Seeing data like a state: A case of Open Government Data in India’s livelihoods program

1.Introduction

India’s livelihood program, the Mahatma Gandhi National Rural Employment Guarantee Act (MGNREGA), is the world’s largest public works program providing employment every day to over 8.5 million workers. Every member of a rural household can apply for work at the local government office. Upon performing “unskilled manual labour” (such as desilting a lake), the government is supposed to pay them their legally guaranteed wages within the stipulated time. The scheme’s claim for success is not merely its rights-based core and massive scale but also its technology backbone, with multiple information systems proactively disclosing data about the operation of the scheme in the public domain. Enthused by the datasets, the first author made easily comprehensible visualisations depicting budgetary outlays and other performance indicators across local governments. The assumption was that such micro-level data analytics would generate conversations in the village communities on the performance of the local government and eventually lead to greater participation in local government. He took these visualisations to Rajeev, an activist of grassroots organisations working on the program for over a decade in dryland regions of Karnataka, to aid him in his grassroots campaign. Rajeev was dismissive of such visualisations and rhetorically asked – “what should I do with this, pickle it?”

In the historical context of the transparency movement, access to datasets such as those we analysed and visualised is no ordinary feat. The official data of the predecessor programs of MGNREGA were not in the public domain. Social movements had, in the past, often struggled to gain access to such datasets relating to welfare programs. Such a sustained campaign led to passing of a Right to Information Act and many pro-transparency measures within MGNREGA.11 Nor was the move towards transparency restricted to India, instead unfolding worldwide. Digital information systems enabled the state to cost-effectively disclose data in the public domain proactively. OGD enthusiasts, including governance reformers and tech enthusiasts, see huge potential in publicly available digital government data. The emergence of digital technologies has significantly expanded our ability to store, compute and communicate digital data on an unprecedented scale. Such “datafication” promises not merely a shift in the material form of data but also promises to alter the functioning of organisations and society. Governments worldwide have been quick to use such affordances of digital technologies to datafy their information systems. While governments have historically been at the forefront of adopting and even contributing to innovations in recording administrative data, what is new is the claims of such datafication enabling citizens’ participation in governance. The assumption is that "opening up" the hitherto hidden government data will lead to an open government, in both the sense of being made visible to citizens and providing opportunities to voice their concerns (Meijer et al., 2012).

The availability of datasets has indeed contributed to greater questioning of government performance. For instance, social activists have used similar datasets of MGNREGA to successfully seek judicial intervention to address delayed payments (Swaraj Abhiyan Vs Union of India, 2016). Similarly, data journalists routinely use such datasets to bring out data stories in the public domain, providing key insights into the performance of the programs. Further, the academic community has extensively used such datasets to critically analyse the performance of the program.

Notwithstanding these positive possibilities of OGD, the evidence of the impact of OGD remains rather weak. Multiple systematic literature reviews on OGD support such a view. Francey and Mettler (2021) review the literature to assess the empirical evidence on the effects of OGD and find that there is little to substantiate that OGD supports citizens’ participation. Yet another literature review by Safarov et al. (2017) assesses the utilisation of OGD. They find that the effects of OGD are often “only proclaimed and not empirically tested”. Similarly, based on a literature survey on the impact of OGD on democratic societies, Ruijer and Martinius (2017) suggest the need to temper our expectations, especially in light of many challenges and barriers, such as lack of data literacy and supporting data infrastructures.

We seek to understand why OGD doesn’t lead to citizen participation and, more broadly, the role of OGD in the everyday engagement of ordinary citizens with state institutions. Our focus is on citizens who are recipients of welfare programs on livelihood security, healthcare, social security, housing and others. Following the emerging literature on the anthropology of the state, we seek to examine the “banal, mundane, routinized …unexotic” practices of citizens’ engagement with what could be called the “everyday state”, i.e. those institutions that engage with citizens daily (Fuller & Bénéï, 2001; A. Sharma & Gupta, 2005). Such emphasis on the everyday state assumes importance, especially due to the schisms within the seemingly monolithic state whereby the policies are often “reinterpreted beyond recognition” by the everyday state (Kaviraj, 1984). While such emphasis on the workings of the everyday state is important, we also analyse the OGD initiative through the lens of what could be called the “distant state”. We use the term “distant” to imply that it is distant in terms of its effects on people’s everyday lives and to indicate the continuance of ways of thinking about governing populations.

