Paper

Towards Platform Democracy: Policymaking Beyond Corporate CEOs and Partisan Pressure

| Oct. 18, 2021

Facebook, YouTube, and other platforms make incredibly impactful decisions about the speech of billions. Right now, those decisions are primarily in the hands of corporate CEO’s—and heavily influenced by pressure from partisan and authoritarian governments aiming to entrench their own power.

We propose an alternative: platform democracy. In the past decade, a new suite of democratic processes have been shown to be surprisingly effective at navigating challenging and controversial issues, from nuclear power policy in South Korea to abortion in Ireland. These processes have been tested around the world, overcome the pitfalls of elections and referendums, and can work at platform scale. They enable the creation of independent ‘people’s mandates’ for platform policies—something invaluable for the impacted populations, the governments which are constitutionally unable to act on speech, and even the platforms themselves.

Executive Summary

The Challenge: Who decides? (On divisive platform policies)

  • Complex policy issues: Online platforms must make policy decisions around controversial issues such as content moderation, political advertising, recommendations, and privacy.
  • Deciders often compromised: Currently, either platform CEOs (and their teams) ultimately determine platform policy or powerful governments do; often neither is rewarded by serving the public.
  • Negligible public mandate: The public is continually impacted by these decisions and cares about their downstream outcomes (e.g. censorship, misinformation, violence, surveillance), but their perspectives are rarely incorporated (beyond one-sided studies).
  • Platforms are stuck: Even platform CEOs often don’t want to be held responsible for these decisions—there may be no action that ‘looks good’ or which can forestall retaliation from partisan politicians or governments.
  • No obvious alternative: Even within functional democracies, governments are often limited constitutionally or by partisan gridlock. Platform-based referendums have been attempted, but had negligible response rates. 

The Context: New democratic mechanisms have handled tough issues at national scale.

  • New democratic decision-making processes have now been shown to make thoughtful decisions and be broadly trusted1 , without most of the damaging political dynamics of referendums and elections, and for a tiny fraction of the cost.
  • When designed well, these processes can work even when no existing powerful actor is trustworthy and when no one wants to be held responsible for a decision.
  • They often involve creating a demographically representative “mini-public” that is compensated for a fixed time period to learn about an issue from the many multi-stakeholder perspectives, deliberate together, and voice their conclusions.
  • This may seem idealistic and implausible. But these new “representative deliberation processes” have now been used to support complex policy-making around the world, tackling issues from abortion in Ireland to nuclear power in South Korea. 

The Opportunity: Platforms can use these processes to tackle controversial issues.

  • Platforms working with governments, civil society, can have experienced and neutral facilitators deploy these new processes for the toughest policy questions
  • Policy decisions will then be made by the impacted populations and informed by key stakeholders, often leading to a strong public mandate (which may even help defend against partisan or authoritarian overreach).

Introduction: ‘Who decides?’ 

Someone has to make the hard decisions for what platforms like Facebook and YouTube can do: 

  • Should political ads be allowed, and if so, with what limitations?
  • What should recommendation engines reward?
  • How should platforms weigh privacy vs. safety?

These are many such platform policy questions—and we are stuck within an endless argument around ‘who decides’ the answers. Should corporations and their ostensibly unaccountable CEOs decide? Or should it be governments with their partisan, special interest, and sometimes even authoritarian motivations?

We propose a third way: Platform Democracy. Governing platforms democratically.

Of course, this sounds hopelessly naïve. Clearly, Tim Cook, Mark Zuckerburg and their CEO colleagues won’t be giving up their seats for a “platform president” anytime soon. Moreover, the last time someone tried to run a platform referendum it was a useless failure.23

However, thankfully, elections and referendums are not the only forms of democracy. In the past decade especially, countries all over the world have begun experimenting with new(ish) forms of democracy—and it turns out that some work very well. In fact, they work especially well for the messy issues that are hyperpolarized or where no powerful actor can really be trusted.

