Good advice: How online platforms can benefit from social media councils

Most very large online platforms have so far shied away from creating such councils, at least as permanent bodies. The academic discussion has emphasised two variables: designing the convening procedure to be inclusive, and taking a multi-stakeholder approach. In terms of designing councils, two other key aspects are: getting technical experts involved and promoting participation.

The debate is not purely theoretical, however: The EU’s Digital Services Act (DSA) also explicitly mentions councils as regulatory and standardising bodies, as does the coalition agreement adopted by Germany’s government in 2021.

Against this background, we have formulated two hypotheses which we examined in a deep dive taken together with recognised experts:

• If designed appropriately, SMCs can be an effective means of ensuring diversity in a platform’s regulatory and standardisation practices. This is the only way to ensure democratic accountability.
• The larger the SMC, the less effective its decision making will be, even if its legitimacy is increased. It is important to strike a balance in the council’s composition between social diversity and having the necessary expertise, without the council becoming ineffective.

These hypotheses led us to four key questions:

1. How can we ensure there is true democratic accountability to the societies in which SMCs operate? How can we maintain a balance between social change and structural institutionalisation?
2. What about scalability? How many SMCs can or should there be? Should they be national, regional or issue-specific?
3. Who should their members be? When it comes to diversity and expertise, which criteria should determine an SMC’s composition?
4. Which role should SMCs play in enforcing regulations?

2.1 Democratic accountability to society: Between social change and structural institutionalisation

To make the discussion of SMCs more concrete, it is worth taking a look at everyday practice. In 2018, for example, Meta created a so-called Oversight Board. This body is made up of experts from all continents and is structured as a “self-governing unit”. The board makes recommendations to Meta on the content moderation happening on Facebook and Instagram. The recommendations are not binding for the company, however, although the board is also able to make binding decisions. Yet from January 2021 to April 2023, it took only 191 decisions – while in 2022 alone, 1.3 million cases were either submitted to the platform for a decision on content moderation or were forwarded to the board following a complaint by users. This discrepancy in the numbers clearly illustrates that the Oversight Board chooses its cases very carefully, or must choose them carefully, due to limited capacity: Ideally, deciding on an individual case could address larger structural problems.

A number of essential questions arise here about the work done by SMCs: Which factors, for example, can be used to prioritise different issues and how effective can comparatively small numbers of decisions and recommendations actually be? As noted above, the decisions taken by Meta’s Oversight Board are not binding for the company. The board can only make recommendations on everyday practice, on rules for content moderation, or on transparency guidelines and then hope that Meta responds accordingly. This is where the second main challenge facing SMCs becomes clear: whether the decisions they take are seen as legitimate and binding.

In our discussions it also became clear that a fundamental tension exists when it comes to striking a balance between social change – and thus structural, far-reaching decisions – and the institutionalisation of SMCs. Progress is being made, but only slowly – and it often still depends on the platform’s interests or its willingness to act on the SMC’s recommendations.

2.2 Scalability of SMCs: Regional, national, global, issue-specific?

Differences quickly become apparent when the discussion turns to the design and scalability of SMCs. Depending on the issue, needs or platform, opinions vary as to whether SMCs should have a regional or national focus, or even have a global reach in order to ensure democratic accountability on globally-active platforms. At the same time, we are far from achieving consensus on whether SMCs should do their work based on location or issue. Do we need special discussion rounds if the topic is climate disinformation, gender-specific hate speech or health issues? Or is it more important how such debates are embedded in the relevant regional or national context?

• Reasons for creating SMCs: In every instance, the need for establishing an SMC must be well founded – few actors (platforms) are interested in having their power curtailed by external bodies. Some experts cite the right to informational self-determination, saying this requires having a balance of power between the platform and its users, which can only be achieved by creating collective self-governance mechanisms. Others advocate for participation, inclusion and representation in light of the need to safeguard basic human rights. Still others note the limits on market power imposed by competition law and call for participatory formats as a means of according legitimacy to socially dominant players like large social platforms.
• European approach: In the European context, it is conceivable that national parliamentary-mandated SMCs could play an advisory role within the framework set by the DSA. Such a council appears in the current draft legislation for implementing the DSA in Germany in conjunction with the creation of the country’s Digital Services Coordinator (DSC). At the European level, SMCs could thus be deployed for extrajudicial dispute resolution (Art. 21 DSA), to develop codes of conduct for online advertising (Art. 46 DSA) or to ensure access for people with disabilities (Art. 47 DSA). As venues for citizen participation, they could also be used to assess and mitigate risk (Art. 45 DSA). Neither platforms, governments nor auditing firms are suitable candidates for conducting risk assessments, given their respective incentive structures and competences. By representing stakeholders from civil society and academia, SMCs could carry out independent risk assessments and recommend appropriate responses.
• In terms of impact and scalability, it is often emphasised that broad-based SMCs on the regional level can also provide incentives for achieving global changes on platforms. Wide-spread public awareness and media pressure are further levers for integrating decisions by SMCs into platform rules. None of these arguments, however, replaces the need to consider power imbalances on an ongoing basis. Moreover, processes of social change are perceived differently in different regions and by different groups, meaning that values and customs within the global framework will be continually subject to negotiation.
• Expertise instead of a broad base? The concept of expert councils follows a different logic. An overarching council that brings together different expert disciplines requires an institutionally neutral home. The theory is that such an expert body (sometimes referred to as a “consilience council”) could create the knowledge base needed for effective governance. In this case, however, classic citizen participation would play a subordinate role.
• Combining elements: Some experts advocate combining different designs with different target areas in practice. For example, a council focusing on citizen participation and a council consisting of experts could work together in a close-knit process of reflection and iteration to explore how effectiveness, legitimacy and social accountability for public values can best be brought into alignment on digital sites. The goal here would be to design an instrument that is sustainable in terms of human rights and also democratically embedded. Questions about resources and feasibility remain open here as well, however.

