Shifting the Spotlight: Understanding Crowdsourcing Intermediaries in Transparency and Accountability Initiatives

"Whereas the users of earlier, targeted transparency initiatives and 'right to know' policies had to rely on disclosers of information and regulators to obtain the information they sought, this wave of collaborative transparency initiatives mobilises citizens to generate flows of information for their own purposes."

This paper presents findings and conclusions from a multi-method study of "crowdsourcing intermediaries", defined in transparency and accountability (T&A) initiatives as civil society and state actors who aim to contribute to more accountable governance through facilitating citizen feedback on government, using internet or mobile phone technology to source opinions. Instead of asking whether T&A initiatives are effective in terms of representation (whose voice is being heard?) and impact (what kind of change is being supported?), the research focused on how crowdsourcing intermediaries mediate citizen voice, and why. It addressed 2 questions: What tools, policies, and practices do crowdsourcing intermediaries adopt to express citizen voice? Who do they mediate between, and why? Key themes covered in the paper are: the role of crowdsourcing intermediaries as gatekeepers of citizen-generated data; the accountability of crowdsourcing intermediaries to citizens who contribute data, especially in terms of data policies; and the factors that influence the pathways of individual crowdsourcing intermediaries.

"By definition, social accountability initiatives seek to encourage engagement between citizens and their government, and although they adopt varied forms and approaches, they all involve information flows. The study of information flows and what defines their character, direction and power is the main focus of gatekeeping theory. It was developed to explain the role of newsprint and broadcast media in shaping the news that lies at the centre of our public life. The rise of a new generation of digital gatekeepers, such as Google and Facebook, has inspired scholars to ask who the new digital gatekeepers are, and examine the extent of their influence."

To that end, the research used an original conceptual framework that combines ideas from the governance and social accountability fields with networked gatekeeping theory. According to this framework, crowdsourcing intermediaries are gatekeepers of citizen voice who can shape and therefore control many different aspects of the flow of information generated by contributors. In the gatekeeping model, the power that intermediaries have is both productive and restrictive. It is productive in that it creates new flows of information about public services between citizens and states, and vice versa. However, it is also restrictive in the sense that it imposes conditions on what information can be collected and how it is presented. This research highlights 2 flows of information: the collection of citizen feedback, and the contextualisation of this information for pursuing accountability.

An example from the open data field is provided in the report to help readers understand what the process of gatekeeping involves in tech-supported T&A and crowdsourcing initiatives. Assume that a government provides budget and expenditure data in an Excel file that can be downloaded from the internet free of charge. A local advocacy group - a "user group" - wants to identify the funds allocated to improving school infrastructure in its region, how much has been spent, and to what effect. To achieve this, budget data would have to be cross-referenced with expenditure data, and the results matched against the current state of the region's schools. This involves different skills and resources: data science skills to combine data from different sources; statistical skills to analyse them; local connections at the grass-roots level to collect information on how the money allocated has been spent; and understanding of the local context to make sense of it all and identify specific opportunities for action. This kind of intermediation, which involves many different stakeholders working towards a common goal, is consistent with scholar Jonathan Fox's idea of strategic approaches to social accountability. In terms of the gatekeeping model, the "gated" in this type of initiative are the citizens who choose to respond to a crowdsourcing effort, while the "gatekeeper" is the crowdsourcing intermediary who decides what information is going to be collected through which channels, and how it is going to be analysed and presented. Every choice that a crowdsourcing intermediary makes has a bearing on whose voice can be expressed, and how.

The research combined content analysis of website text with qualitative case studies. (See Table 1 on page 10 for the 20 websites, representing a sample of crowdsourcing initiatives in developing countries aimed at improving citizen representation or supporting social monitoring, and for the sub-sample of 4 interview-based case studies, 2 focusing on social monitoring and 2 on voice. The examination focuses on crowdsourcing initiatives that are open, in theory, to every citizen with a mobile phone and/or access to the internet.) The picture that emerges reveals a great deal of fluidity and experimentation in the way that crowdsourcing is defined and used as part of the political strategies of crowdsourcing intermediaries. For instance, the analysis of 20 websites showed 3 broad approaches to the tools, policies, and practices of crowdsourcing in the selected initiatives:

