Collaboration without consensus: Building resilience in sustainable agriculture through ICTs

Abstract

In this article, we draw on a case study of a weather information system in West Bengal, India to examine how the design and usage of ICTs can play a much larger role in steering holistic changes in everyday practices of smallholder farmers as they cope with effects of climate change. In doing so, we focus on the role of the information system in forging collaborative practices across diverse set of actors with varied interests. We show how collaborative practices facilitate village capacity building to make effective use of outside expertise, convert them into locally relevant knowledge, and imbue them with new meaning in their everyday practices. We argue that these capacities are a necessary means by which smallholder farmers can build resilience in the face of the global climate crisis.

Keywords:

• Capacity
• collaboration
• ICT4D
• resilience
• sustainable agriculture

Introduction

Global climate change poses a serious challenge for agriculture – particularly for small and medium holding farmers in developing countries. They find themselves especially vulnerable in the face of changing natural conditions, with limited resources to adapt. While mainstream approaches to agricultural development – rooted in the Green Revolution of the late 20th century – have assisted a large number of people in rural areas to improve their incomes and livelihoods, they were not designed to address the needs of the most marginal areas for agriculture (Pingali 2012). The high costs of mainstream approaches (e.g., “Training & Visit”) in face of limited budgets and insufficient political support has long weighed down agricultural extension services, making it near impossible for them to reach the large majority of small-scale and subsistence farmers in low and middle income countries (Feder, Willett, and Zijp 1999). For instance, in India, despite adopting a variety of approaches to agricultural extension, less than 10% of farmers have accessed and made use of the agricultural information and resources the government provides (Glendenning, Babu, and Asenso-Okyere 2010). In addition to the struggle to reach out to farmers across large geographical distances, delivering sufficiently localized advice has been a big challenge for extension services and for long there has been a discrepancy between the information they provide and what information farmers need (Anderson and Feder 2004; Glendenning, Babu, and Asenso-Okyere 2010; Pingali 2012). More importantly, in the context of the climate change, when it comes to ecological, social and economic sustainability, the practices promoted by agricultural extension programmes have shown limited or sometimes even negative impact on smallholding farmers (Altieri 2002; Altieri and Nicholls 2005; Evenson and Gollin 2003; Francis et al. 2003; Parayil 1992; Pingali 2012; Pretty 1995).

In response to these challenges, there has been a great deal of interest in finding ways in which information and communication technologies (ICTs) can help address agricultural development. In the last two decades, following the wide-spread availability of mobile phones, significant efforts have been made to use them to support farmers (Aker 2011). Tele-centers, Internet Kiosks, and call-centers were established to help farmers make use of new information channels (Cecchini and Raina 2002; Das, Basu, and Goswami 2012). SMS-based and interactive voice response (IVR) systems such as, Reuters Market Light (Fafchamps and Minten 2012) and Avaaj Otalo (Patel et al. 2010) disseminated crop advice, market price information, and also connected farmers to extension agents and each other. Other initiatives included the use of participatory video production and screening (Gandhi et al. 2007), multimedia systems (Dearden, Matthews, and Rizvi 2011), and social networking (Medhi-Thies et al. 2015). With advances in technology, we now see systems operating on large geographical scales – such as whole regions or even countries (Cox 2002; Gomes 2009; Raghavan et al. 2016). They include geo-spatial analysis, sensor-driven precision agriculture, and model-driven big data analysis. When it comes to supporting climate resilience among farmers, services based on agrometeorology and climate science are becoming increasingly prevalent (Christensen, Ahsan, and Akand 2018; Chaudhuri et al. 2017; Sala 2009). They make use of meteorological and climate models to forecast weather and climate and disseminate it along with appropriate agricultural advice – information considered critical for farmers to be able to adapt to new climate realities.

