Patching data governance: what can India’s rural employment guarantee scheme teach us about designing participatory data governance?

On a short family holiday just before I started at Connected by Data, the one bit of holiday reading I managed to finish (from the ever-optimistic stack of 5 books I carted around) was Rajesh Veeraraghavan’s new book ‘Patching development: Information Politics and Social Change in India’.

The book is the product of years of in-depth study and ethnographic field-work looking at the implementation of social audits within the National Rural Employment Guarantee (NREGA) scheme in Andhra Pradesh, India. I’d got hold of a copy of it not only because I’ve heard snippets of the story of the field work from Rajesh when we’ve talked over the last few years, but also it just so happens I spent time looking at the creation of NREGA in an undergraduate studies module on the politics of India in 2005, so it’s a topic I’ve loosely followed over the years.

However, what I hadn’t expected was to find just how useful the story of regulating the ‘last mile’ delivery of a policy offering rural families 100 days of manual labour a year might be to the work thinking about the design of participatory data governance systems that I would be focussed on in my new job.

Patching: governance is an ongoing process

At the heart the book is a novel framing of how policy can be put into practice: what Rajesh terms ‘patching development’. This idea, borrowing from the idea of the software patch as a “routine and continuous process”, highlights that policy rarely gets all the details right the first time, and that implementing policy can drive various actors to change their behaviours in ways that thwart the original policy goal, requiring ongoing response and adaptation. The book explores a variety of forms of patching, including:

• Patching institutions - through iterative updates to shape who carries out the social audits of the NREGA scheme, how often audits take place, who presides in public hearings and how auditors are hired;
• Patching documents - playing out in struggles over who has ‘read’ and ‘write’ access to critical records; and
• Patching technology - resulting in decisions such as which fields in the software used to plan and record rural works programmes can be locally edited, and which are centrally fixed.

When we think about developing new data governance models that can‘put community at the centre of data narratives, practices and policies’ , it’s really useful to think about this as a process of patching governance. If we approach this as a one-off task of introducing new discussion or decision-making fora, we’re likely to fail to take into account the learning, feedback loops, and, at times, pushback, that these interventions create. Instead, the model Rajesh introduces suggests that we’ll need to think carefully about how to adapt and fine-tune the institutional design, the points of interface, and the platforms for participation. There’s already some good thinking going on around this in projects like Understanding Patient Data, whose sketched model of ‘learning data governance’ explores the multiple stages needed to manage ongoing community involvement in data decision making.

Importantly, Patching Development highlights that iterating on a governance design through patching is not apolitical, and does not call for tech ’solutionism’: instead, it recognises the messy politics of participation when real resources and power are at stake. As Rajesh puts it: “Patching development is also an account of the informational politics of resistance and conflict that are central to building and sustaining participatory bureaucracy wherever it flourishes.”

People: thinking carefully about who is, or is not, involved

At the heart of the ethnographic story presented in the book are the detailed decisions over who should be involved in various aspects of delivering, and auditing, the NREGA offer of 100 days employment on demand for every rural household in India.

It draws a distinction between ‘upper level bureaucrats’ who set policy, ‘lower level bureaucrats’ who manage programme implementation, and the last-mile delivery through field assistants, engineering consultants and computer operators. At each level, these actors face different incentives, and there are different risks of resource leakage, or actions that are not aligned with the overall programme goals.

Similarly, in the parallel audit process, there are layers of actors, from a state-wide director, through to the village social auditors hired from amongst the children of workers in neighbouring villages to balance concerns of literacy and understanding of the programme, with avoiding conflicts of interest or capture of the process.

If we think about introducing more participatory data governance in a large organisation like an NHS trust, or even a sector like the health sector, we need to think about the range of actors involved and their incentive structures. For example, setting a really strong data governance policy framework at the top level, and having patients on an oversight board may not necessarily lead to better decision making by individual projects and data stewards unless thought is given to the incentive structures and decision making freedoms at this middle layer. At the same time, if bottom-up participatory processes don’t give adequate consideration to the recruitment, role and responsibilities of ‘community members’ taking part, then there are risks of both missed opportunities, and process capture.

It may seem odd to some for me to explore what learning from a policy aiming to deliver greater flow of funding from state to rural households, and where many the governance interventions are focussed on reducing risk of corruption and money leakage, might have to say about projects seeking to govern the flow of data in contexts where we rarely talk explicitly about corruption risks. However, I find that the NREGA case brings people and their incentives really usefully into view. The particular interests might be different in many data governance cases: at times, being able to re-use or restrict data in particular ways might have more impact on an individual or organisation’s reputation and future prospects than on their finances. Yet taking the value of data seriously can help us to see that the governance of data doesn’t take place in a neutral landscape of benevolent actors, but frequently involves the interplay of many personal and political interests.