1. Crowdsourcing for addressing specific issues and supporting structural change (I Change My City, Fix My Street, Maji Voice, Stop Stock Outs Project, Sauti ya Mtaa, CGNet Swara (later stages), Check My School, Citizen Feedback Monitoring Programme, Corruption Watch (early stages), Daraja Maji Matone) - Initiatives taking this approach seek primarily to connect citizens and either government authorities or elected representatives for resolving specific issues or answering particular queries / concerns. The process of reporting is often highly structured, with participants needing to provide specific information in a specific format, allowing data to be quickly manipulated and analysed.
2. Crowdsourcing for improved representation (Tracka, U-Report Uganda, Sauti Za Wananchi, Sauti ya Mtaa, I Paid a Bribe, CGNet Swara (early stages), Report Xenophobia, Not in My Country) - These initiatives use crowdsourcing as a platform for expression and as a way of opening up a window on the views and experiences of a broad audience, including traditional media. Initiatives in this group maintain a high degree of control over how an issue is framed and reported.
3. Flexible reporting for collective action (Praja, amandla.mobi, Vouli Watch).

Equally varied were the roles and relationships that crowdsourcing intermediaries engaged in as collectors and analysts of citizen feedback. For instance, various factors influence an intermediary's decision to mediate in T&A initiatives, and whether and how to use technology. These include their: implicit theories about what it takes to influence government behaviour and how different approaches to intermediation influence this; personal background, character, and technical ability; links to government and civil society, and perceptions of how policy and decision-making work; and capacity to establish and communicate their trustworthiness.

A key finding is that in crowdsourcing initiatives, it is difficult to distinguish between the interpretive aspects of intermediation, which comprise the collection and analysis of citizen feedback, and political aspects of intermediation, which involve using the collected information to support positive change. "The picture that emerges from the case studies departs from previous accounts where crowdsourcing initiators believed that the availability of information alone could support positive change. Instead, it suggests that 'infomediation' - sometimes framed as a purely technical activity - cannot be separated from the broader and more political strategies of which it forms a part." In other words, crowdsourced information is inherently political. "Previous accounts of intermediation have emphasised the gap between information provision and action, arguing that technical proficiency and the establishment of new information flows does not necessarily support collective action and positive change. We argue that, equally, political savviness and good intentions may be undermined by poorly-thought-out data policies and failure to communicate clearly to citizen contributors the implications of their participation."

The analysis of website content also highlighted privacy concerns and challenges about how transparent and accountable crowdsourcing intermediaries are to their participants on the basis of the information they make available on their websites. Only 10 out of the 20 websites studied included a privacy policy; 6 did not explain the sequence of actions that were triggered by submitted reports; and only 2 made the information they collected available in a format suitable for further analysis. "On the whole, this lack of clarity on what citizens should expect as a result of their participation can seriously undermine the aims of these initiatives to engage citizens and make their voices count, or their downward accountability. For example, the submission form for I Paid a Bribe includes an option for participants to share their report on bribery with the relevant authorities and the media, but doesn't explain what citizens that agree to this should expect."

These and other findings suggest that there is much to be done to ensure that those who seek accountability on behalf of others are equally accountable to them. Suggestions offered here include encouraging citizens who have signed up to a service using short messaging service (SMS, or text) to visit a website or listen to a recorded message that explains clearly and succinctly how their anonymity is protected. An alternative approach might be to provide spaces for intermediaries to discuss their approaches, showcasing innovative applications of technology for this purpose. Bringing actors together physically or virtually to do this could encourage the co-creation of appropriate standards.

Based on these and other conclusions, the paper presents a number of further questions for future research:

• "Given that the use of technology for crowdsourcing falls within broader political strategies, how can these be best studied and understood?
• How can we begin to increase the responsiveness and accountability of infomediaries to their users and / or constituents?
• How are crowdsourcing platforms, their processes and effects experienced by citizens?
• How do service providers negotiate the requests put forward by crowdsourcing intermediaries?
• How do existing, readily available tools and platforms for supporting citizen participation 'gate' crowdsourcing intermediaries?"