Often, research on OGD begins with data availability and examines its use (and non-use) and impact. Drawing on Avgerou (2010), such a perspective can be called a “transfer and diffusion” perspective. In contrast, we adopt a “socially embedded” perspective that examines the social context in which such initiatives are embedded and sheds light on what is “locally meaningful, desirable, or controversial”. Our findings on OGD use and non-use by citizens are based on a 14-month ethnography of India’s livelihood program under the Mahatma Gandhi National Rural Employment Guarantee Act (MGNREGA).

We argue that OGD initiatives present data that is desired by the state. Drawing on James Scott’s work on “seeing like a state” (Scott, 1999), we show how these datasets are state-centric in both their objectives and their form of presentation (Section 3). A key assumption underlying state-society relations is that citizens are willing and able to monitor the society’s functioning and further contribute to the decision-making process. OGD initiatives assume such “citizen-auditors” to monitor the functioning of the state. Further, seeing data like a state introduces a techno-official language that citizens are expected to master.

We show how not all citizens see themselves as “watchdogs” or anti-corruption crusaders but as recipients of welfare services (Section 4). Their key objective is to “get things done”, through an engagement with the local state. Yet, the implicit assumptions of OGD design – that they are primed to be crusaders – contribute little to the citizens’ engagement with the state. Further, since OGD represents data as seen by the state, the citizens are expected to speak like a state. We conclude by highlighting our research’s implications and suggesting a “citizen-centric” OGD to complement the existing state-centric OGD initiatives.

2.Research design: An ethnography of India’s livelihood program

Our findings are based on an ethnographic case study of India’s livelihood program, the Mahatma Gandhi National Rural Employment Guarantee Act (MGNREGA). MGNREGA is a universal program open to any rural household. It mandates that a minimum of 100 days of guaranteed wage employment in a financial year should be provided to every rural household whose adult members volunteer to do unskilled manual work (Ministry of Rural Development, 2013). This program has two key objectives: to provide employment to wage-seeking rural citizens and, secondly, to create durable assets, improve water security, soil conservation, and agricultural productivity. Together, these two objectives are intended to provide livelihood security to rural citizens.

MGNREGA is best suited to understand the role of Open Government Data (OGD) in citizens’ participation for the following reasons. MGNREGA has an explicit commitment to deepening democracy by enabling citizens’ participation and making the everyday state transparent to ordinary citizens. Further, MGNREGA has multiple information systems, all being real-time, transaction-based systems with much of the data available in the public domain. Notably, the publication of such data online precedes the popularity of OGD initiatives such as data.gov portals and Open Government Partnership.22 Also, as described previously (Section 1), social movements had long campaigned for such datasets to be accessible, even before the digitisation of information systems. Thus, the datasets exist with the key assumption that they will enable citizen participation in their everyday engagement with the state. Also, the precarious livelihood situation, especially in dryland regions, makes the program extremely relevant to rural households. In terms of budgetary outlays, too, MGNREGA stands out among all other programs with a cumulative expenditure of over US $ 100 billion (Rs 7,77,514 crore) since its launch in 2005.

MGNREGA is a centrally designed program, with the union government providing most of the funds, scoping the program, and designing the digital technology backbone to administratively manage the program. However, MGNREGA follows a decentralised program implementation with reasonable autonomy for the Gram Panchayats33 in program implementation. Thus, the Gram Panchayat is responsible for registering job seekers, identifying works to be undertaken, providing employment to job seekers, and disbursing payments.

We study the OGD initiative of MGNREGA using an ethnographic research design since we seek to understand OGD from a socially embedded context where the phenomenon under study cannot be easily delineated from its contextual factors. It thus requires an in-depth understanding of these factors. We conducted our fieldwork in the Gram Panchayats of the rural dryland district of Chikkaballapur in Karnataka, South India, between January 2018 and March 2019. The rural livelihood distress, especially in these dryland regions, makes MGNREGA relevant to rural households. We used participant observation, interviews, and document analysis as the key methods of data collection. The first author, whose social position as a male urban dominant caste researcher gave him significant access to most sites, conducted the ethnography predominantly. The ethnography engaged with three key actors – social activists and citizens, Gram Panchayat and local elite, and the social audit team. However, for the purposes of this paper, we have drawn predominantly on the findings from the social activists and citizens. A brief overview of the site and events observed and participants interviewed is shown in Tables 1–3.

