From Digital Sovereignty to Digital Agency

From Digital Sovereignty to Digital Agency

Brief

Shutterstock

Akash Kapur

At a Glance

• In recent years, governments have increasingly pursued variants of digital sovereignty to regulate and control the global digital ecosystem. The pursuit of AI sovereignty represents the latest iteration in this quest.
• Digital sovereignty may offer certain benefits, but it also poses undeniable risks, including the possibility of undermining the very goals of autonomy and self-reliance that nations are seeking. These risks are particularly pronounced for smaller nations with less capacity, which might do better in a revamped, more inclusive, multistakeholder system of digital governance.
• Organizing digital governance around agency rather than sovereignty offers the possibility of such a system. Rather than reinforce the primacy of nations, digital agency asserts the rights, priorities, and needs not only of sovereign governments but also of the constituent parts-the communities and individuals-they purport to represent.
• Three cross-cutting principles underlie the concept of digital agency: recognizing stakeholder multiplicity, enhancing the latent possibilities of technology, and promoting collaboration. These principles lead to three action-areas that offer a guide for digital policymakers: reinventing institutions, enabling edge technologies, and building human capacity to ensure technical capacity.

Every year, China holds a gathering known as the World Internet Conference in Wuzhen, a scenic town on the Yangtze River. At the first conference, in 2014, delegates woke to find a draft joint declaration slipped under their hotel doors overnight. The declaration, whose language had neither been discussed nor debated, included a determination to "respect Internet sovereignty of all countries." Delegates were informed they had until 8 a.m. to send an email to the conference organizing committee if they wished to change any of the language.

In the event, many delegates pushed back, and the declaration, whose origins remain mysterious to this day, was never officially agreed upon. The whole event, however, was a harbinger of things to come: A growing number of countries now seek to wall off their own segments of the digital realm, and questions of sovereignty have taken center stage in the arena of global digital policymaking.

In retrospect, the conference was an inflection point in the shift away from the uncontrolled, Wild-West digital domain of the 1990s, where cyber-libertarians thumbed their noses at the power of nation-states and old-fashioned regulation. Since then, geopolitical power has become more diffuse, the post-Cold War faith in liberal democracy and free markets has waned, and the global digital ecosystem has shed much of its emancipatory promise. Partly in response to this changing context, nations and global institutions have increasingly called for greater digital sovereignty. After years of all-but-abdicating authority, regulators and policymakers are rushing to reclaim previously-ceded territory, whether in the name of individual rights, control over their populations, or national economic and security interests. This trend is likely to intensify as artificial intelligence (AI) continues its rapid spread into virtually every aspect of society, culture, and economy; 69 countries have already enacted over 1,000 laws and policies governing AI.

Governments often call for digital sovereignty as a means to contain the many risks and harms-such as misinformation and cybercrime-of the digital ecosystem. In truth, though, digital sovereignty is not only insufficient to achieve these goals, but can in fact exacerbate many harms. The scramble for greater state control risks replacing one form of undesirable centralization (corporate) with another (governmental), potentially leading to greater authoritarianism and surveillance, undermining human rights and the global nature of the internet and hurting the economic and other interests of the very countries asserting sovereignty.

Instead of digital sovereignty, consider a different organizing principle: digital agency. While the former seeks to reinforce the primacy of nation-states, digital agency asserts the rights, priorities, and needs not only of sovereign governments but also of the constituent parts-the communities and individuals-they purport to represent. In a multi-layered digital ecosystem, agents exist at all levels: individuals, groups, communities, organizations, corporations, states, alliances of states, and more. Digital agency extends the promise of technology to all levels. As an organizing principle for global digital governance, it has the potential to reinvigorate the multistakeholder governance of the internet's earlier years, reviving the multiplicity, heterogeneity, and inter-sectoral collaboration that were part of the network's original vision.

In this brief, Section I examines how the emphasis on digital sovereignty arose in response to the failed-or at least underfulfilled-promise of the internet, as governance in the digital ecosystem shifted from emphasizing multistakeholder participation to the primacy of national governments. Section II examines the multiple meanings of sovereignty, in particular digital sovereignty, a "'fuzzy' concept" signifying different things to different people. To be effective, policymakers will need to understand and distinguish between its various forms and their risks and advantages. Section III unpacks the many implications of these variants and their policy consequences, intended and unintended.

