The knowledge of crowds and the power of networks: Using the internet to challenge corruption in South Africa
How effective can the internet be as a tool to combat corruption? The potential is exciting: applying approaches such as crowdsourcing and data-mining; exploiting the power of social networking services and creating platforms to increase accountability and transparency. But the challenges of using the internet, especially for countries in the South, are also daunting: poor information and communications technology (ICT) infrastructures; restricted access to state data; risks to whistleblowers and resource and skills constraints. Online anti-corruption initiatives are relatively new, and understanding how these pioneers are utilising the potential of the internet may help in mapping out future directions for developing and scaling up these initiatives.
The launch of Corruption Watch
On 26 January 2012, a new South African anti-corruption organisation was launched at Constitution Hill, a former political prison and the home of the country’s Constitutional Court. Present were the head of the major trade union federation, the justice minister and leaders from political parties, civil society and the major business associations.
Corruption Watch urged the public to report cases of corruption to the organisation through its website1 and other channels. This was key to the mission that Corruption Watch set itself: not only to research and advocate but to enable citizens to engage directly in combating corruption. As the chairperson of the board, Vuyiseka Dubula, put it:
It will bring back the voice of citizens to the political realm of debate around equality, accountability and transparency … empowering citizens to take a stand against corruption.2
Within three days of the well-publicised launch event, more than 70 reports had been received. Some described terrifying incidents: “A metro (traffic) cop said if I did not pay [a bribe] he would have me followed and would rape my wife and kill her. So I paid him.” Many others concerned public sector contracts. A businessman described his interaction with a municipal procurement official: “…it was made clear to me … the bill of quantities would be adjusted to accommodate R800k [about USD 100,000], which would pay all the relevant Municipal officials.” Others described nepotism in the public service: report number 71, for example, accused a senior official, Dr. M. A Seakamela, the deputy director-general in the North-West Department of Education, of nepotism. It offered detailed information with dates and salary grades and provided the names and positions of the officials concerned.
After the first month, around 500 reports had come in. David Lewis, Corruption Watch’s executive director, commented at the time: “Our principle objective is to give voice to the public. No government or business leader who looks at our data and at our Facebook page can be left in any doubt as to the level of outrage on the part of ordinary members of the public.”3
From outrage to action
By July nearly 2,000 reports had been received and processed. Around 4% of the reports received were passed on to media partners for further investigation and wider publicity. Case number 71 – the report of nepotism in North West that had arrived in January – was referred to a national newspaper, the Sowetan. After an investigation which unearthed evidence substantiating the initial report and showing that the official had broken a number of rules in appointing and promoting his girlfriend and now wife, the paper published an article in June: “Naked nepotism exposed - From bedroom to top office”.4 Three months later the official was suspended pending an investigation initiated by the provincial minister for education.5
A small number of reports led to in-depth investigations by Corruption Watch’s small team of investigations specialists, which were then passed on to state agencies capable of taking further action.
Other reports formed the basis of public campaigns. Those that concerned traffic police asking for bribes contributed to a campaign to challenge paying traffic bribes with hundreds of thousands of bumper stickers handed out to motorists and a detailed research report handed to the Johannesburg authorities.6
The challenges of using the internet
The key platform that Corruption Watch developed in order to “give voice to the public”, as the executive director put it, was a web-based reporting tool built on the Drupal open source platform. The original technical specification of the website created by Corruption Watch and the website developers7 set out an ambitious framework for the reporting tool: first, a reporter questionnaire that would guide respondents through a set of questions to categorise the type of corruption, locate it and provide details that could lead to further investigation; second, a means of storing this information in a secure database; and third, the ability to present the information in aggregate via maps and charts8 on the website.
The specification also allowed for registration and categorisation of users to enable the management of groups of users so that they could track the progress of actions on reports. Developing and then managing these online tools meant dealing with a number of challenges.
Managing multiple channels of communication
According to the South African Network Society Survey, one in three South Africans uses the internet.9 In comparison, over 80% of adults use a mobile phone. Among internet users, by far the most popular applications are social network services. There were an estimated 5.1 million Facebook users in South Africa in July 2012.10 Corruption Watch aimed to address this reality by enabling a variety of communication channels for those wanting to report corruption. The specification called for access to the reporter via email, short message service (SMS) mobile texting and Facebook.11 A short-code SMS gateway was secured and a Facebook page created.12 Analysis of the reports after six months shows that 46% of reports came via the website and 42% via SMS. Email accounted for 8% of reports. The remainder came via phone, post and personal contact.