Performance: creating spaces where governance is seen

Some of the most fascinating (and, dare I say with the distance of an observer, entertaining) vignettes in the book come from the public meetings Rajesh attended (with at least one he had to escape at high-speed by motorcycle after things got tense!).

Central to the social audit process of NREGA is the idea that audit findings should be read out at public meetings that the, often illiterate, workers can attend. At certain meetings, senior officials should be present to hear, and respond to, the findings. Although the book reports numerous meetings that ended up held just as a formality, with very limited attendance, the idea of the public meeting is an important one. It creates a structuring ‘moment’ to the participatory process, and a performative point of accountability.

Thinking about these mandatory moments of feedback and publicity in the Andhra Pradesh social audit model raises interesting questions about the kind of open public activities that might take place as part of collective and participatory data governance processes. For example, how should findings from a citizens jury be fed back to the wider population affected by data governance recommendations; or what public sessions should a data governance board host? Even if these routine open fora have relatively limited attendance in most instances, the holding of the space can be important both to emphasise the idea that governance is taking place on behalf of, and for, a wider population, and to create a space for accountability of data governance processes.

Thinking about moments of participation can also help us consider how participatory governance fits into citizens lives. It can be hard to keep track of, and keep engaged with, many ongoing open participatory processes, but having defined moments of open governance that anyone can step into, without requiring lots of background knowledge or prior literacy, can usefully complement standing mechanisms of engagement.

Power: individual and collective; in- and out-process negotiation

In India, the colonially-shaped caste system plays a significant role in access to work and income. One of the significant impacts of the national rural employment guarantee has been not only to directly provide minimum wage labour to landless households, but also to change the negotiating power of Dalits and other workers vis-a-vis local agricultural landholders: providing workers with the ability to turn-down agricultural labour unless it provides at least minimum wage.

At the same time, the social audit process that house-by-house verifies workers have been paid for the work they have done, and includes performative moments to call out discrepancies, creates protections against wage theft by mid-level bureaucrats or local political actors. Critically though, Rajesh highlights that not all the issues that might be raised through a social audit process are resolved within the process: oftentimes issues are settled outside of the formal mechanisms, but drawing on the rebalancing of power created by the governance design.

There are many different lessons to draw from the focus on power in ‘Patching development’. Firstly, it describes, in essence, an alliance between powerful ‘higher level’ bureaucratic actors, and the intended last-mile beneficiaries of a programme, created through careful institutional design and iteration, to avoid fatal distortion of a programme by bureaucrats and actors in ‘the middle’. Rajesh notes, however, that it leaves alone any substantive scrutiny of higher-level bureaucrats and their goals and decisions.

Secondly, it highlights the relationship between the individual and the collective. Whilst the social audit process primarily highlights individual harms (e.g. workers unpaid for their work), it takes a collectively oriented process (peer-led auditing; open meetings etc.) to secure redress. Individuals making private appeals against underpayment would be far less likely to overcome the power-imbalances that work against them, and wider systemic issues would go undetected.

Thirdly, it shows that the impacts of a ‘participatory bureaucracy’ are not always direct: but the reshaping of power relations can have wider and longer-term impacts. When we are thinking about the collective data governance work of Connected by Data, some of this may be about the long-term narrative shift so that ‘data subjects’, ‘decision subjects’ and ‘data stewards’ have a clearer instinctive and embedded understanding of the right of communities to be part of the data governance discourse.

Layered governance

The biggest lesson I take away from Patching development is perhaps the easiest to state: effective governance involves more than just one thing. It’s how the pieces fit together that decides whether we get good governance and meaningful participation.

As I’ve been starting to build out the Connected by Data case study database, I’ve quickly found that it doesn’t make sense to talk about cases in terms of a single participatory method or approach. Instead, even the simplest cases often deploy multiple methods of participation, feeding into each other, or, at best, acting as checks and balances on each other.

Over the coming weeks I’m going to explore how to better capture and explore this, and thinking about what it might mean for the practice strand of Connected by Data work, understanding the demand for practical and simple guidance on adopting collective decision making, but also the need to always see each instance of collective governance within the wider context of power, politics and ongoing process that it sits.