Section IV unfolds the concept of digital agency: a renewed commitment to multistakeholder governance, though in an enhanced-more inclusive and more responsive-form. As an organizing principle, it allows policymakers to retain the benefits of digital sovereignty while limiting its harms. We identify three cross-cutting principles: multiplicity, technology enhancing, and collaboration, which lead to three action areas for better digital governance.

I. The Rise of Digital Sovereignty

The internet's early years were characterized by a faith in free markets and private enterprise and a determination to keep governments and regulation at bay. During this time, the internet became the vehicle for hopes of self-regulating markets, the end of the nation-state, and a borderless global future.

Among a host of often ill-defined aspirations, two tenets stood out. First, activists and technologists were determined to remove the authority of nation-states from the equation: Digital governance should take place without governments. And second, the internet community sought to develop something it called multistakeholder governance, a more inclusive and participatory model that would bring together voices from a wide range of sectors and geographies. The approach was perhaps best embodied by what internet governance scholar Milton Mueller called the "I* institutions": ICANN (Internet Corporation for Assigned Names and Numbers), tasked with assigning domain names and managing the internet protocol (IP) system, and the IETF (Internet Engineering Task Force), responsible for technical standards and protocols. While not without controversy, these organizations embodied the hope that digital governance might supersede nation-states and include a wider variety of actors.

In 2014, activists, policymakers, and technologists descended on the NetMundial conference in Brazil. In response to revelations by Edward Snowden (and others) about government surveillance, the conference was meant to reemphasize consensus-driven multistakeholder governance as an alternative to state control. A moment of "peak multistakeholderism," in Mueller's words, the dream was short-lived. NetMundial soon devolved into inter-sectoral recriminations; within a few years, the promise of a self-governing internet collapsed into revelations of privacy violations, misinformation, the Cambridge Analytica scandal, Brexit, and Donald Trump's 2016 election.

In the decade since, policymakers and citizens have become increasingly disenchanted with the digital ecosystem-and increasingly convinced of the need for some form of state intervention. Chief among the risks they seek to address is the problem of corporate consolidation (what Tim Wu called the "Curse of Bigness"). In addition, governments have grown concerned about the problem of geographic consolidation and the dominance, perceived and real, of the West, especially the United States.

Digital sovereignty rose to prominence within this context, gaining credibility as a form of state pushback against digital centralization, and more generally as a way of potentially rescuing some of the internet's original promise. Whispers about the need for "cyber-sovereignty" in China began as far back as the early 2000s and gained greater attention when China released its 2017 International Strategy of Cooperation on Cyberspace. A range of jurisdictions-from India to Vietnam to the European Union-now use the concept, albeit with sometimes incongruent interpretations.

Despite lingering questions about its desirability (and feasibility), digital sovereignty is today arguably the dominant trope of global digital governance-a new article of faith among policymakers and other stakeholders that has replaced the earlier belief in multistakeholder governance.

II. The Many Faces of Digital Sovereignty

Digital sovereignty may have recent provenance, but it taps into a much older notion of sovereignty that rose to prominence during the seventeenth century in Europe, amid the Peace of Westphalia. Two principles were at its core: indivisibility (the idea that power rests with one authority, i.e., sovereign governments) and non-interference among nations. Today, when policymakers assert nation-states' fundamental right to control data flows or content dissemination over and within their borders, they are reasserting these principles.

The scope and meaning of sovereignty have long been contested and riddled with ambiguity. For one thing, the notion has proven flexible over time, adapting to new geographies and circumstances (the principle of indivisibility has, for instance, weakened in response to globalization). At its heart exists a core tension between an inward-looking, defensive form of sovereignty, and a more outward-looking, hegemonic version that emphasizes states' rights to cast their authority abroad. These tensions-already apparent in the Peace of Westphalia, which enshrined the principle of non-interference while upholding states' right to intervene in the affairs of others to maintain a balance of power-appear in recent treaties and conventions. The United Nations Charter upholds a principle of "sovereign equality" (Article 2) while allowing for intervention on several grounds (Chapter VII). Contradictions also appear in numerous other documents, notably the Universal Declaration of Human Rights (1948) and the (non-binding) Tallin Manual (2009), which seeks to reinterpret sovereignty in the context of cross-border cybercrime.

