In April 2022, the United States, the European Union and other partners launched a Declaration for the Future of the Internet, which articulates a vision for the internet that is open, widely accessible, commercially vibrant and protective of human rights. This vision has been expressed by various states and civil society organizations, and has been reflected in policy documents of different international forums over the last 20 years, such as WSIS, NetMundial, the G7, and the OECD. Internet openness is frequently opposed to the phenomenon of fragmentation, also referred to as the "splinternet" – the gradual fragmentation of the global internet into local sub-networks, each with their own protocols and subject to different domestic regulation. Fragmentation, in turn, tends to be associated with notions of digital sovereignty and national interests. As noted by Milton Mueller: following his recent participation in the 2022 Internet Governance Forum, held in Addis Ababa:
“[d]igital sovereignty, we are now beginning to realize, is essentially the same thing as fragmentation. It is about the de-coupling of the digital political economy due to a reassertion of power exclusivities by nation-states.”
The concern with internet fragmentation no doubt stems from a proliferation of domestic internet regulation initiatives in recent years – from domestic content moderation requirements (UK, Germany, Australia), to country-specific delisting requirements pursuant to the GDPR’s “right to be forgotten”, to bans of certain applications, to more extreme measures such as China’s “great firewall”, wide-scale internet shutdowns, de-connection from the global internet and full internet censorship. Admittedly, one cannot equate delisting obligations for privacy purposes to wide-scale internet shutdowns. However, these policies and domestic court decisions can result in differences in content, user experience and internet availability between jurisdictions. Taken collectively, they result in an internet that is, generally speaking, more fragmented.
- Reframing the dilemma
What, then, should be done? As the 2022 IGF discussions showed, the dilemma does not lend itself to simple solutions. Part of this has to do with diverging views on what fragmentation actually means and which types of fragmentation are most problematic (other than the obvious case of wide-scale internet shutdowns). Different efforts have been taken to address fragmentation. For example:
-the IGF's Policy Network on Internet Fragmentation, as well as the Internet Society, have been producing detailed analyses of the different components of fragmentation (fragmentation at the level of user experience vs. technical layer), providing granularity and context in the discussions, and proposing some red lines;
-Internet & Jurisdiction has produced, in coordination with governments and industry, a set of toolkits to address jurisdiction issues with respect to cross-border access to evidence, DNS abuse and IP infringement;
-the ongoing work of standards organizations like IETF and W3C remains crucial in protecting global internet interoperability.
Taken as a whole, these reports suggest that, while not all forms of fragmentation necessarily have significant negative effects, some types of fragmentation, particularly fragmentation of internet governance and coordination, fragmentation of the user experience and fragmentation of the internet's technical layer, could pose a risk to the internet as we know it.
To address these concerns, we would like to suggest an additional, complementary avenue which, to the best of our knowledge, has not yet been explored: we propose to encourage the multi-stakeholder community to adopt an "experimentation approach" in developing internet governance policy. This suggestion has its roots in a regime of governance that predates the internet's "multi-stakeholder model" by hundreds of years: federalism. In the 1932 case of New State Ice Co. v. Liebmann, US Supreme Court justice Louis Brandeis famously referred to states within the United States as "laboratories" for democracy:
“It is one of the happy incidents of the federal system that a single courageous state may, if its citizens choose, serve as a laboratory; and try novel social and economic experiments without risk to the rest of the country.”
Under this framing, the federative model of governance empowers states to each adopt different policies and laws to tackle similar issues within their competencies. For this model to work as efficiently as possible, there must be a free flow of information, so that other states can learn from each other’s successes and failures, replicate successful policies and avoid common pitfalls. To be sure, the theoretical model has its practical limitations. One scholar has argued that it if anything, it produces a “race to the middle”, in which states, rather than innovate, are content to replicate generally satisfactory policies produced by others. Perhaps this is so in certain cases, but the idea that states within a federation can act as laboratories for new policies and regulation remains relevant to date, drawing studies in fields such as corporate law, environmental protection and Covid-19 pandemic response. Informed by the “laboratories of democracy” metaphor, one think tank, New America, has published a website tracking and comparing state reforms on various issues. What these studies show is that state innovation does occur to varying degrees, resulting at least sometimes in regulatory convergence .