While a large number of information services for agricultural development have been deployed (Gakuru, Winters, and Stepman 2009; Saravanan and Bhattacharjee 2014; World Bank 2011), their impact in terms of improved agricultural practices and correspondingly livelihoods has been limited (Aker and Fafchamps 2010; Aker and Ksoll 2016; Fafchamps and Minten 2012). One way to understand the limitations of these initiatives is to look at their underlying theory of change – the way they frame the problem of agricultural development and sustainability and how they see the technology intervention addressing that problem. Kendall and Dearden (2018b) identified “market efficiency” and “knowledge dissemination” as two common framings used for conceptualizing the role of ICTs in agriculture. In the “market efficiency” approach, the analysis centers on information asymmetries in the market for agricultural commodities and posits ICTs as a means of addressing them. While ICTs enable increased access to information, access does not necessarily translate into improved livelihood opportunities (Aker and Ksoll 2016; Cole and Fernando 2012; Fafchamps and Minten 2012; Kendall and Dearden 2018b; Srinivasan and Burrell 2015). For instance, provision of price information via ICTs does not necessarily fit with the decision-making practices of farmers and the situated complexity of their livelihoods (Burrell and Oreglia 2015; Srinivasan and Burrell 2015). On the other hand, “knowledge dissemination” approaches focuses on access to and quality of digital content (Aker 2011; Karanasios and Slavova 2019). However, researchers have drawn attention to differences between the way farmers learn and the way ICTs are being leveraged to deliver knowledge (Oreglia 2013; Wyche and Steinfield 2016). The limited success of ICT-based initiatives suggests that both the market efficiency and knowledge dissemination approaches have significant shortcomings.

Taken together, the crisis of sustainability in smallholder agriculture; the limitations and difficulties of mainstream extension services; and the shortcomings of ICT-based initiatives suggest that there is a need for alternative approaches to ICTs in agriculture (Karanasios and Slavova 2019; Kendall and Dearden 2018a, 2018b, 2018c). Agricultural development efforts need to incorporate critiques of the top-down and universalizing practices of rural development in general (Altieri and Nicholls 2005; Chambers and Conway 1991; Pretty 1995) and of ICT-based initiatives in particular (Burrell and Oreglia 2015; Kendall and Dearden 2018b; Srinivasan and Burrell 2015; Wyche and Steinfield 2016). They also need to address the question of long-term sustainability of the agricultural system, especially how any intervention contributes to the ability of the farmers to cope with and adapt to the external shocks or changes that climate change brings.

In this article, we contribute to the discussion on the role of ICTs in development of sustainable agricultural practices by examining the design and usage of an agricultural information system (IS) for agrometeorology in West Bengal, India. We specifically seek to examine this case through two main concepts, namely, resilience and collaboration.

Resilience and collaboration in sustainable agriculture

We define resilience as presence of three, interrelated, capacities in a system. One, the capacity to withstand external shocks or changes (Ospina and Heeks 2010; Heeks and Ospina 2019), e.g., the ability of farmers to take actions – such as protecting crops or livestock – in the face of an extreme weather event. Two, the capacity to recover from such events, e.g., having a financial buffer to re-sow or replace damaged crops. Three, the capacity to change in response to new external conditions – the capacity to re-actively respond to changes that are chronic in nature and also proactively make changes to forestall adverse affects in the long-run (Folke 2016). For farmers, this can entail changing cropping patterns, delaying seasonal planting or diversifying their seed varieties. While our ultimate aim is to understand how information systems can enhance agricultural resilience, in this article we draw on intersectional literature on sustainability, agriculture, and ICTs to identify the collaborative practices that provides the basis for such resilience.

Collaborative approaches to improve social and institutional learning are necessary for building the capacity for resilience (Berkes 2017; Kark et al. 2015; Pahl-Wostl et al. 2007), as they bring together heterogeneous and multiple actors with conflicting yet shared interest (Cole 2015; Ostrom 2009; Termeer and Bruinsma 2016). It requires breaking of routine boundaries and bridging them by effectively translating needs and interests of different stakeholders across the boundaries (Termeer and Bruinsma 2016). ICTs could facilitate this by creating a support infrastructure for collaboration between heterogeneous actors (Nyella and Honest 2015; Puri 2007). At the same time, one needs to be mindful that collaboration does not necessarily mean a depoliticized or harmonious approach to development (Escobar 2011; Ferguson 1990; Li 2007). Rather, it calls for a careful tackling of organizational power structures, local politics, social hierarchies and so on (Cooke and Kothari 2001; Mosse 2004; Slater 2013; Williams 2004).

