Paradox of Open Redux

January 23, 2023

Open is both a challenger and an enabler of concentrations of power, we wrote two years ago as we launched Open Future. Today, after many conversations on what we called then the Paradox of Open, and a series of responses to our essay, the statement still rings true. However, we also learned that open activists are divided on the issue of addressing the imbalances of power that concern open ecosystems, technologies, content, and data.

More importantly, there are ways forward that help resolve this paradox. Ones that stay true to the visions of openness defined years ago. Yet, at the same time, harden and protect the information commons against harm and abuses. There are also new narratives that frame the value of openness in new ways, in the context of the digital environment in its current shape.

Addressing the power imbalances present in open ecosystems, and devising means of value creation that take the paradox of open into account, will be key tasks for advocates of open in the 2020s.

An expanding space of copyright activism

In his response, Derek Slater writes that there is a need to expand, and not to switch, the focus of copyright advocacy: “Advocating for openness must account for how circumstances have changed, choosing new strategies and tactics.” So for copyright activists (with whom we identify, and who were one of the key targets of our essay), there is still a need to make sense of how our activism and its stakes could be expanded, beyond the initial scope defined twenty years ago.

This is the lesson that became clearly visible in our AI_Commons initiative as we explored the use of openly licensed photographs of people for training AI systems (more precisely, face recognition technologies). The case has often been portrayed as one of the major challenges to open licensing frameworks in the last decade. Indeed, it is one of the strongest examples of reuse at scale – with 100 million images, or a quarter of all openly licensed photographs, being used in AI training datasets. And at the same time, the potential risks were identified as higher than ever before.

This was no longer the hypothetical risk, so often told, but rarely materializing in reality, of someone reusing a photograph of a person in an offensive advertisement. This was millions of CC-licensed images of people’s faces churned into code tools used for technologies that – according to privacy advocates – should be banned.

As we discussed this issue with open advocates, there was a collective sense that the problems, while real, are not copyright issues. That copyright law, and copyright-based voluntary measures like CC licenses, are not the correct tools for addressing these challenges. But the paradox is that these challenges, although external to the open frameworks, ultimately need to be addressed by their stakeholders: stewards of open licenses and administrators of open platforms. And by all of us who openly share content.

Thus open frameworks face the challenge between maintenance and innovation. There is still a need to maintain and grow the environments, blueprints for which were created twenty years ago. And to secure copyright reforms that will support the commons and the Public domain. As a case in point, a broad Open Access policy for publicly funded research was introduced in the United States only last year. At the same time, we need to innovate and transform these very frameworks. We need to go beyond the well-tested tools of the last two to three decades of building the information commons. And investigate tools that deal with such issues as monopolies and competition, privacy, collective rights of communities, disinformation or harms that emerge from the use of experimental AI technologies.

Our theory of change worked well for years with a focus on copyright-related issues. In advocacy work, this translated into a focus on copyright reform and the promotion of policies supporting open licensing. Digital rights were something done by others. Today, these perspectives increasingly cannot be separated from each other. Focusing on market competition, surveillance or disinformation does not mean shifting the focus of open advocacy. Instead, it is about acknowledging the conditions under which the information commons functions today.

Addressing complexity

Addressing this complexity is a matter of developing new ideas and frameworks. But it is also a practical matter: open organizations and advocates need to acquire new expert knowledge and capacities. New coalitions need to be built, bridging activism silos (as stakes defined previously as distinct begin to mesh). And funders, key actors supporting open ecosystems, need to adapt their strategies as well.

In Europe, advocacy needs to be coordinated across a broad range of policy files that constitute Europe’s “digital constitutionalist turn” and that seek to ensure a balance of power in the digital environment. This turn started with the adoption of the GDPR in 2016, followed by the Copyright Directive in 2019; the recently adopted DMA, DSA and DGA, and the Data Act and AI Act that are still in the works.

Today, there are significant efforts in place to build connections and new perspectives: on openness and climate, openness and sustainability, to name just a few. And at stake is a greater capacity to operate, as advocates, in the complex space of European policies. Yet, there is a need to build even stronger connections between open activism and digital rights activism. And to bridge a difference in perspectives between a digital rights outlook that’s ultimately focused on protection and control of uses and an open mindset that is more generative and also considers productive uses of the same resources.

This is also a matter of redrawing some of the maps that chart the open movement, and the connections that are drawn between its different nodes. For example, traditionally there has been a clear distinction between open content and open source / free software advocates. While the ideology was the same (and the same ideological battles were fought), there was a sense that these were two different communities or networks. Today, this distinction is much less clear, as advocacy on open infrastructures and the narrative of the digital public space brings together builders of open source solutions and stewards of the information commons.