While the model has emerged in the context of discussions on state powers within a federation, we believe it can also be relevant in framing and addressing questions of international internet governance. We do not suggest that policy experimentation among countries will, by itself, foster a bold new age of internet policy innovation. Rather, our modest claim is that embracing an experimentation approach at the international level can create new pathways for the multi-stakeholder community to address some of the issues – and specifically, fragmentation – around which a general consensus has not yet emerged. In some cases, successful policies adopted in one jurisdiction could be replicated by other jurisdictions, leading to a diffusion of norms in a bottom-up manner, thus contributing to a harmonization of approaches over time, across the globe. This, in turn, could result in a more optimal balance between internet openness and the preservation of domestic interests, such as consumer protection and privacy protection obligations, regulations on unlawful content, and content moderation requirements.
- Advantages (and disadvantages) of the experimentation approach
The field of internet governance, like other technology-intensive fields of activity such as AI, cybersecurity and blockchain, is highly dynamic. Every few years, we witness successive waves of disruptive technologies and new social phenomena that harness these technologies in unexpected ways. Geopolitical forces both influence, and are influenced by, the deployment of disruptive technologies. As a result, it has been difficult for the multi-stakeholder community to develop long-term, future proof policies to address fragmentation challenges.
An experimentation-based approach can help address global phenomena characterized by dynamic changes, by fostering a long-term yet flexible approach to policy development. Under such an approach, States can dare to diverge from conventional wisdom and adopt policies that reflect their domestic interests and cultural sensitivities. This allows the international community to “incubate” original approaches, see how they pan out and then decide whether any particular approach should be scaled up and turned into international policy recommendations, norms or rules. Thus, harnessing state experiments can allow for more responsiveness to the policy needs of an ever-changing and diverse digital world.
This approach is obviously not a panacea, and it has some drawbacks. The notion of experimentation at the international level in the context of internet governance has not been formally studied or tested (as far as we know). Importantly, the idea of conducting international policy experiments might need to be accompanied by a normative framework. Intuitively, it is apparent that regulatory experimentation is not necessarily appropriate for every situation, for example when urgent, coordinated action is needed at the international level, or when core human rights need to be protected. In addition, we are aware of the fact that the digital space is not a standalone sphere of activity but rather, forms part of the global market. As such, the model’s practical utility will no doubt be limited by the different forces at play, notably geopolitical forces and by economic and technological considerations. These limitations should not per se prevent states and international organizations from applying the model, but they must be borne in mind.
In light of the above, we propose some precautionary measures:
Firstly, there should be a good, fact-based understanding of the problems with fragmentation. A good starting point is the paper prepared by William Drake, Vint Cerf and Wolfgang Kleinwächter for the World Economic Forum, which identifies three broad categories of internet fragmentation:
“Technical Fragmentation: conditions in the underlying infrastructure that impede the ability of systems to fully interoperate and exchange data packets and of the Internet to function consistently at all end points.
Governmental Fragmentation: Government policies and actions that constrain or prevent certain uses of the Internet to create, distribute, or access information resources.
Commercial Fragmentation: Business practices that constrain or prevent certain uses of the Internet to create, distribute, or access information resources.”
Technical fragmentation can directly impact the very foundations of the internet, the so-called “infrastructure” and “logical” layers, whereas government fragmentation tends to relate to the “social layer” of interactions (B2B, B2C, G2C, etc.). When government fragmentation is highly localized and has limited spillover effects, it is unlikely to raise significant concerns for global internet openness. In our view, domestic initiatives that reflect local values, priorities and sensitivities, and that have a minor fragmentary impact on the global internet, should be accommodated and positively acknowledged in the international discourse. For example, price disclosure regulations in consumer protection law do not affect the global internet in the same way as global delisting requirements do. The experimentation approach provides the space for different domestic initiatives to take place (within some limitations, which we discuss below).