We document the ways in which the forementioned agrometeorology IS in West Bengal is supporting collaborative practices across three sets of actors – Development Research Communication and Services Center (DRCSC), a non-governmental organization, small-holding farmers, and technical experts who generate content for the IS. We ask whether these three actors could forge collaboration through their involvement with the system, what were the nature of these collaborative practices, and in what way the IS allowed these practices to emerge.

Methodology

DRCSC, which conceptulized and managed the information system, has around 30 employees at its Kolkata head office and another 60-70 at field office locations across 11 districts in West Bengal. The weather information system was conceptualized to produce, curate and disseminate localized weather forecasts and agricultural recommendations to smallholder farmers. Day-to-day activities include the collection of raw weather data, use of such data to generate five-day forecasts and curate locally relevant agricultural recommendations, as well as dissemination of this information to farmers. The system was implemented in two blocks in two separate districts¹ located agroclimatically in the dry-zone part of the state. We collaborated with DRCSC but conducted the study independently.

The IS was conceived of as a way to support marginal and small-scale farmers adjust to the impacts of climate change. It was funded by the UN Adaptation Fund, in India administered by India’s National Bank for Agriculture and Rural Development (NABARD). The proponents – DRCSC – advocated the project in terms of issues such as water availability, drought, extended monsoon rain breaks, depletion of natural resources, and erratic climate. In the project proposal they stated that “crop-weather advisories will help to reduce the risks and damages caused by climate change. It will capacitate the farmers to take more effective decisions regarding farm management”(DRCSC. 2013, 28). Further, they highlighted the need for engagement with scientific experts to minimize crop losses and optimize the use of inputs such as fertilizers, irrigation, seed, and pesticides. To explain the need for a locally managed weather system, they pointed out that the weather information from the Indian Meteorological Department (IMD) was only available on a district level² – it was not fine grained enough to be actionable by farmers. DRCSC also noted that it was hard for farmers to translate forecasts into agricultural practice.

We employed participant observation, formal and informal semi-structured interviews, group discussions, and also gathered secondary material such as project documentation, pictures, and working materials from the field offices. We sought to triangulate data from different geographical locations (head office, field offices, and villages), actors, and data sources. While we had access to most of the sites and system-related data, we faced limitations when it came to our interaction with the experts. The project’s technical expert, a meteorologist located in a different state, was also the point of contact for other agricultural experts involved. Consequently, the project team relied on the meteorological expert for providing weather forecasts as well as corresponding agricultural advice. We interacted with this expert on multiple occasions over three years through face-to-face and telephone interviews as well as in workshops. While these interactions with the meteorologist informed this article, we could not engage in participant observation of this expert’s work practices. The expert’s work was obscured not only to us but also to the team at DRCSC which managed this initiative. Hence in our analysis, we only include the parts of the expert’s work which we could observe through our engagement with the head and field offices of DRCSC. To help validate our interpretations we shared and discussed them with villagers, field offices, and the head office staff.

Our focus was on three primary groupings: staff and volunteers working on behalf of DRCSC, the farmers who were the intended beneficiaries of the system, and the experts involved in interpreting the collected weather data and generating forecasts. In the case of DRCSC, we studied head office staff based in Kolkata as well as field office staff and unpaid volunteers at the field sites in the two different blocks. The volunteers were villagers selected and trained by DRCSC for tasks related to managing collection and dissemination of weather data. In some cases, they were also involved in farming. The field officers were full-time, paid, staff of DRCSC, primarily recruited from the local area, with some coming from other districts. We should keep in mind that these groups were neither cohesive nor mutually exclusive.