As a concrete example, consider how France deployed one of these new democratic processes, the “Citizens’ Assembly”, to develop new climate policy in the wake of the Yellow Vest protests.4 Similar methods have been used by Ireland to untangle political deadlocks on abortion5 and in South Korea to decide on nuclear power policy.6 In each case, a specific highly controversial and complex set of issues was tackled over a fixed time period by a “mini-public”—a statistically representative group of hundreds of ordinary people who were trained by neutral facilitators and briefed by stakeholders. The result of the processes were broad mandates across the impacted population.7  

While one can imagine many potential issues with such processes, we now have significant practical and empirical evidence that these issues can be overcome, as we describe below. These new democratic processes are not a panacea, but they can enable platforms to take some of their most challenging issues to their ‘governed’ populations—and this can benefit those populations, the platforms, and functional democratic governments (while perhaps even weakening the mandates of authoritarian governments).

Many purported fixes to platform problems involve giving individual users more options; e.g. if you want to see less sensationalist and divisive content, a platform might let you tweak your personal recommendations. But this individual agency “solution” does not solve the collective problems that the sensationalism and divisiveness might cause for a community, nation, or the planet—and it could even make those problems worse.8 Platform democracy provides a way forward—democratic processes and legitimacy for exerting the collective agency necessary to address such thorny problems.

Contributions

In this working paper, we focus primarily on adapting the citizen assembly process to platform democracy—the platform assembly; later work will touch on alternative and complementary processes in more detail. We first provide a brief overview of the democratic rationale behind these governance processes, and citizens’ assemblies in particular. We then summarize the standard process and provide a concrete example of what a “platform assembly” involving Facebook and political advertisements might look like. Finally, we provide a supplement with answers to frequently asked questions from stakeholders around platform democracy and the platform assembly process we focus on here.

Democracy beyond elections—and why platforms might care

Democracy is generally associated with elections and elected representatives, but this is a means, not an end. The essence of democracy and the strength of its mandate arguably relies on the deeper ideals of representation and deliberation. In fact, existing implementations of electoral democracy may lose trust as people see them diverging from these ideals.

As an alternative, modern democratic processes9 such as citizens’ assemblies explore different mechanisms for ensuring representation and deliberation, addressing many of the pitfalls of electoral democracy (e.g. poor knowledge of the issues, low engagement, politicization, little incentive to understand others perspectives, destabilizing electoral campaigns, lobbying, etc.). Instead they rely on the hypothesis that if one takes a representative sample of a population, and provides them compensation, training, expertise, facilitation, and stakeholder perspectives, they can make good decisions that provide a broad mandate. Moreover, unlike electoral processes which often incentivize polarization, deliberative processes can support common ground. The significant successes of these approaches around the world10 have helped validate these hypotheses—even though it is surprising to many that people chosen by lottery are sufficiently capable (in the right environment) such that they can be key parts of a policy-making process.

Why this might actually happen

Importantly for the purposes of effective (and pragmatically feasible) platform democracy, many of these processes are difficult to game, do not require unsustainable levels of public engagement, are comparatively inexpensive, are rapidly implementable, are easily iterable, and work at any scale. 

Citizens’ Assembly Primer

While there are a number of new processes that hold promise, for simplicity and concreteness we focus here on citizens’ assemblies.11

There are three core phases to most existing Citizens’ Assemblies:

  1. Lottery selection of participants to reflect the variety of backgrounds and experiences found within a population (via statistical sampling). Participants are offered compensation for their time (and ideally other context-specific support to lower the burden of participation, such as childcare, travel, and connectivity support if requested). 
  2. A learning phase where participants are brought up to speed on the topics (e.g. political ads) they will be exploring from different stakeholder groups, and where they get to know each other through discussing the material and the process in small facilitated discussion groups. {Participant time: 1-3 weekends}
  3. A deliberation phase where participants discuss, consider, and debate in the facilitated groups, explore potential proposals, and come to decisions (for example, via voting across the proposals). {Participant time: 1+ weekends}

The assembly acts as a mini-public—it has the same demographics as the population it is convened to serve due to representative sampling, so most of that broader population can identify with a participant who has had a similar life experience or background to them (compared to e.g. politicians who generally have a narrower set of life experiences and demographics). Throughout the assembly process, plenary hearings are conducted with stakeholders, experts, and assembly participants and are broadcast publicly. These combine to bring the broader public along on the same journey as the assembly, supporting a greater understanding of and a stronger mandate for the decisions (small group deliberations remain private). 