These tensions carry over to claims for digital sovereignty; in fact, the internet's borderless nature arguably exacerbates them. As a result, claims to and applications of digital sovereignty are hard to assess. Just what does the exercise of digital sovereignty mean, and what are its likely impacts, for a country, its government, its economy and security, and its people? Is digital sovereignty likely to alleviate or exacerbate the risks of technology? The short answer to these questions is that it depends. A nation's assertion of digital sovereignty means different things in different contexts and can take at least five major forms.

1. Defensive: China's early claims to cyber-sovereignty-such as in its 2010 white paper "The Internet in China" and at the 2014 Wuzhen conference-rely on a primarily defensive version that emphasizes non-interference in countries' internal affairs. Others have taken similar stances: Vietnam in its Law on Cybersecurity upholds a principle of "national sovereignty in cyberspace" to maintain "social order and safety," while Belarus asserts "information sovereignty" as a way to resist Russian influence.
2. Expansionist: More recently, China's approach has turned outward, closely tracking the country's broader geopolitical ambitions-and pointing to longstanding tensions in international law between defensive and hegemonic sovereignty. China's landmark 2017 policy document, "International Strategy of Cooperation on Cyberspace," laid out the country's efforts to expand its interests through initiatives such as the Digital Silk Road. Similarly, Russia's concept of information sovereignty includes the right not only to control domestic information but also to use cyberoperations to influence elections and information systems abroad.
3. Normative: In the European Union, digital sovereignty is arguably less about inter-state contestation than resistance to the financial and normative power wielded by technology companies, especially U.S. Big Tech (though the two motives can overlap). The EU approach is emblematized by its landmark 2016 General Data Protection Regulation (GDPR), which protects consumer privacy and human rights as key goals of the community's distinctive approach to digital sovereignty.
4. Postcolonial: Historically marginalized countries in the global economy, for example in Latin America, Africa, and Asia (sometimes referred to as the Global South or Majority World), have advanced a version of digital sovereignty that seeks to embed contemporary technology policy within the broader arc of history. They emphasize the importance of combatting "digital imperialism" or "AI colonialism" and avoiding dependence on Western companies.
5. AI: AI sovereignty can be considered a fifth variant (or sub-variant) of digital sovereignty. Definitions vary, but it typically focuses on a set of elements critical to the AI stack. These include at a minimum data, chips, compute, and cloud capacity, though sometimes also non-technical elements such as regulatory capacity, digital literacy, electricity, and "workforce development." Examples include Abu Dhabi's efforts to create a company (AI71) to commercialize indigenous large language models (LLMs); Emmanuel Macron's backing of Mistral; America's $50 billion investment to boost domestic chipmaking; and Singapore's SEA-LION LLM project.

III. Disentangling Digital Sovereignty

For policymakers seeking to reassert a measure of control over the digital ecosystem, the varied types of digital sovereignty pose a conundrum. What version of digital sovereignty is likely to prove beneficial and freedom-enhancing, and for whom? How to maximize the positive potential while limiting the harms of an approach that is so ill-defined?

Part of the solution is to disentangle the various meaning of digital sovereignty: to understand their different implications and possible consequences and to help identify ways in which policies based on digital sovereignty may enhance or limit the rights of both governments and of their populations. There are at least three major benefits and four risks.

Key benefits include:

1. National freedom and autonomy: At a time when governments and many others feel increasingly constrained by consolidations of corporate and geographic power, "the ability to have control over your own digital destiny" (in the words of the World Economic Forum) is undeniably appealing. The European Union refers to digital sovereignty as a means of ensuring "strategic autonomy in the digital field," and China declares that "no country should pursue cyber hegemony." While calls for greater freedom can be fronts for other agendas or part of a more complex web of motivations, assertions of digital sovereignty often arise as genuine responses to perceptions of national, regional, or corporate dominance.
2. Global equity: Some of the most compelling claims for digital sovereignty emanate from the Global South. For nations whose marginalization pre-date the digital era, sovereignty is likely to offer a particularly appealing response to concerns over digital colonization or more general exclusion. When American companies use African data to train their LLM models without compensation, it summons memories of the expropriation of minerals and other wealth during colonialism. Some scholars argue that the use of low-paid content moderators in countries like Kenya and the Philippines is a form of "modern slavery." One doesn't have to agree entirely with the analogies to understand their historical and moral weight: Claims to digital sovereignty in the name of greater global equity are not easily dismissed.
3. Local context: Digital sovereignty can lead to digital ecosystems that are more attuned to local needs and priorities. This benefit is likely to appeal to policymakers concerned about the impact of technology on local cultures and societies, as well as those seeking to maximize the potential benefits (while limiting harms) from systems, platforms, or apps. Such concerns have particular salience in the Global South, given a long history of misapplied and often harmful technological interventions in the name of development. The advent of AI adds greater urgency: A lack of diverse and relevant training data can exacerbate AI biases and hallucinations and lead to policy responses that are at best irrelevant and often harmful. The World Bank points to the risks of using data based on European building patterns for disaster relief in Africa; evidence similarly suggests the limitations of using Western road data in Tanzania, where 95 percent of roads are unpaved. In this context, policies based on digital sovereignty are closely linked to outcomes such as economic and social development, disaster relief, and community empowerment.