The open internet is a narrative and vision that allows the open movement to properly frame its current goals and address the challenges to open. James Boyle argues that the fight for the open web needs to be seen as a key part of the open movement’s history. There is a need to defend existing principles — such as net neutrality — that have prevented capture at the network layer. At the same time, there is a need to advocate for new protections that deal with the consolidation of power at the services layer. To protect the open internet, we need to double down on open standards and interoperability mandates – on policy issues that add one more layer of complexity to open activism. And higher up in the internet stack, at the services level, there is, in turn, the need for direct public investment into public digital infrastructures designed with societal objectives in mind.

There have been debates for years about the importance of infrastructure. In the field of Open Access, this is the issue of institutional repositories. In Open Data, public data repositories are the key tool of institutional change driving this paradigm. In Open Culture, key developments were around new platforms for digital heritage, both public and private (but with commons-based aspects). Yet, the current focus on infrastructures, and the spaces that they enable, is unprecedented. As a policy topic, infrastructures will be front and center for open movement advocates. The broad digital public space narrative is one that connects openness with other causes, as it aims to address power imbalances through advocacy based on the idea of public space.

For a stronger commons

In our own work over the last two years, we have increasingly considered how stronger forms of commons governance can be layered on top of what we have been calling (after Balázs Bodó) Open Access Commons. Again, the term “layering” is important here. There is space for the “basic” open ordering, which should be applied, for example, to a lot of public, non-personal data or to publicly funded media content. But there is also a need for other commons-based arrangements in cases where the basic open model does not suffice.

This is the lesson that we have learned from our Data Commons advocacy in the last year, as we have engaged in policy debates on different files in the European Data Strategy. Data governance issues are fascinating precisely for this reason: because in this field, unlike in many others, different commons-based models are being explored. We have ourselves created a Data Commons framework, which helps design governance models that still feel “open,” and serves to attain a balance that secures both individual rights and collective public interest. We need to design the gatekeepers of open, even if that sounds like another paradox.

This can be seen as an outcome of a decade of conceptual and activist work that explored and promoted ideas such as data trusts, data cooperatives and other types of data intermediaries. And all of these approaches share the common denominator: they aim to address the Paradox of Open. Public policy plays a key role in making these ideas a reality. The decisions made by European policymakers created an opening for securing these new data institutions through law and public policy (in policy files like the Data Act or the European Health Data Space).

Yet the legislative discussion about the Data Act has also shown that our ideas about commons-based, public-interest data governance are difficult to reconcile with the political realities currently dominating discussions about data. With strong opposition from both the right – which perceives sharing requirements as government interference with private property — and the left — which is opposed to data sharing on data protection grounds — there is little space for public access to privately held data (which we defined as a core policy measure furthering commons-based data governance).

This results in a failure to define data as a commons and ultimately weakens the public sector. As a result, the very institutions underpinning democratic and civic societies are cut off from access to data – while such access is increasingly becoming a precondition for evidence-based policy making and effective service delivery. We are still hopeful that commons-based data institutions will be secured through sectoral rules for different European data spaces.

AI governance debate and opportunities to reimagine Open

Artificial Intelligence regulation and policies are another space, in which important debates about openness are playing out. In the last years, advocacy on these issues has been driven by digital rights concerns – the European debate on the AI Act being a case in point.

Last year it became clear that this debate is one that the Open Movement cannot avoid. 2022 was the year when important AI models were released openly. But also one in which a new approach to open licensing aimed to reconcile a traditional open source approach with concern for responsible uses of AI. Finally, policymakers considered regulating open source AI systems, showing that open advocacy needs to consider how bottom-up norms interplay with top-down regulation. Looking more broadly, openness is increasingly seen in the field of AI as a measure that can not just ensure generativity and productivity gains, but also address social justice concerns.

Conversations around openness in the field of AI once again force open advocates to address the greater complexity of digital ecosystems. In order to open up the AI stack, one needs to think across differences between, and the consequences of, the sharing of data and content, code and models.

And recent developments in the field of AI and ML create an opportunity to bring the idea of commons-based governance to this field. In the coming year, questions around the governance of AI datasets and models, inputs and outputs, will be one of the key conversations on the open agenda.

This is a conversation that does not concern just the field of Open AI, no matter how groundbreaking AI-powered systems will turn out to be. New frameworks for licensing, attempts to balance openness with other social values, and ways of limiting power concentrations, will help address the Paradox of Open across digital ecosystems, and the different fields of open.

In all of the fields of open, there are people re-imagining openness and thinking about new strategies and shared advocacy goals. We will be releasing soon a mapping of these fields, followed by results of a survey of open movement leaders, regarding the future of open.

Alek Tarkowski
Paul Keller