The evidence base for the effectiveness of deliberative approaches 

Most people are understandably extremely skeptical when first presented with the concept of the citizen assembly and related representative deliberative processes. While it is out of scope in this whitepaper to address each of these critiques in detail, and how they are overcome in practice, we provide some excerpts from a recent paper in Science, each citing additional research from deliberative processes and experiments around the world:

  • “Evidence from places such as Colombia, Belgium, Northern Ireland, and Bosnia shows that properly structured deliberation can promote recognition, understanding, and learning”12
  • “Ordinary people are capable of high-quality deliberation, especially when deliberative processes are well-arranged [...] Analysis of the transnational ‘Europolis’ deliberative process—a demanding multilingual setting—found that ‘the standards of classic deliberation are far from being utopian standards that only very few citizen deliberators can achieve’”13
  • “Ordinary people thinking together can see through elite manipulation of symbolic political appeals”14
  • “Deliberation can overcome polarization. The communicative echo chambers that intensify cultural cognition, identity reaffirmation, and polarization do not operate in deliberative conditions, even in groups of like-minded partisans.”15
  • “Deliberation leads to judgments to become more considered and more consistent with values that individuals find that they hold after reflection.”16

The Platform Assembly: An example with Facebook political ads in the United States

Initial implementation of a platform assembly would likely be focused on a regional or nation-specific issue, in a location with a single dominant language, and where the platform has deep regional knowledge.17 The lessons learned can thereafter be applied to broader and more diverse assemblies; including fully global assemblies appropriate for a global platform. 

Here is an example of what the end-to-end process might look like for a citizens’ assembly on political advertisements in the United States on Facebook:

  1. Initiation: Mark Zuckerberg, in line with his professed goal of increasing independent governance, provides a fixed lump sum to an impartial assembly facilitation organization in the Democracy R&D network for implementation of a “Platform Assembly” on a predefined issue, e.g. What types of political ads should be allowed on Facebook in the United States, and with what targeting options?
  2. Commitment: He either commits Facebook to abide by all of its recommendations with e.g. over 70% of the assembly vote18 , and/or to provide detailed explanations of why any recommendations of over e.g. 50% approval are not implemented.
  3. Stakeholder invitations: The facilitating organization invites congressional and executive leadership, civil society actors, platform decision-makers, academic experts, and other stakeholders to make the case for their preferred policies to the assembly. 
  4. Participation signup: All platform users in the relevant jurisdiction (and potentially non-users impacted by the decisions) are invited to be in the pool from which e.g. 100 assembly participants will be selected to create a representative body.
  5. Assembly Process: The remainder of the process continues as described previously (lottery selection, a learning phase, and a deliberation phase), analogously to processes used by the assemblies used to decide national policy around the world. During this process, assembly participants may identify additional stakeholders and experts to learn from, and they discuss, modify, and develop proposals amongst themselves through small group deliberations with the help of trained facilitators. The assembly concludes with a vote on each proposal.
  6. Commitment fulfillment/response: The policy proposals are then implemented or responded to as specified by the initial commitments (from step 2).

A few things worth noting about this example: First, there is a specific scope and predefined commitment, which are then entrusted to a neutral third-party for implementation. Secondly, this is a simplified example for a single platform and country. It would be ideal to have global assemblies for global policy decisions. Multiple platforms could also choose to bind themselves to the outcome of a single assembly held in collaboration with multiple governments. Finally, the entire process from initiation to conclusion for a given issue could be completed in under three months (depending on the scope).