Despite these benefits, policymakers need to be wary of at least four "sovereignty traps"-unintended consequences that could limit the positive social impact of technology or undermine the broader goals of the very nations seeking sovereignty. These include:

1. Authoritarianism: Some of the most vociferous and earliest calls for digital sovereignty have come from some of the world's most repressive regimes; the rising tide of digital sovereignty has been accompanied by an unprecedented increase in internet shutdowns and restrictions on digital freedoms. The flipside of sovereignty's freedom-enhancing possibility is that it may also enable authoritarianism, repression, and surveillance. These consequences may be intentional in the hands of unsavory governments, but perhaps even more dangerously, well-meaning regimes may unintentionally unleash authoritarian tendencies through policies (such as data localization) that enable intrusive law enforcement. Stakeholders seeking to promote greater freedom and autonomy in the global digital ecosystem would do well to guard against a host of far less desirable outcomes that may lurk alongside the potential upsides of digital sovereignty.
2. Homogenizing and hierarchical: One way of understanding the "double-edged" nature of sovereignty is that calls for freedom and autonomy are being made on behalf of states and governments rather than the populations and communities they encompass. This tension is implicit in contradictions within international law, notably the difficulties of balancing the right to non-intervention against the right to intervene, for example in the name of human rights. This tension is apparent in much contemporary thinking on human rights, which emphasizes the rights of citizens and challenges the notion of national sovereignty as an "absolute shield." For all its claims to freedom and autonomy, sovereignty contains a deeply hierarchical and totalizing strain, one which often homogenizes the needs and priorities of populations and communities. For policymakers seeking to wield digital sovereignty as a means of promoting greater freedom, the question to ask is: Whose freedom?
3. Balkanizing: Digital sovereignty can have balkanizing or fragmentary tendencies; greater national control is often inherently in tension with the borderless nature of the global digital ecosystem. For all its undeniable flaws, the internet is a collaborative project that relies on openness, access, and connectivity across national borders. This does not mean that sovereign control is always unjustified, but there is little doubt that restrictions on cross-border data flows, firewalls, and other domestic checkpoints risk weakening the global public good nature of the internet-in the process, often stunting the very commercial, educational, and other transactions that would otherwise benefit the governments asserting digital sovereignty.
4. Self-undermining: Policies based on digital sovereignty are often self-undermining, working against the interests of the very nations upholding the principle. While countries may feel the need to control domestic data to promote local business and entrepreneurship, restrictions on cross-border data flows and other similar actions may instead harm domestic economies. Susan Ariel Aaronson, for instance, cites evidence to suggest that "AI nationalism" may reduce innovation and competition, encourage the development of domestic monopolies, and distort trade. One recent example comes from Pakistan, where a national firewall erected in the name of sovereignty has instead weakened industry and dampened technical innovation. Calls for national AI models-a core tenet of AI sovereignty-can be similarly self-undermining. In fact, the decision on whether to allocate the tremendous resources required to build domestic AI capacity requires a balancing of national priorities and needs. The choices are especially complex for resource-poor countries from the Global South, but even wealthy countries may have trouble keeping up. Among other risks, over-investing in the chimera of AI sovereignty could undermine more urgent priorities, such as education and health care.

The truth is that those most likely to benefit from a "drawbridge up" approach to digital sovereignty are existing powers that have the technical, financial, or data heft to stand on their own. Smaller countries with fewer resources and less technical capacity face the very real danger of "over-indexing" on sovereignty. Rather than shutting out the broader digital ecosystem in the name of national control, many nations might be better off participating in a revamped, more inclusive, and more participatory global governance system. What might such a system look like?