Our initial foothold in the fieldsite was through an organisation we call NPO, a loosely structured organisation working from a rights-based perspective. It consists of a team of 3 core social activists and other freelancing volunteers. This team is led by Satish, who has a long history of working on campaigns for the right to food and livelihood. While the organisation is three years old, Satish and his team have worked pro bono for at least ten years to facilitate people’s access to MGNREGA. Satish has a very strong informal social network and has friends in various quasi-official roles within MGNREGA. Satish emphasises the socio-economic backgrounds of his team, who, unlike many of the “city-based activists” hail from the very same villages and communities they engage with. According to Satish, these are the “real social activists” who deal with “abhadrathe” (insecurity) about their livelihood every day and have faced similar forms of discrimination that the communities deal with. All three social activists of NPO belong to socially marginalised scheduled caste (SC) and scheduled tribe (ST) families involved in dryland agriculture on small pieces of land. Some are from families involved in bonded labour until a decade ago. NPO has previously intervened in getting people to work under MGNREGA, facilitating access to veterinary doctors for livestock and securing institutional bank linkage and loans for self-help groups (SHG). The social activists stay in the nearest44 town and travel to the villages for community meetings. We rented a small room next to the ones used by social activists of NPO, which allowed us to observe and participate in their organisational meetings and accompany them in their community meetings at the villages. Along with NPO, we also observed the functioning of two other non-profit organisations working predominantly on gender and community commons, respectively, and on the livelihood security concerns of MGNREGA. One of the villages, which we call Venkatapalli, became the key observation site. The first author visited the village in the evenings, teaching English to school students and participating in community gatherings, festivals, and other events.

While access through NPO helped us understand citizens’ perspectives, we also independently engaged with the local government, i.e. the Gram Panchayat. A Gram Panchayat is the lowest tier of the elected government and covers a population of about 4000, and consists of 10 villages that are represented by 15 elected members. While we visited and observed five Gram Panchayats, we spent much of our time in the Venkatapalli Gram Panchayat office. The Data Entry Operator (DEO)’s room is one of the most active sites within the Gram Panchayat, where most transactions related to MGNREGA occur. The administrative processing of MGNREGA is heavily dependent on the use of information systems. The extensive use of information systems makes the DEO one of the most important actors in the MGNREGA ecosystem. Along with the DEO’s office, two other sites became our primary source of observations and semi-structured interviews. One was the courtyard outside the office, where the elected members and elite gathered for a chit-chat. Some key decisions would also be taken here, such as the allocation of houses under the government housing scheme. Another site where we conducted most of our semi-structured interviews was the local tea stall next to the Gram Panchayat. While sipping a cup of tea in a plastic cup, the engineers, watermen, bill collectors, and social audit volunteers would share their challenges with the “system”. We also followed up the observations by studying the work practices of Block Panchayat officials, including the MIS (Management Information System) Coordinator, Technical Coordinator, and Engineers.

The third key set of actors included the social audit team. A social audit is a participatory mechanism to ensure transparency in the program’s functioning. A social audit involves a door-to-door verification of households, physical inspection of all worksites and verification of all the documents to ensure compliance with the rules of MGNREGA. The Village Resource Persons (VRPs) who conduct the social audit are not government employees but are considered volunteers. We followed the social audit team in multiple Gram Panchayats to better understand the practice of social audit.

As common in ethnographic research, we let field events guide our site choice and participants. Yet, a routine began to emerge in the first few weeks of our fieldwork. We spent the mornings with the social activists of NPO in their office, often cooking food together. Such informal conversations helped us understand their challenges with the local government and citizens. We spent the official working hours (10:00 AM to 5:00 PM) on most days in government offices, observing and interviewing the officials and the visitors. Social audits took place during these working hours and allowed us to visit multiple villages and Gram Panchayats. Village visits were reserved for late evenings when the citizens, mostly engaged in casual labour, would return home. We taught English to children in the evenings, which was followed by semi-structured interviews and late-night dinners. We would return to our room in the block late at night to transcribe the interviews and observations of the day.

