What would it take to turn Facebook into a democracy?

by Severin Engelmann and Lisa Herzog*

When the relation between “Facebook” and “democracy” is discussed, the question usually is: what impact does Facebook – as it exists today – have on democratic processes? While this is an urgent and important question, one can also raise a different one: what would it mean to turn Facebook into a democracy, i.e. to govern it democratically? What challenges of institutional design would have to be met for developing meaningful democratic governance structures for Facebook?

At this point, Facebook is run as a corporation, and users and employees have hardly any voice in its strategic decision-making. Facebook has indeed experimented with user deliberation and voting: In 2009, it had set up a comment-and-vote governance system with the declared purpose of enabling user participation in its policy decision-making. As a recent paper demonstrates, these experiments have not lived up to procedural standards of democratic decision-making.

In 2012, Facebook ended its participatory governance model by putting the right to vote on Facebook policy to a vote itself. Since the necessary quorum of 300 million voters was decisively missed (only 668,872 users voted), users effectively lost their participation right. Proposing the termination of its participatory governance process, Facebook claimed: “We deeply value the feedback we receive from you during our comment period but have found that the voting mechanism created a system that incentivized quantity of comments over the quality of them. So, we are proposing to end the voting component in order to promote a more meaningful environment for feedback” (see Facebook screenshot).

After this announcement, however, Facebook never implemented another participatory governance process. Recently, following a plethora of privacy scandals, the question as to whether more stringent forms of social media governance are indeed possible and viable has gained global public interest. Some – including CEO Mark Zuckerberg – have openly supported the idea of an independent Supreme Court tasked primarily with the specification of free speech and hate speech. Others have again suggested to shift some power from the platform operator to social media users. However, since the termination of user participation governance in 2012, no tangible changes towards democratic forms of social media governance have been realized.

The basic rationale for developing a democratic governance model is similar to that often quoted as an argument for workplace democracy (see e.g. here): if certain people have power over others, they should be held democratically accountable, especially if important interests are at stake. For Facebook, the most important group whose interests are at stake are the users, who share their data on Facebook (arguably a form of work, and the basis of Facebook’s economic success!), but whose data and privacy are at risk of being abused by Facebook (see e.g. here and here). Other groups, such as investors and employees, also have interests at stake, and so should probably also have some kind of say.

The question thus is: what kind of democratic governance model might be appropriate for an international social media platform such as Facebook? We put aside, for now, the question of whether its (advertisement driven) business model would still work, and focus on the structures that a “democratic Facebook” should have. This questions turns out to be rather complex, and requires taking into account a number of technical and non-technical features of Facebook. In what follows, we discuss a number of these challenges, without a claim to completeness.

– Individuating participants: Currently, verification of a Facebook account is based on a unique Facebook ID linked to a single email address. This opens the door to participation manipulation since one user can have many email addresses connected to multiple Facebook accounts. Thus, any form of user participation requires a system distributing unique user identifiers. One option would be to change the terms of service and to only allow for accounts with ID-based verification (but the form of ID would have to be chosen wisely in order not to discriminate against certain groups, e.g. homeless people). Moreover, the voting process must ensure authentication and anonymity of the user but at the same time reveal the content of the voting transaction (one system that fulfills these criteria is the e-Petition system proposed by Gürses et al).

Accounts that belong to legal persons, e.g. companies, would presumably not be permitted to vote; it should be noted, however, that companies often have fan pages rather than accounts. Fan pages can have unlimited amounts of followers, where personal accounts can have a maximum of 5000 “friends”. So one would also have to consider what rights the owners of fan pages would have during campaigns, because they could, potentially, have a massive influence on voters or participants in deliberative formats, even if they could not vote or deliberate themselves. In principle, fan pages run by independent bodies could also work as spaces for deliberation by users about the issues they vote on.

– Designing constituencies: At least for some issues, a democratic governance structure for a social network like Facebook would probably require representation. This raises the question of how representatives could be elected – how would one design the election mechanisms and design the constituencies? One possibility would be to have one representative per, say, 100.000 users in a designated region. Another possibility would be to use features such as home region or user language as additional criteria for carving out constituencies. One could also imagine a federal structure with regional parliaments, e.g. one for Iceland and one for France, which would be integrated, e.g. through delegates, into an overall structure. In any case, these constituencies should be designed such that “gerrymandering” would not be possible.

– Allocating responsibilities: The questions to be decided about by a democratic Facebook would have to remain within the scope of the legal framework set by different states (in other words, we assume a priority of political democracy). With regard to internal issues, however, there are further questions about who should decide about what. For example, when it comes to working hours and the design of job rotation, there are strong reasons to give employees a say. When it comes to the management of user data, it is mainly users who are concerned. Assuming Facebook remains a business (rather than, say, being turned into a public infrastructure), one could imagine a multi-cameral system, with different chambers for employees, investors, and users (along the lines described for corporations by Isabelle Ferreras). Last but not least: which entity has the final say over how to allocate responsibilities? This question concerns the basic structures of power, and unless the constituents have a say here, one must doubt whether the governance structure as a whole could be called “democratic”.

– Protection of minorities: All democracies have to find ways to protect minorities. Often, this is done by legal mechanisms, or sometimes by ombudspeople who can speak for minorities and mediate in cases of conflicts. Something similar would probably be needed for Facebook, e.g. to protect linguistic minorities.

– Expertise and democracy: For many debates about issues that concern Facebook users, high levels of technical and marketing expertise would be needed; this include the details of how profiles are created, what data is collected, processed, and analyzed for people-based marketing, or questions about the role of “external” data brokers such as LiveRamp. Hence, data scientists and other experts who work at Facebook would probably play an important role in presenting information, explain alternative options, and implementing decisions. Therefore, one would need mechanisms of checks and balances that prevent manipulation or distortion when they insert their expertise – which only other experts can judge – into the democratic process. But these data scientists would still be employees of Facebook, so one might question their independence. One possibility would be to have independent expert bodies as review boards. Or one might consider delegating the design and execution of online deliberation and voting to external bodies, to make sure there can be no manipulation.

 As these considerations show, designing a democratic governance structure for an entity like Facebook is not straightforward – but neither is it impossible. Political theory can deliver ideas about mechanisms, while technical expertise is needed to understand the technical challenges and the possibilities of implementation. Thus, this is a case for interdisciplinarity! A second challenge, of course, concerns the transition: how could one develop the kind of political pressure and design transition mechanisms for turning Facebook democratic? We take it, however, that the latter question can only be addressed if reasonable proposals for a democratic governance structure are on the table. Therefore, this is a debate that we need to get started!

PS: If readers happen to know about other articles or papers that discuss this topic, please let us know, we have not found any so far!

*Both authors work at Technische Universität München. This work has not been externally funded.