IV. Toward Digital Agency

Various alternatives to digital sovereignty have been proposed as ways of rethinking technology governance. These include "networked sovereignty," "networked multilateralism," "the digital commons," and "digital self-determination." Building on work by Pablo Chavez, the U.S. State Department has advanced a notion of digital solidarity to bring together like-minded nations with shared democratic values. These approaches differ from digital sovereignty in several respects, but many (not all) share an emphasis on the role of governments and nation-states as primary actors. In contrast, this section explores a non-state-centric approach: digital agency.

Digital agency retains much of what is appealing about sovereignty, particularly its emphasis on freedom and autonomy, especially for the Global South. At the same time, digital agency avoids some of sovereignty's limitations, notably its homogenizing and hierarchical tendencies. At its core, digital agency can be understood as the foundation for a revamped and reinvigorated multistakeholderism. In this sense, it represents an effort to recuperate (while refreshing) some early principles of digital governance, especially a recognition of the importance of diverse, global, and cross-sectoral representation.

Three cross-cutting principles underlie digital agency.

1. Recognize stakeholder multiplicity: Digital agency is fundamentally about recognizing the multitude of stakeholders, at all levels of governance, that make up the digital ecosystem. It acknowledges that different stakeholders have different priorities and values, and it begins from the premise that a healthy digital ecosystem emerges when governance seeks to maximize freedom and autonomy-and other rights-for all levels in the ecosystem. Nation-states are indisputably part of this heterogeneous assortment-major parts of it. While digital agency is at its core a reaffirmation of multistakeholderism, it does not represent a return to cyber-libertarianism, which sought to altogether exclude sovereign governments. Instead, digital agency posits a governance framework in which nation-states participate and exert their authority alongside a multitude of other voices.
2. Use policy to enhance technology: Technology is a force for both social good and potential harm. Understandably, policy has often been directed at limiting harms, but the result has frequently stifled the positive potential of technology by focusing on misuse rather than missed-use. While recognizing and addressing the tremendous harms caused by technology, perhaps especially in the Global South, a digital agency approach seeks to use governance as a tool to harness the positive potential of technology. In particular, digital agency seeks to maximize the technical properties of openness, decentralization, and interconnection to promote inclusion, collaboration, and a revamped multistakeholderism.
3. Prioritize collaboration: One of the chief shortcomings of digital sovereignty is that it seeks to impose artificial borders on an essentially borderless technology. This approach not only works against the underlying potential of technology, but also undermines the interests of the very governments asserting national sovereignty. In contrast, a digital agency approach seeks to enhance inclusion, accountability, and trust through collaboration. It posits a network-of-networks approach to governance, building links and alliances among stakeholders with common interests and complementary capacities. Sometimes these stakeholders may be nation-states; shared interests may be advanced, for instance, by collaborative policymaking in a given region. At other times, the alliances may supersede region-for example, when several grassroots organizations come together to advocate for greater freedom of speech or gender inclusion on the global internet. In all cases, digital agency suggests an approach driven by shared values or desirable outcomes rather than geographic or political boundaries.

Three Action Areas

These three principles provide guidance for a range of actions, policies, and other measures in a revamped global digital governance framework. In particular, three areas of action are crucial, grouped into three layers of governance: policy, technology, and society. Conceiving of technology in this more holistic manner recognizes the way it is embedded into our broader social and governance structures.

Policy Layer

Action Area 1: Reinvent Institutions

Many institutions are by their nature somewhat cumbersome, slow to adapt to changing contexts and circumstances. This is one of the key sources of friction in digital governance; the pace of technological innovation often outstrips institutions' abilities to keep up. To harness technology's potential while minimizing its harms, we need institutions that are more agile and responsive, fit to purpose, and able to accommodate the heterogenous needs of stakeholders. In particular, advancing digital agency requires institutions that can:

• Encourage meaningful participation by a wider range of stakeholders. Civic engagement platforms have had some success, both for inclusion and in building consensus. Policymakers could also explore the creation of deliberative polling and of regional multistakeholder digital councils that seek input on priorities and needs. Encouraging grassroots participation is especially important (and challenging); digital agency is firmly committed to a principle of digital subsidiarity, which emphasizes decision-making at the lowest possible levels of authority.
• Adapt to the rapid pace of technological innovation and be capable of mixing-and-matching to achieve desirable outcomes. Effective policy is flexible and non-doctrinaire; the pursuit of sovereignty should not be absolute. In the case of AI sovereignty, policymakers should aim for a "Jenga AI tech stack"-some elements may promote national control (e.g., data localization), while others may acknowledge the importance of offshore elements (e.g., international LLM models to derive insights from local data).