We systematically analysed the data after the completion of data collection through the standard practice of coding, sorting, local integration, and internal integration (Weiss, 1995). We adopted an issue-focused approach to data analysis within each site. The initial codes were a mix of inductive and deductive codes (Weiss, 1995). We followed Miles and Huberman (2014) to develop codes clustered together as meta-codes or pattern codes.

The following sections describe the state-centric and worker-centric understandings of Open Government Data. Section 3 describes two key OGD initiatives within MGNREGA. Using document analysis of state’s archives, we show how the distant state assumes that the workers will use the OGD initiative to confront the everyday state. However, as our ethnographic findings suggest (Section 4), workers do not always see themselves as activist-citizens confronting the everyday state.

3.Seeking data like a distant state: OGD and the quest for legibility

The state plays a significant role in addressing many development challenges concerning poverty, malnutrition, health, rural roads, social security and housing. Despite substantial budgetary allocations and spending, a perpetual concern has been the on-ground implementation of such programs, especially by the lower officials who directly engage with the citizens. Such concerns are evident in the official documentation and the detailed accounts of social movements working on governance. The state has often responded to such concerns by datafying these programs to make the functioning of the everyday state legible. A large number of datasets belonging to many government programs, for instance, on education, health, nutrition, livelihoods, and agriculture, are now available in the public domain. While the publication of such datasets is significant, we show how the datasets represent state-centric thinking on OGD. Following James Scott’s “seeing like a state”, we show how the datasets seek to ensure the legibility of the everyday state and citizens.

As described earlier, MGNREGA is considered landmark legislation for its rights-based approach and has an explicit commitment to enable citizens’ participation and government transparency. Direct participation is encouraged by giving importance to the Gram Sabha, an assembly of all adult voters. Such forums are intended to provide opportunities for citizens to seek information and justification from elected members and officials. MGNREGA further provides a legal right for citizens to access key government documents such as the wage list and muster roll. The program design also provides every household with a “job card”, a key document which contains the details of the household with the program. The commitment to transparency is so explicit that it was often called the sister legislation of India’s Right to Information Act, 2005 (cited in (Mathur, 2012)).

Despite the existence of such transparency measures, concerns about corruption began to emerge in the initial few years. For instance, the lower officials and elected members would falsify data to include non-existent worksites and misappropriate the expenses. Similarly, workers would be paid less than their wages due to them. Non-existent workers (known in official documents as ’ghost beneficiaries’) would be included in the program to misappropriate their wages. The distant state responded to these challenges by deploying digital technologies, which enabled higher officials to monitor the functioning of the everyday state. Further, such administrative data began to be published online to ensure citizens’ participation in monitoring the program. We describe two such initiatives relevant in the context of Open Government Data – the NREGASoft information system, which publishes data on workers and worksites and the Janmanrega app, which visualises worksites on a map.

4.Seeing “data” like a worker

As noted in the previous section, many of the datasets on MGNREGA are proactively published online. Such datasets have been widely used by both the officials within the state and by reformist activists outside the state. Discussions based on NREGA data have taken place in multiple forums, including academia, media, parliament, judicial courts and most importantly, the villages. While we recognise the importance of data to debates on MGNREGA, our key focus for this research is the use of data in the village, especially in the engagement of ordinary people with the everyday state.

There is evidence, especially based on ethnographic studies, of social activists mobilising village communities to question the everyday state (Sharan, 2021; Veeraraghavan, 2021). However, our findings (focussed in this paper on workers and activists rather than other significant actors such as bureaucrats) attempt to shed light on the participation of citizens in the last mile, especially in regions where there might not be sustained investment from organised social movements. Also, as we show below, social activists often temper their normative ideals based on their understanding of workers’ lived realities. Thus, the social activists shift their emphasis from a right-based anti-corruption agenda to somehow “getting things done”. We elaborate below on the subtle distinction between an anti-corruption agenda and getting access to one’s own entitlement. We also highlight how the OGD might not be helpful in such contexts and might have to be reoriented to contribute to citizen participation. There are three key reasons why the current design of OGD is not being found directly relevant by workers. The nature of misappropriation is blurring the difference between legally correct and incorrect data. Two, the costs of confronting the state on corruption continue to be non-trivial and have severe repercussions for workers, as evident in our findings but also supported by other ethnographies. Three, the social acceptance of corruption takes away the moral right of citizens to demand accountability from the elected members. The implication is that workers value data that will enable them to identify the “face and place” of officials and office, to “get things done”, and to better negotiate with the everyday state.