Secondly, there needs to be broad common understanding of the shared values and rules that underlie any experimentation that a country undertakes. One baseline is a commitment to human rights and the rule of law. That is not to say that policy experiments that affect human rights in some way cannot be conducted. Indeed, many internet governance questions can impact human rights such as privacy and freedom of expression, and it is often necessary for states to balance between competing interests in order to achieve legitimate policy goals. However, the experimentation approach that we suggest must not be construed as a general license for untethered policies. Another shared value is a commitment to the overall stability, resilience and global nature of the internet, such that domestic fragmentary policies that have broad impact on how the internet operates as whole – its fundamental architecture, its general openness and its wide accessibility – would not, generally speaking, be characterized as mere “policy experiments”.
Thirdly, "experimentation by design": the multi-stakeholder community must be able to learn from both the successes and failures of states in their experiments. For this to occur, mechanisms should be developed within organizations like ICANN, IETF, IGF etc., so that states can share good practices as well as practices that did not succeed as planned, in an open and receptive environment.
Fourthly, the model we are suggesting must be subject to ongoing scrutiny, in order to ensure that it is working as intended. This could be done through committees within the various organizations, tasked with collecting information, processing it, and issuing non-binding recommendations to ensure a proper flow of information between participants.
- Examples of international regulatory experimentation
In practice, states can and do experiment with different policy and regulatory approaches, in a wide array in all spheres of activity. Take for instance, the development of international investment law. While admittedly different from the framework of internet governance, it nonetheless enables useful comparisons. In the absence of a global multilateral treaty that regulates the treatment of foreign direct investments, international investment law is mainly based on national investment legislation, international investment agreements (notably, bilateral investment treaties), and ad-hoc investment contracts. Accordingly, changes in international investment policies tend to occur from the bottom up: national laws and investment agreements, taken together, reveal global trends. In this environment, states retain some flexibility in adapting domestic policies and bilateral agreements to changing circumstances. As the figure below, produced by UNCTAD, illustrates, states have responded to different security, economic and other changes, by adjusting the level of regulatory restrictions in their investment policies over the years.
States are also experimenting with different policies in the technological sphere. For example, with respect to the regulation of the cybersecurity of critical infrastructure, the EU’s NIS2 Directive, the US Cyber Incident Reporting for Critical Infrastructure Act, and the Australian Security Legislation Amendment (Critical Infrastructure Protection) Act all recently came into effect. The UK is also currently working on new measures to protect its critical digital services. It remains to be seen whether a given approach will emerge as the “most appropriate”, but in the meantime, it is interesting to see how countries are tackling the same issue in different ways.
Arguably, experimentation is also taking place with respect to domestic privacy regulation. Following the adoption of the General Data Protection Regulation (GDPR) by the EU in 2016, several jurisdictions updated their data protection legislation, notably, California, New Zealand, Israel, and South Korea.
It should be noted that not all cases of international policy convergence necessarily stem from one country’s deliberate desire to replicate another country's policies. In some cases, it can be attributed to the “Brussels Effect”, a term coined by Anu Bradford to describe the EU’s regulatory influence across the globe. Theodore Christakis has shown that this influence is apparent in many areas of digital policy. Another related phenomenon that partially explains the diffusion of norms across jurisdictions is “norm entrepreneurship”, whereby States (and the EU) actively use their influence to promote the development of certain norms in other countries and international organizations. Even so, these phenomena do not necessarily account for all instances of norm diffusion.