We began the study by mapping the system, first with the head office staff and then with the field office staff. We employed collaboratively created rich pictures (Monk and Howard 1998) to document different actors, physical spaces, and processes undertaken to generate and disseminate the forecasts. Based on this mapping, we then identified individual actors for deeper follow-up interviews and observation. We decided to focus on five villages from one of the two blocks where the system was operational. Our selection criteria were based on these villages’ access to the system (e.g., whether there was an automated weather station nearby, whether there was a local volunteer managing a manual weather station, whether there was an information display in the village) and their participation in the practices of the system (e.g., whether the village had provided a local volunteer, how regularly the forecasts were posted, whether or not there were frequent group meetings). Three of the villages had a high degree of access to the system and also high degree of participation, the other two villages had a low degree of access and participation in the system. Additionally, we conducted shorter formal, semi-structured and informal, unstructured interviews with farmers across the remaining villages where the system was operational. Furthermore, we conducted participant observation of field office staff as they collected data, received, and disseminated forecasts and recommendations to farmers’ groups. Interviews – both formal and informal – focused on different actors’ participation in the system, their role in data collection, and their use of the information generated by the system. We sought to inquire into interactions between different actors in the system, the meanings they attached to the system, and changes following the introduction of the system.

During a total of 6 months engagement, approximately one week per month was spent in the districts where the system was being implemented. In each of the villages studied, we attended at least three monthly meetings of both female and male farmers’ groups – in which an average of between 10-15 group members participated. During these meetings we conducted group interviews and engaged in participant observation. In addition to group interviews, we conducted shorter informal, semi-structured individual interviews with 35 farmers. During these interviews – which were not recorded due to farmers’ discomfort with recorded interviews – we discussed their farming practices and landholdings, their knowledge and use of the system, their understanding of its operating procedures, and their relationships with other actors in the system. All farmers interviewed belonged to either Scheduled Tribes, Scheduled Castes or Other Backward Classes³ and had small landholdings on which they primarily engaged in rain-fed, marginal farming – the major crop being monsoon season rice. In many cases they also kept a small number of livestock and maintained kitchen gardens for growing vegetables for personal consumption.⁴ The two districts where the system is present belong to the dry agroclimatic zone of West Bengal, and were therefore heavily dependent on monsoon rainfall. In the preceding couple of years, the farmers interviewed had increasingly faced issues of non-seasonal and insufficient rainfall, and this had provided one of the motivations for the introduction of the IS: to bring some sense of stability in local agricultural practices.

In addition to participant observation conducted across all sites (villages, field office, head office), we conducted semi-structured recorded interviews with a total of nine individuals in different roles − 8 people employed by DRCSC (3 from the villages, 3 from the field office, and 2 from head office) and the meteorologist. These semi-structured interviews were transcribed, and some were translated from Bangla to English to aid analysis. When it came to farmers, we drew on a total of 35 informal interviews and 3 group discussions for this article. Three years after this fieldwork, we conducted three more interviews to follow-up on the development of the project. These interviews were with the head office project manager, the field office project manager, and the meteorologist. Material such as project documentation, generated forecasts, and agricultural recommendations were also collected and cross-referenced with the information gathered through interviews.

We analyze: (a) information processes designed to operate for diverse actors; (b) the information practices that emerged when these actors engaged with the system; and (c) the meanings these actors attached to the system.

Information processes in the weather information system

The information system was designed to involve multiple actors:

• a consultant meteorologist located in a different state

• agricultural experts consulted by the meteorologist

• coordinators (staff) based in the head office in Kolkata

• field office staff (project manager and community mobilizers) located in an office in each block

• village volunteers based out of their home villages

The starting point for the system was its weather prediction model. As input to this model, local weather-related data was collected. This data collection process took two different forms, centered on two different artifacts – automated weather stations (AWSs) and manual weather stations (MWSs). AWSs were digital devices that collected data on rainfall, wind speed, temperature and humidity (Figure 1). They were installed in three locations in each block. Weekly, the field staff would go to the weather station to download the data on to a laptop. The MWS set-up consisted of a rain gauge and a small hand-held temperature and humidity monitor (hygrometer) (Figure 2). Village volunteers would take note of the readings twice a day and community mobilizers would photocopy these notes on a weekly basis. This process resulted in new artifacts – spreadsheets generated by the AWSs and notebook pages generated by the volunteers. These artifacts would then be sent to the meteorologist to become part of two other information processes; first, as input into the prediction model-based process of generating a 5-day forecast, and, second, as a benchmark to evaluate the accuracy of previously generated weather forecasts.