Managing this diversity of channels has been time consuming and complex. While SMS was clearly an important channel and enabled broader access than would be possible with the internet alone, 83% of SMS reports were deemed invalid due to insufficient information, a far higher proportion than was the case online. This often required follow-up calls with reporters. Even where there was considerable information provided, it was not structured and so it had to be entered manually into the database.
SMS is also relatively expensive as a communication medium. Each message to Corruption Watch costs the user R1 (approx. USD 0.15) and users often have to send multiple messages to include the information they want to provide.
Crowdsourcing corruption data
An objective for the website was to enable sufficient quantities of rich data to be able to provide useful indicators and insights into the nature and scale of corruption in South Africa. For example, it was hoped that it would be possible to identify “hotspots” – particular places with high levels of reports. This follows crowdsourcing data projects like Ushahidi, which was developed originally to monitor post-election violence in Kenya. Data on corruption is notoriously difficult to get for obvious reasons. In general it happens out of sight and those who know about it are often those who have a stake in its remaining hidden. Transparency International’s corruption index,13 for example, does not measure corruption itself but rather people’s perceptions of corruption. Since reporters are self-selected, it is not possible to use Corruption Watch’s data as a representative sample of corruption in the country. However, analysis of the report data has helped the organisation gain insights that have contributed to informing its priorities and strategies. For example, the largest category of reports concern abuse of public funds by state officials – including, for example, using state resources for election campaigning and using state vehicles for private business. There is also a surprisingly high share of reports coming from small towns outside the major cities, often concerning corrupt procurement. While it is not possible to draw conclusions on where corruption is most prevalent, this information does help direct the organisation’s attention to issues that its community of reporters know about and care about.
Managing anonymity, verification and confidentiality
The online reporter gives respondents the opportunity to remain anonymous or to provide their names and contact details, which Corruption Watch undertakes to keep confidential. As of July, 60% of those reporting provided their names and contact details. This suggests a high level of trust in the organisation. While anonymity is necessary in some contexts, it has proved highly challenging in other crowdsourcing initiatives. Ushahidi, for example, has reported on the challenge of verification where reporters are not identified.14 The Crimeline SMS line established by South African media group Primedia in 2007 is entirely anonymous and faces the problem that reporters often do not provide enough details for action.15 The fact that most of those reporting corruption have provided their names and contact details has enabled Corruption Watch staff to follow up with requests for further information. It also enables the organisation to engage these reporters in campaigning. But it also creates other risks that have to be managed. Most important of these are the risks of breaching the confidentiality of reporters. This has required physical, operational and legal approaches.
Managing communication and conversation
As David Lewis had said at the organisation’s launch, Corruption Watch aimed to be more than a research institution or think tank. He wanted it to be the hub of an activist network. This required promoting a two-way conversation. But it also required trying to manage the direction and tone of this conversation. Amongst the concerns the team had were that the discussions could become dominated by unproductive “whinging and whining”, or that they would move off-subject into general attacks on government failures. Another concern was that the online community would not be sufficiently diverse, especially in terms of race and class, to fairly represent the constituency that Corruption Watch was aiming to reach.
What may not have been clear prior to Corruption Watch’s launch is the extent to which norms and even definitions of corruption are contested. When the Sowetan published their first story on case number 71 online, it elicited almost 100 comments. The debate was extensive and explosive. Leutler16 responded: “I think this man is doing what everyone is doing, government officials are all enriching themselves and their family starting with Zuma, so give that man a break. I would do the same if i get opportunity.” Blackthought was equally cynical: “Lol....happens everywhere around us. Just too scared to do anything about it. White people do this all the time.” However, he or she was challenged by WakeUpSouthAfrica: “Blackthought: Shame, is that how you defend nepotism? White people do it?”
Others disputed whether there was a corruption case here at all: “thers no Story here !!! so, what if the lady was well perfoming, a hard worker & doing her job accurately ? so acording to you she should not be employed wether she has the credentials or not?” asked MissBhakajuju. Dzel responded with: “You can't be serious Bhakajuju, by using her connections she has jumped the queue… If she was competent then she should have competed for the position with everyone else... It could have been your sister who was overlooked, is that fair?”