Technology Layer

Action Area 2: Enable Edge Technologies

Technology-enhancing policy seeks to harness the potential of technology to achieve socially desirable outcomes. Edge technologies, a potent area for such intervention, cover a range of specific applications, including Internet of Things (IoT) sensors for environmental monitoring, mobile phones for remote health care, smart wearables for personalized health insights, and decentralized storage networks for secure data management. Edge-based technologies have particular potential in AI, including through the deployment of model orchestration, hybrid AI models that combine local and cloud-based processing, and federated learning, all of which can also help reduce costs and the need for compute capacity.

All these approaches have the potential to foster inclusivity and multiplicity. In particular, by encouraging low-cost and innovative uses, they can weaken the centralization that has long marked technology and technology governance, helping to bring grassroots and marginalized stakeholders into the conversation.

Social Layer

Action Area 3: Build Human Capacity to Promote Technical Capacity

Digital agency seeks to bring marginalized communities into global governance. We must not, however, be naïve about these communities' ability to participate meaningfully, even if formal structures and institutions are made more inclusive. Meaningful participation requires various forms of capacity. Digital agency seeks to supplement institutional redesign and technological innovation with a commitment to education, awareness-raising, financial support, and digital as well as formal literacy, with a focus on three particular forms of capacity:

• Formal education: Digital literacy is built on traditional literacy and numeracy. Communities without these foundations may have difficulty participating meaningfully in global policy conversations.
• Digital education: Formal education must be supplemented by digital skill-building and literacy to build the specialized knowledge often required in technical policymaking.
• Financial capacity: Financial support for stakeholders, for example, in the form of travel grants and other support mechanisms, is necessary to enable meaningful participation by grassroots voices in national and global policymaking fora.

Conclusion

Since its emergence in the early 2000s, digital sovereignty has evolved from a fringe notion to a core principle at the center of global governance. This stems in large part from nation-states' determined pushback against centralization in the digital ecosystem, but it also reflects technology's increasingly pervasive nature. Digital governance is no longer simply about digital technology. It is an arena of contestation over the very structure of societies, politics, and economics-all core concerns of national governments.

The trend is only likely to intensify as AI extends its tentacles into daily life. As a horizontal technology, AI has the potential to dramatically transform a wide variety of sectors, for better and worse. How AI is governed is likely to be one of the most important questions facing policymakers in coming years. AI governance will have broad ramifications for economic growth, social inclusion, political and cultural polarization, the structure of work, geopolitical contestation-and much, much more.

Understandably, governments and other stakeholders are determined to avoid a repeat of the consolidations of power and the resulting harms that characterized earlier generations of policymaking. Digital sovereignty, now recast as AI sovereignty, seems to provide a solution. Many believe that exerting early and definitive authority over the AI stack could help ensure that this latest round of innovation does a better job of promoting freedom and inclusion, fostering economic growth, and supporting the needs of populations.

AI sovereignty may indeed lead to some of these benefits. But as the above suggests, exertions of digital sovereignty can also backfire, in ways both anticipated and unanticipated. If there is one takeaway from this brief, it is the need for a more granular and nuanced approach: one that balances the rights of sovereign states with those of their populations, that considers various dimensions and levels of freedom, and that acknowledges the differential needs of the various constituent parts of nation-states. Because the AI stack has so many components-AI is multiple technologies, not one-a more granular approach would also recognize that different elements of the stack need different approaches.

To achieve this granularity, we need a better understanding of what digital sovereignty is and what it does. We need to understand where it helps achieve the goals of policymakers and populations and where it falls short. This may require a new organizing principle for digital governance, an approach to national and global policymaking that is more capacious, inclusive, and agile. Digital agency offers such an approach. It suggests an alternative route through the dilemmas and difficult choices posed by technology-and maybe, just maybe, it offers a way to ensure that the coming round of technical innovation serves the needs of all populations rather than just a few and that at least some of the original promise of the internet is at last fulfilled.