In any event, how the trade-off is handled between expert knowledge and participation seems crucial. Experts often work more routinely and can draw on a wealth of knowledge and experience. A balance must be found in the SMC’s composition that reflects both society’s diversity and the required expertise and allows the SMC to work effectively.

2.3. Diversity in SMCs

Other key questions regarding SMCs concern their composition: Which criteria targeting diversity and expertise should apply here? What role do SMCs play when it comes to including marginalised groups and their needs? How can SMCs meaningfully influence the enforcement of regulations?

There is general agreement that local and participatory solutions are essential elements of democratic accountability. Resource-poor regions and socially marginalised groups and issues still receive too little attention in the context of platform governance. The lack of consideration given to local languages, for example, has been shown to result in algorithms filtering content inadequately or incorrectly and to too little access being provided to effective mechanisms of conflict resolution. The phenomena of digital violence and attempts to threaten marginalised groups still generate little substantive response. Participation and diversity are therefore essential, since they are the only way the concept of SMCs can gain legitimacy and have a broad impact.

In the German context, experts in our discussion cited the reform of the country’s broadcasting councils, which have been heavily criticised for their overrepresentation of established political groups at the expense of marginalised civil society groups. According to the critics, the councils have generally reflected the more visible social structures of the 1980s and have changed little since then. The country’s broadcasting councils need to do more in this regard and become more flexible as a result.

2.4. SMCs and regulatory enforcement

The major online platforms influence how we can exercise our basic rights. They create the spaces in which we communicate and they define the rules according to which communication takes place – through their terms of use and their algorithmic recommendation systems. They and their in-house decision making are not directly legitimised by their users, and even less so by the people who do not use the platform. Yet the latter can be impacted just as much by the decisions platforms make. This becomes clear at the latest when a site’s algorithmic design begins promoting social polarisation or even human rights violations or when inadequate moderation in less frequently spoken languages leads to an infringement of human rights. When a platform’s rules have an impact on society as a whole, they should also be subject to regulation.

• Control requires access: To work effectively, SMCs must have access to resources. This fact was strongly emphasised in our discussion with the experts: The minimum requirements here are data generated by social media companies, access to platforms to carry out monitoring and research, and measurements of the impact that interventions have had.
• Faster reaction times: Especially in its role as a forum for coordinating the relevant stakeholders in platform governance, an SMC can be more agile than traditional government authorities. It creates a direct feedback loop to users and other stakeholders, allowing the issues addressed and solutions found to be geared to current challenges. One key prerequisite remains, however: Any recommendations made must be binding for the platform.
• Input from civil society: As an alternative to the options outlined above for integrating SMCs into platform activities, the suggestion was made to involve civil society organisations in analysing the platforms’ systemic risks. The DSA envisages such risk assessments for very large online platforms. This also answers the question of how a global SMC could address aspects unique to specific countries or regions. After all, civil society organisations generally have a high level of expertise and enjoy widespread acceptance in the social contexts in which they operate.
• Technical expertise and independence: In addition to participation, the issue of technical expertise must also be considered. Having in-depth knowledge and understanding of the logic driving platforms and their structures is crucial if solutions are to be developed that are as well-adapted and sustainable as possible. Another precondition is that experts must be free to provide advice independently of the company itself in order to protect the SMC from lobbying attempts and to ensure its public legitimacy. Data protection rights must also be taken into account. It will be interesting to see how the data access rights for researchers laid out in Art. 40 of the DSA are impacted by the competing interests in this area.

The exact composition of each SMC and the relationship between participation and expertise will depend on the chosen priorities and objectives. If an SMC is meant to represent civil society, then participation is important even during the phase when the council is being created. Similarly, basic ethical and structural decisions (e.g. on composition and funding) should not be taken by political or academic elites alone. When it comes to optimising the interplay of rules and algorithms, expert knowledge is indispensable.