The debates on Corruption Watch’s Facebook site have often been equally robust. In September Corruption Watch published a report17 on Facebook and Twitter concerning the costs of corruption based on information from one of the state’s investigation agencies. Helen Mudge responded: “Why do you take 1994 as the start of your Corruption Watch statistics? There was a great deal of corruption in SA long before 1994. I find it offensive, narrow-minded and racist that you take 1994 as the starting date for these figures.” Erwin Schwella responded within minutes: “@ Helen Mudge We all expected the National party to participate in corruption as the system of apartheid was systemically corrupt. However we had/have different expectations from the state government of constitutional democracy. We can therefore be doubly disappointed…”
These exchanges expose the divergent social norms and ideological perspectives that exist in South Africa in relation to corruption. These differences are not as clear in mass media where discussion is more mediated by gatekeepers, and this demonstrates an important ability of the internet to extend participation in the public sphere.
Crowds, networks and organisation
The “theory of change”18 that Corruption Watch has been developing is complex and has multiple components.19 The organisation values the gathering, investigating and then publicising of reports of corruption in itself, following the famous statement of Louis Brandeis that “sunlight is said to be the best of disinfectants.”20 It prioritises analysis and research as a basis for campaigns – aiming to harness the widespread anger and frustration that can be seen frequently in South African mass media and transform them into public action. It has also identified changing social norms and behaviour as important in changing the environment for corruption. And it has identified a need to reposition corruption as an issue that affects everyone – especially the poor – thus aiming to connect it to broader political debates on social development and service delivery. Lastly, it sees improving accountability and transparency in the agencies responsible for investigating and prosecuting corruption as a means to increase their effectiveness and improve deterrence.
The range and complexity of the organisation’s ambition and its analysis of the complex problem of corruption imply a corresponding complexity in how it uses the internet and other communication tools. While data is valuable in itself, the organisation’s strategies require doing much more than gathering and publicising data. It aims to build a network of partners capable of taking action on the reports it receives and holding prosecuting agencies responsible. It also aims to use online platforms as a means of encouraging and organising action – as organisations such as Avaaz.org have done successfully. This will require further development of the online tools it uses in the future.
Conclusions and action steps
Corruption Watch is not the first anti-corruption organisation to use the internet to gather or disseminate information. The Indian I Paid a Bribe21 website, the Russian navalny.ru site and others in Nigeria and Kenya have all used the internet to create platforms for increasing transparency and building anti-corruption communities. Corruptionwatch.org.za, though, is one of the most ambitious projects developed to date, aiming not only to record and crowdsource data but also to network citizens and a wide range of institutions together towards action.
The organisation’s strategies and the online platforms themselves are still in development. However, its work so far enables us to identify a number of themes that may be important not only for this project but for others following similar paths:
Even where internet penetration rates are relatively low, websites and Facebook pages can be used effectively – but this requires integration with other communication channels, which can be challenging and resource intensive.
As online communities become more representative, there is an opportunity for enabling a public sphere where social norms around corruption can be debated and influenced, especially where mass media have limited reach or are tightly controlled.
Moving beyond anonymous reporting has allowed Corruption Watch to engage with those reporting – strengthening its capacity to build social activist networks – but it also creates risks that the organisation has to actively manage.
Corruption Watch’s developing theory of change identifies creating and disseminating information on corruption as important steps, but not sufficient in themselves. Its focus on building strong networks and partnerships is likely to drive further development of how it uses online tools to further its objectives.
7 The website developers were Hello Computer, a leading South African web development and design company.
8 The presentation tool has not been enabled at time of writing.
9 South African Network Society Survey, Media Observatory, Wits Journalism (publication forthcoming).
11 Not all these functions have been implemented at the time of writing.
14 “Ushahidi support for verification has until now been limited to a fairly simple backend categorisation system by which administrators tag reports as ‘verified’ or ‘unverified’. But this is proving unmanageable for large quantities of data.” blog.ushahidi.com/index.php/2011/11/11/what-is-the-next-step-for-ushahidi-verification
16 Like most news websites, the Sowetan’s site requires those who wish to comment to register, but displays only user names.
18 Theories of change are the explicit and implicit theories underlying an organisation’s strategies and priorities that connect their objectives (the change they are seeking) and the means by which these changes will come about. The Corruption Watch staff have been working on developing and articulating a theory of change to inform their strategies. For more on theories of change see Weiss, C. (1997) How Can Theory-Based Evaluation Make Greater Headway?, Evaluation Review, 21 (4), p. 501-524.
19 These observations are based on numerous conversations with Corruption Watch staff and founders as well as public statements to be found on the Corruption Watch website.