Frequently Asked Questions

Shouldn’t democratic governments be doing this via existing regulatory processes?

Existing democratic governments have a crucial role to play in ensuring that online platforms provide value to their constituents. However, some issues may be specific to particular platforms and their unique properties, and broad lawmaking may have negative unintended consequences. Moreover, even governments do not always have jurisdiction (especially with regards to some speech issues), and because these decisions may affect their electability, they may not be seen as impartial either. It is of course possible that governments themselves independently run assemblies that develop policy for all platforms, but they may also have trouble getting the political will for this. Conversely, this is a case where platforms can help accelerate collaboration by taking advantage of their relative speed to initiate a process, and then the facilitating organization can fully involve the government as stakeholders. 

How would platform democracy decisions interact with existing law, regulation, and precedent?

At least initially, this is not substantially different from the way platform policies currently interact with existing law, regulation, and precedent. The only difference would be that whatever proposals a representative assembly settles on after learning from stakeholders, facilitators, and each other would have more of a mandate than unilateral choices by platform employees. In some cases, this might arguably violate existing law, and then the platform (according to whatever policy it has decided on) may choose to challenge that law, abide by it, or ignore it—analogously to how WhatsApp and Twitter are currently engaging with new policies in India

What about authoritarian governments?

The ability to safely involve representative participants living within an authoritarian or pseudo-authoritarian regime will likely vary on a case-by-case basis, as will the influence of platform assembly mandates. It seems plausible that a global assembly could make it more difficult for such a regime to argue that a national law at odds with the outcome of such an assembly actually serves their population; whether that makes only a small difference in the mandate of that law or a large difference will also likely depend on the specific case.

What about those without access or connectivity?

Many global online platforms impact everyone, even if indirectly, and so we might strive for an ideal where everyone in a population would have an equal chance to serve in a platform assembly. Moreover, the mandate and legitimacy of an assembly becomes much stronger with better representation. This means that organizers of such assemblies should seek to ensure that the initial pool is as comprehensive as possible, and that sufficient resources are allocated to participants to ensure that everyone selected can join. Some international pilots of assemblies include not only those with poor connectivity, but even provide support for participants who are not literate, so it is likely that access obstacles will become more surmountable as our experience with such processes increases. While perfect representation is likely unattainable, this model is likely to be more representative than many traditional forms of democracy. 

Where is the platform assembly approach most applicable?

Scholars, advocates, and practitioners have different perspectives on where this approach is most useful—some believe it should be used everywhere while others believe in a much narrower scope. The OECD guide to this sort of deliberative process (based on hundreds of implementations) describes it as “best used for addressing values-driven policy dilemmas, complex problems that require trade-offs, and long-term issues that go beyond the short-term incentives of electoral cycles.”19

How would a platform assembly be more helpful than an expert advisory board?

Like an expert advisory board, a platform assembly would convene many experts and stakeholders in order to inform decision-makers and get to the best possible decision. However, unlike an advisory board, the “decision-makers” are the actual people being affected, with minimal monetary or reputational incentives to take actions that might be better for the company, a political party, or a particular interest group.

How would a platform assembly be different from the Facebook Oversight Board?

The Facebook Oversight Board is more similar to an expert advisory board with judicial-style powers. Its independence from Facebook provides significant value, and the “elite experts” doing the decision-making appear to be taking their responsibilities very seriously—but it is still an elite body.

With a platform assembly, one still gets the ‘technocratic’ benefit of elite experts, since they will be brought in to inform the assembly members. However, the people being impacted are those who ultimately make a recommendation based off of that input, leading to choices that take into account the experiences of ordinary people—and thus giving a much stronger “people’s mandate” to the outcome. This democratic component may be less critical for the Facebook Oversight Board’s judicial-style decisions, but is crucial for policy creation. That said, a body like the Oversight Board could be a complement to platform assemblies, acting as an independent agenda-setting body to identify issues to be brought to them.