2.5 Observations and recommendations for democratic accountability through SMCs

Our discussions with the experts during the workshop and beyond have given us inspiring insights into all four areas covered by our questions. The concepts and contexts pertaining to SMCs are diverse, the arguments about their necessity are quite controversial – yet the discourse is engaged and expedient. Despite all the diversity, it is still possible to provide a number of observations and recommendations for the future of platform governance: namely, on the inherent tension between social change and structural institutionalisation, on scalability, on diversity and the potential for participation, and on options for supporting regulatory enforcement. Based on our analysis, here are the most important recommendations:

• Respond to the broad need for making digital spaces more democratic: Social platforms have created spaces for communication which are organised by the private sector but which impact society as a whole. The greater the influence they have on public discourse and human rights, the greater the degree to which different stakeholders and fundamental elements of the public interest must be included when a platform’s rules and practices are being designed. Various models of SMCs are increasingly finding their way into regulation (e.g. the DSA) and practice (e.g. the Meta Oversight Board).
• Align composition and structure with the goal: The idea behind SMCs is to increase diversity and participation when decisions are made and public discourse is taking shape on private platforms. SMCs made up of experts and/or select users can adopt different roles depending on their composition. More expertise supports issue-specific legitimacy and faster decision making; broader participation by different groups leads to greater democratic legitimacy and to decisions that have an impact even beyond their more limited context. Effectively addressing complex and multifaceted challenges could require cooperation and coordination among the councils’ different levels.
• Learn from examples: Meta’s Oversight Board represents one of the first attempts to open up the decision-making system at a commercial platform to the outside world. Important normative milestones can be set here, in that an ongoing influence is exerted on Meta’s rules and regulations through the strategic selection of cases, for example, or in the form of “collective decisions”. Nothing has yet become evident, however, in terms of speed or long-term impact beyond individual cases.
• • Recognising and preventing threats: While an SMC can lend a platform’s rules and algorithmic practices more legitimacy, certain drawbacks and trade-offs must also be considered. They include the weakening of government regulators, the diversification of responsibilities, a potential “fig-leaf effect” – in which good deeds are showcased even if they have not led to any demonstrable change – and a dominant approach to language rules that does not take local and regional practices into account.
• Consider further models of democratic accountability: If challenges and interests are to be addressed effectively both globally and locally, governance structures and the composition of SMCs must reflect everyone’s needs, including those of non-users. Platform structures and SMCs could conceivably be accompanied by public, even governmental, advisory mechanisms and by strategic consulting from highly engaged civil society organisations.
• Discuss resource requirements honestly: The desire for democratic accountability is widespread. Its implementation, however, requires not only openness on the part of corporate platform operators and expertise when designing the relevant mechanisms, but, above all, adequate resources. Only if incentives are created that lead less privileged groups to participate constructively can such models promote social discourse.

The following are recommended readings on SMCs and related developments. If you have any suggestions to add to the list, please let us know!

• The Platform://Democracy report, which contains the project findings that informed our workshop and impulse paper, was published in May 2023 and offers 35 perspectives on the design of SMCs from researchers from around the world.
• Zur Grundfrage, welchen Einfluss digitale Kommunikationsforen und Plattformen auf demokratische Prozesse, Vertrauen und Zusammenhalt haben, findet sich in Nature eineMetastudie.
• Published in 2021, the paper Die Demokratie plattformfest machen – wie geht das überhaupt? by Matthias C. Kettemann and Martin Fertmann explores the basic concept of SMCs.
• Aviv Ovadya suggests using “citizens’ assemblies” whose members are chosen at random: ‘Platform Democracy’ – a very different way to govern big tech.
• Alicia Wanless examines the CERN Model for Studying the Information Environment – a body promoting cross-disciplinary information research.
• Matthias C. Kettemann takes a closer look: Designing Digital Democracy discusses different approaches to SMCs, deliberative democracy and public-sector initiatives.
• In Musks Willkür Grenzen setzen (November 2022), Wolfgang Schulz explains how complex regulatory structures interact with each other in online communications and the role SMCs can play here.
• Niklas Eder, a key contributor to our June 2023 discussion, examines how systematic risk assessments work by drawing on his experience on the Meta Oversight Board.
• Meta’s Community Forum on the metaverse is summarised here by the Stanford Deliberative Democracy Lab.
• How to randomly select citizens for participatory processes is the topic explored by Christian Huesemann and Stefan Roch here.
• Agora Digitale Transformation aims to open a space that promotes debate on the future of regulation and self-determination in the digital sphere.
• Germany’s draft legislation designed to facilitate implementation of the DSA also envisages an advisory council – even if it will not advise platforms, but public authorities.

Conversation Starters:

• Niklas Eder, Meta Oversight Board
• Dominik Piétron, Humboldt-Universität zu Berlin

Experts:

• Martin Fertmann, Leibniz Institut für Medienforschung (HBI)
• Sabine Frank, YouTube / Google
• Henrike Gudat, Gemeinsame Geschäftsstelle der Medienanstalten
• Bettina Hesse, ver.di Berlin
• Dominik Hierlemann, Bertelsmann Stiftung
• Sandra Hoferichter, EuroDIG Sekretariat
• Friederike Mohrat, Tagesspiegel Background Digitalisierung & KI
• Katrin Ohlmer, .berlin
• Francesca Schmidt, Bundeszentrale für politische Bildung
• Julia Tegeler, Bertelsmann Stiftung

Moderation:

• Matthias C. Kettemann und Katharina Mosene

Translation from German by Tim Schroder.