Figure 1. Automated weather station.

Figure 2. Manual weather collection: device and notebook.

Once the 5-day forecast had been generated, a new information process would begin. The meteorologist would consult agricultural experts to create a locally relevant list of problems that might emerge, along with remedies in the form of specific plant treatments, fertilizer and pesticide applications.

When the agricultural experts had formulated their recommendations, DRCSC translated the recommendations into practices that aligned with the kind of agricultural practices they promoted. DRCSC staff members would remove specific references to pesticides and fertilizer and replace them with instructions for the application of bio-pesticides such as neem oil and trichoderma. They would also remove references to crops more commonly cultivated by commercial growers than the small-holding farmers they worked with. Combined with the 5-day forecast, this edited recommendation would form the final agrometeorology advisory (Figure 3).

Figure 3. Sample agrometeorology advisory: weather forecast and agricultural advice.

Finally, they would send the combined agrometeorology advisory to the field office staff who would begin the dissemination process, which consists of two steps. First, they would transmit information via SMS (Figure 4) and WhatsApp to both farmers and village volunteers. Then, on receipt of this information, the field staff would make printouts of the advisories (Figure 5) and distribute them to the villages. In each village a volunteer would take the printout and write the information on a DRCSC-installed blackboard in central location in the village (Figure 6). While DRCSC had originally planned to use SMS as the main vehicle for dissemination, challenges in encoding Bangla characters on the farmers’ “Chinese”⁵ phones compelled them to search for other solutions. As a result, DRCSC placed greater emphasis on the use of blackboards, group meetings, and printouts to disseminate the weather forecasts and reports.

Figure 4. Sample SMS message.

Figure 5. Printouts of weather report.

Figure 6. Village volunteer updating blackboard.

As an additional step, community mobilizers and volunteers would bring forecasts and recommendations to meetings in the villages where they would discuss them with the farmers. In the villages separate meetings were held for men’s and women’s groups, each with about 10 to 30 members. Through these meetings as well as individual interaction with villagers, community mobilizers and volunteers would collect feedback on the accuracy and usefulness of the agrometeorology advisories. Furthermore, they would collect “case stories” providing narrative accounts to raise awareness and mobilize support for the project and DRCSC’s mission.

In Figure 7 we illustrate the workflow of the IS information processes, as they had been formally conceptualized and designed by DRCSC. To understand the consequences of these information processes we need to look at two things; first, how these information processes became embedded in the everyday practices of those involved; and, second, how and why these practices were meaningful to them.

Figure 7. Information processes in the weather system.

From information processes to information practices

In this section, we discuss the everyday practices that take place around the system. We focus here on the sites that we could directly observe through our participant observation.

In the head office

At the head office, we found two main practices of the staff. First, they would identify and coordinate with relevant actors who could produce weather forecasts and provide agricultural recommendations – the meteorologist and agricultural experts. Second, they would curate and edit the recommendations received from the meteorologist and agricultural experts.

In the case of weather forecasts, they would translate them from English into Bangla. The head office staff would pick appropriate words and expressions that could be easily interpreted by the farmers. In the initial phase of the project, they had worked together with the meteorologist to identify a vocabulary of colloquial terms that could be used in the forecasts. In effect, there are two levels of translation – first, from English to Bangla, and, second, from formal Bangla to local idiom. In this work, DRCSC faced a challenge in finding terms that could balance the specificity and accuracy of expert language vis-a-vis its local appropriateness.

