowledge exchange in tackling global challenges. By harnessing the power of data and advancing data-driven methodologies, the Data for Policy community is committed to driving progress in these critical areas. It promotes the responsible use of data to inform policy decisions, develop innovative solutions, and foster collaboration among diverse stakeholders. Through its initiatives, the community strives to contribute to building a more sustainable, inclusive, and resilient future, where global challenges are met with collective action and data-driven insights.

Beyond the six focus areas described, data for policy research is significantly shaped by broader, fundamental factors grouped under the environment and context of data for policy research heading. This encompasses a variety of challenges, including diverse research and innovation cultures, limitations of legacy infrastructure, and physical environmental constraints. It also considers the dynamics between public and private sectors, evolving collaboration models, the emergence of new inequalities alongside a widening digital divide, societal concerns such as job loss fears tied to economic shifts (like those seen with the sharing economy), and demographic constraints. These factors underscore the need for ongoing, in-depth discussions that extend beyond the scope of this paper.

4.1. Advanced computation versus policy and practice

This axis delves into the interplay between advanced computation in data science and practical policy implementation. Advanced computation, grounded in computer science, statistics, and mathematics, contrasts with policy practice, which tackles socioeconomic and political issues through social sciences and humanities. The Data for Policy community champions a transdisciplinary approach, integrating diverse insights into pragmatic solutions for societal challenges. Our initiatives include:

• • Strategic landscape reviews to monitor the evolution of the data for policy research and practice to support all stakeholders in this space;

• • Promoting “translational articles” that bridge research and practice;

• • Incorporating multiple perspectives in our committee and peer-review processes;

• • Employing a multimodal dissemination strategy, including an open-access repository, the Data & Policy journal, YouTube, and social media.

Through interdisciplinary and cross-sector collaboration, we aim to evaluate and leverage data-driven methods in relevant socioeconomic contexts. Our goal is to foster dialogue, knowledge exchange, and impactful policymaking.

4.2. Local versus global

This axis in data for policy initiatives encapsulates the interplay between local/national and global perspectives. The global aspect focuses on wide-reaching challenges and solutions spanning multiple disciplines, sectors, and geographies, fostering international collaboration and resource sharing. In contrast, the local perspective prioritizes bespoke solutions attuned to specific regional needs, respecting unique social, cultural, and political contexts. Our key initiatives include:

• • Regional conferences—our 2022 conference edition expanded to regional events in Hong Kong, Seattle, and Brussels, reflecting distinct digital cultures of China, the United States, and Europe. Plans to extend to more locations, especially in the Global South, are underway.

• • Our Sub-Saharan Africa Engagement Programme, in collaboration with the Bill & Melinda Gates Foundation, promotes a two-way knowledge exchange. It supports the SSA data and AI ecosystem to address local challenges while integrating regional insights into the global discourse, particularly in agriculture, food security, and climate change adaptation.

These initiatives aim to balance global collaboration with local action, ensuring contextually relevant and sustainable data for policy practices and outcomes.

4.3. Open versus proprietary

This axis explores the continuum from open, collaborative innovation to traditional, proprietary methods. Open innovation values transparency, accessibility, and equity, encouraging knowledge sharing and co-creation. Conversely, proprietary innovation, with its clear financial models and control, often conflicts with open practices, raising concerns over safety, security, and ownership. Our key initiatives along this axis include:

• • Ensuring all our publications and dissemination channels are available on an open-access basis, and support unfunded authors through a proactive campaign.

• • Facilitating connected discussions between industry stakeholders and public interest research groups.

• • Encouraging and supporting new GovTech ventures to embrace open innovation practices.

These steps reflect our commitment to blending open and proprietary innovation for ethical, inclusive data-driven policies and technology innovation.

Through our commitment to these axes and the six areas of data for policy, we aim to drive global debate and inspire local solutions. By facilitating constructive dialogue, promoting knowledge exchange, and enabling effective and impactful policy decisions and practices, we envision a future where data-driven technologies and methodologies contribute to a more informed, equitable, and sustainable world for all.