Note: This is a working paper; parts of which may be used in future publications and outputs, but it can also be cited directly. It is an expanded version of a 2-pager that has been shared privately since early fall 2020.


 

1 John S. Dryzek et al., “The Crisis of Democracy and the Science of Deliberation,” Science 363, no. 6432 (March 15, 2019): 1144–46, https://doi.org/10.1126/science.aaw2694; OECD, Innovative Citizen Participation and New Democratic Institutions: Catching the Deliberative Wave (OECD, 2020), https://doi.org/10.1787/339306da-en.
2 David Sarno, “Facebook Governance Vote Is a Homework Assignment No One Did,” Los Angeles Times, April 23, 2009, https://latimesblogs.latimes.com/technology/2009/04/facebook-governance….
3 Adi Robertson, “Facebook Used to Be a Democracy — but Nobody Voted - The Verge,” The Verge, April 5, 2018, https://www.theverge.com/2018/4/5/17176834/mark-zuckerberg-facebook-dem….
4 “The Citizens’ Convention on Climate – Participedia,” accessed October 6, 2021, https://participedia.net/case/6044.
5 “The Irish Citizens’ Assembly – Participedia,” accessed October 6, 2021, https://participedia.net/case/5316.
6 “South Korean Citizens’ Jury on Nuclear Construction – Participedia,” accessed October 6, 2021, https://participedia.net/case/6875.
7 OECD, Innovative Citizen Participation and New Democratic Institutions.
8 In practice, when given options like this, comparatively few people notice or tweak them—there is rarely more than marginal impact. Moreover, while some people will choose to see less sensationalist and divisive content, those most likely to fall into extremism or depression often actively seek communities that encourage and reward it. Finally, for issues like privacy, individual choice is often insufficient, as Shoshana Zuboff eloquently illustrates in “The Age of Surveillance Capitalism.”
9 Many of these processes are bucketed under the terms deliberative democracy (e.g. citizen assemblies) and participatory democracy (which generally aim for breadth of engagement over depth).
10 Dryzek et al., “The Crisis of Democracy and the Science of Deliberation.”
11 There a variety of alternative terms for very similar processes or components of them including democratic lotteries, citizen juries, citizen panels, people’s juries, deliberative polls, etc., all of which are a form of mini-public formed via sampling or sortition.
12 E. Ugarriza and D. Caluwaerts, eds., Democratic Deliberation in Deeply Divided Societies:: From Conflict to Common Ground (Palgrave Macmillan UK, 2014), https://doi.org/10.1057/9781137357816.
13 Marlène Gerber et al., “Deliberative Abilities and Influence in a Transnational Deliberative Poll (EuroPolis),” British Journal of Political Science 48, no. 4 (October 2018): 1093–1118, https://doi.org/10.1017/S0007123416000144.
14 Simon Niemeyer, “The Emancipatory Effect of Deliberation: Empirical Lessons from Mini-Publics,” Politics & Society 39, no. 1 (March 1, 2011): 103–40, https://doi.org/10.1177/0032329210395000.
15 Kimmo Grönlund, Kaisa Herne, and Maija Setälä, “Does Enclave Deliberation Polarize Opinions?,” Political Behavior 37, no. 4 (December 1, 2015): 995–1020, https://doi.org/10.1007/s11109-015-9304-x.
16 Niemeyer, “The Emancipatory Effect of Deliberation.”
17 It is possible to implement this much more broadly, including for rules with an international scope. However, this adds additional challenges, e.g. around language, that are likely best avoided when exploring initial best practices for such a body.
18 With the exception being where this violates local law, as with the Facebook Oversight Board.
19 OECD, Innovative Citizen Participation and New Democratic Institutions.
For more information on this publication: Belfer Communications Office
For Academic Citation: Ovadya, Aviv . “Towards Platform Democracy: Policymaking Beyond Corporate CEOs and Partisan Pressure.” Paper, October 18, 2021.

The Author