5.Conclusion

Governments worldwide have begun to proactively publish hitherto inaccessible datasets online, throwing light on the functioning of their different interventions in society. Such initiatives hold promises of enabling citizen participation, improving state accountability and contributing to achieving developmental outcomes. Multiple literature reviews including Safarov et al. (2017) and Francey and Mettler (2021), however find very little empirical evidence to such promises of OGD. They, therefore, suggest that “qualitative studies using in-depth interviews may be able to trace causal mechanism between the utilization of certain OGD initiatives and its effects”(Safarov et al., 2017). Responding to such a call, we have sought to understand the role of OGD from a socially embedded perspective (Avgerou, 2010) in enabling citizens’ participation. While we recognise the role of OGD in contributing to debates in the arenas of academia and media, our specific focus is on its role in shaping ordinary citizens’ engagement with the “everyday state”.

Our findings are based on the ethnography of India’s livelihood program, the MGNREGA, in the rural dryland district of Karnataka, in south India. Borrowing from James Scott (1999), we highlight how a “seeing like a state” logic underlies the design of the OGD initiative. With digital information systems, the metaphorical “seeing” has become real with unprecedented legibility through direct, unmediated access to digital records. The state-centric design of datasets often consists of aggregate datasets on the performance of different program attributes. Such datasets might be useful for researchers or data journalists to contribute to policy analysis. Yet, they are of little value to ordinary citizens in their engagement with the everyday state. Further, even when the datasets are granular, they are administratively ordered. By this, we mean that the key assumption of the datasets is to fight corruption, especially by the everyday state. In the process, the datasets introduce a “techno-official language” often far removed from the everyday realities of the citizens.

We show how such assumptions underlying a state-centric OGD might not match the lived realities of ordinary citizens who do not share the anti-corruption agenda for at least three reasons. One, the changing nature of corruption has blurred the difference between legally accurate and “fudged” data. Social movements have often sought to confront the everyday state using access to such “fudged” data containing “ghost” and duplicate entries (Sharan, 2021). However, with the practice of “adjustment”, administrative data is free of such errors while perpetuating newer forms of corruption. Second, the costs of confrontation with the everyday state and the local elite are non-trivial. Challenging the well-entrenched practices of the local elite can have violent consequences. Our findings reiterate other similar ethnographies on the use of data to confront the local elite (Roy, 2018; Sharan, 2021; Veeraraghavan, 2021). Finally, the moral economy of corruption implies that confronting the everyday state on corruption doesn’t have much social appeal. As Webb (2012) points out, citizens might not always share the idea of an “ethical citizen” who must be rescued from the “dirty politics” of parties, patronage, and brokerage.

Instead of confronting the everyday state, the citizens often prefer to “get things done”, i.e. access their entitlements (Ruud, 2001). Citizens are often caught in an administrative maze and value knowing the official and the office to negotiate and “get things done”. Yet, the state-centric OGD doesn’t support such efforts mainly because it is designed to make the functioning of the everyday state legible to the distant state. The administratively ordered datasets containing aggregates are of little value to ordinary workers in getting their work or payments on time. Even when granular data exists, it is presented in a distinctively “techno-official” language. A parallel can be drawn between the OGD and the Right to Information Act. Sharma (2013) shows how the practices of RTI “bureaucratises social lives and activism, forcing ordinary individuals to become competent in a specific kind of technical literacy”. The OGD requires competence with the “digital” in addition to the “official” keywords drawing the citizens into the language of the state. Further, the citizens are required to “speak like a state”, i.e., grapple with the techno-official language to engage with the datasets put out by the state.

One of the key implications of our findings is the need to reorient OGD initiatives towards citizens’ needs. While we acknowledge the relevance of state-centric OGD, we suggest the need to complement it with a citizen-centric framing. Our findings also highlight the role of social movements in preparing the ground for the use of OGD. A citizen-centric OGD and social movements can mutually reinforce each other. Our findings, especially on the blurring of differences between legally accurate and fudged data through the creative process of “adjustment” caution us against “data worship”.1111 Such caution is warranted especially when data is used to claim policy successes prematurely, including by the state. Finally, our study highlights the need for a research agenda on OGD from a socially embedded perspective, especially from underrepresented regions that do not have a strong activist network.