State initiatives in the regulation of artificial intelligence are also instructive. In 2019, the OECD published a set of non-binding recommendations for trustworthy AI, which may be characterized as a consensus framework for AI policy. These recommendations have influenced the development of AI policies in OECD countries. For example, the EU and the Council of Europe's working definition of "artificial intelligence" borrows from the OECD's (see here and here). AI policy papers from the UK, Japan and Israel explicitly refer to the OECD principles and track them very closely. At the same time, there are several differences in terms of the breadth and depth of proposed regulations. For example, Canada has published an AI bill that applies to high-risk private sector uses of AI, but does not contain extensive regulatory requirements like those of the EU’s AI Act. The White House published a Blueprint for an AI Bill of Rights that is similar to the OECD principles but is not legally binding. The OECD thus created a common baseline and taxonomy which jurisdictions have applied and built upon, albeit in different ways. The OECD also established an AI Policy Observatory, which fosters the kind of policy comparisons that we suggest in the context of internet governance.
However, save for the OECD AI Policy Observatory, there seems to be a disconnect between regulatory experimentation happening in practice, and global policy development mechanisms. What we are proposing is to harness regulatory experimentation by deliberately integrating it into global internet governance discussions.
- International instruments to the service of the experimentation approach
As noted above, the need for global policy experimentation stems from a tension between international harmonization and domestic interests, between the universal and the particular. This tension is not unique to internet governance. Indeed, international policy-making efforts in other spheres of activity are generally based on the same dialectic, and different types of instruments and mechanisms are used to achieve the desired balance. These include international declarations, policy statements and non-binding resolutions, which set broad policy goals and call for certain actions or outcomes, without creating legally binding obligations. In the field of internet governance, the Declaration for the Future of the Internet isa recent example. By avoiding binding language, such “soft law” instruments provide states with flexibility to conduct policy experiments. Another example is the work done by Internet & Jurisdiction: the toolkits it developed demonstrate how learning from existing practices in a bottom-up manner can lead to broad harmonization, without having to impose stringent top-down rules.
The experimentation approach can also be proactively enabled or encouraged through international instruments. This can be useful where states cannot agree on the exact norms or obligations applicable in a certain domain. In such cases, multilateral forums can foster experimentation by providing a common baseline and allowing states to determine their own commitments. For instance, in the field of climate change, the Nationally Determined Contributions (NDCs) regime, established under the Paris Agreement, consists of targets that each state sets for itself, in order to cut emissions and adapt to climate impacts. These targets are monitored and discussed in international conferences. This principles-based approach creates a defined and agreed upon international framework for mitigating greenhouse gas emissions, which allows states to decide for themselves how to meet the goals of the agreement while considering other factors, such as ending poverty. While we do not suggest that a binding treaty dealing with internet governance should be the outcome of the experimentation approach, a similar concept is at play here: internationally accepted baseline principles and frameworks coupled with a mindset that allows flexibility and adaptability. At the same time, one key issue that will need to be born in mind is how to promote mechanisms so that the end-result is a race to the top.
Turning again to the example of international investment law, we note the emergence of diversified international investment agreements as a critical tool in shaping the principles of international investment law, in the absence of a global multilateral treaty. This allows states to reach common ground on certain issues that are still being debated in multilateral forums, as illustrated by the table below, prepared by UNCTAD.
Another example of policy flexibility in international decision making is the use of reservations in multinational treaties. When such a treaty is being discussed, there is often a tension between the desire to make the treaty as coherent and uniform as possible, which entails having fixed provisions that cannot be derogated from, and as global as possible, which entails relaxing some of the treaty’s provisions. Reservations allow countries to join a convention and assume the main obligations imposed by the treaty, while leaving themselves some freedom to diverge from a particular provision (so long as the reservation is not incompatible with the object and purpose of the treaty). For example, the Budapest Convention on Cybercrime allows countries to make reservations with respect to criminalization of intellectual property infringements, subject to certain conditions. When aspiring to achieve a global consensus on a certain topic, reservations in a treaty can play a crucial role in enabling global accession.
The above examples illustrate how the international system is replete with mechanisms that provide states with flexibility. Such flexibility can be harnessed to promote an experimentation mindset, which would empower states to act as policy “laboratories” while remaining tied to an overarching, common set of global policy goals. How can these mechanisms be used to address the openness-fragmentation debate?
- A proposed way forward
Applying the experimentation approach to global internet governance policy requires a multi-layered perspective. In its most basic form, the experimentation approach requires international forums dealing with internet governance to be both passive and active. They should:
-refrain from imposing overly prescriptive obligations on issues that should be tested out;
-actively encourage sharing good practices and those that fell short of expectations, successes and failures, so that states can learn from each other’s policy experiments;
-monitor developments over time to determine which issues and approaches are ripe for scaling up at the global scale, whether as binding obligations, common political agendas, technical standards or otherwise.
This approach requires active support by states, to design agendas that allow for open discussions to take place, and to integrate discussions on the topic into the workflows of international organizations like the OECD, the UN, the ITU, as well as standards organizations like the IETF, IEEE and W3C. In our view, the IGF should build upon the central role it already has, by expanding the activities of the Policy Network on Internet Fragmentation so that it can become a platform for the types of exchanges we are suggesting. For example, the Policy Network or a new group within the OECD could establish a general framework for internet fragmentation testing and analysis. It could select one of several topics in which the question is particularly acute (.e.g content removal requirements), invite states to share their policy and experience – both negative and positive – and identify some common denominators. Alternatively, they could start by choosing a topic that is not overly controversial, the “low-hanging fruits”. They could follow up on the outcomes of the different policies at a later stage, after enough information has been compiled, and then trigger a global discussion based on such information. Global policy recommendations would be informed by data-driven, systematic research, resulting in a consensus position regarding a given issue (e.g. which forms of content moderation requirements are generally permissible).
Another aspect of the approach entails sponsoring solid academic research on the basis of this information, and translating the findings of this research into actionable policy recommendations.
We are aware of the limitations of this "international experimentation" model for internet governance. For one thing, it has not been formally tested, and it will surely need to be refined. Given the red lines we have proposed – specifically, respect for human rights and preserving basic interoperability of the internet – it might be the case that practical use cases of experimentation are limited. It will also be important to examine whether and under what conditions experimentation can lead to a race to the top, the bottom, or somewhere in the middle. International organizations might need to experiment with the experimental model itself, to cautiously introduce it in their discussions and methods of work, monitor how it unfolds and measure its impact. It would be interesting also to explore possible linkages with another governance approach in public administration and international law: the principle of "subsidiarity", according to which " central authorities should perform only those tasks which cannot be performed effectively at a more intermediate or local level." We suspect this framing might also be useful for internet governance as well.
The proposal we make is not a paradigm shift. It does not seek to displace the current axioms and broad understandings of global internet governance policy. Instead, it seeks to introduce into the discussion a new framing that can complement current global discussions. Still, even if used in a more limited fashion among broadly like-minded countries, the experimentation approach could prove to be useful, even beyond internet governance. For example, the OECD’s AI Policy Observatory and its annual blockchain policy conference, could incorporate an "experimentation by design" component to the discussions. Going forward, as the metaverse and quantum technologies begin to take an increasingly international dimension, some of the global governance challenges associated with these technologies might also be addressed by refraining from imposing top-down rules and observing the different approaches taken by states.
Adv. Cedric Yehuda Sabbah is a Director in the Department of Emerging Technologies, at the Office of the Deputy Attorney General (International Law), Ministry of Justice, Israel. Adv. Sabbah is also a Ph.D. candidate at the Hebrew University Faculty of Law.
Adv. Mirit Sharabi Holds an LLM in International Dispute Settlement and Arbitration from Leiden University.
This blogpost was written in the author’s personal capacity and does not represent the views of the Ministry of Justice or the government of Israel.