In the case of agricultural advisories, recommendations were based on conventional agricultural practices, such as identifying specific pests that might emerge during certain weather conditions. Responses to these problems from the experts would involve application of specific pesticides, appropriate synthetic fertilizers, and cultivation practices. However, this would often be contrary to the agricultural practices DRCSC sought to promote. Therefore, the head office staff would modify the experts’ recommendations to align with their preferred practices of agroecological farming. This process was not without difficulty, as there was no natural mapping between agroecological and conventional agricultural practice. The head office staff, informed by agroecology, rather than targeting specific problems related to pests and disease, would encourage farmers to address underlying systemic issues of ecological interactions and farm design. Thus, DRCSC provided very little pest- or disease-specific advice.

There were constant debates among DRCSC staff about agroecological practice not matching well with reductionist “problem-solving” approaches of conventional agriculture. At the same time, they would lament the lack of availability of specific, scientific knowledge and advice to provide. In the ranks of DRCSC management there was one senior staff member who was very experienced in agroecology. When the programme was first implemented this agroecology expert would be sent the edited recommendations by the team leader implementing the information system to provide feedback on them. However, this practice was discontinued after a while. When asked about it, the expert noted that the implementing team had difficulty with their critical view of the agro-advisory recommendations. The implementing team preferred the more targeted and specific advice provided by the agronomist to the more holistic and general advice that the expert on agroecology could provide.

The process of creating the agrometeorology advisories as outlined in Figure 7 may seem straight forward. However, when looked at through the everyday practices of the actors, it highlights the politics of information and translation. Decisions like what words are appropriate for communication at different levels, whose understanding is prioritized, and which worldviews⁶ are preferred are political in nature. This also leads us to question who makes these decisions and where these decisions are made. For example, the information about relative humidity, translated into the Bengali word “adrota” in the weather forecasts report, needed further translation into concrete implications such as amount of dew or evaporation. We see the choice of only using the formal word “adrota” in the weather forecast as a product of a top-down process of translation.

Moreover, the head office staff was responsible for sending SMS messages to farmers to disseminate the advisories. Initially, since the plan was to use SMS as the main communication channel and for this purpose, they collected a large number of phone numbers from farmers. However, they soon realized that sending bulk SMSs not only entailed a huge cost but was also ineffective in reaching their intended beneficiaries, as mobile phones were not as ubiquituous as they had thought and, as we mentioned earlier, their screens could not adequately display the advisories. Furthermore, a large number of farmers – especially female farmers – were illiterate which added to the difficulties of accessing forecasts via SMS. Hence, the head office opted to send a much smaller number of SMSs to “lead farmers”⁷ and hoped that they would in turn share with others in the village.

As the information process evolved, many of the practices undertaken by the head office gradually started moving to the field office. Therefore, the practices that we will discuss in the next section have some overlap with those we have just covered.

In the field office

When taking the recommendations and forecasts to the farmers, community mobilizers would reinterpret and re-frame the recommendations on the basis of their own understanding of what would be appropriate for their farmers’ groups. For instance, we observed a community mobilizer introducing the system to a group of female farmers, describing the information that would be provided and giving examples of uses to which they could put it. Rather than discussing the agricultural recommendations provided by DRCSC in the IS, he told them that the IS provided details on rainfall and that forecasts of rainfall could help the women decide when to repair their mud houses.

Project managers, who were recruited because of their skill and comfort with computers and other digital technologies, tended to be from outside of the region, unlike most other field office staff members. With these skills, they collected and managed the data from the automated weather station and also collated manual weather station data gathered by community mobilizers via the village volunteers. This put them directly in touch with the meteorologist and made them part of the communication channel between the head office and the experts. Consequently, every week they received the forecast and agricultural recommendations at the same time as the head office did. This direct access to the expert advice, in practice, made it possible for project managers to edit and prepare the final agrometeorology advisories at times when the head office staff was busy. On such occasions, the project managers – without being requested to do so – edited the advisory following the patterns of edits done by the head office previously. Eventually, editing and translating of agrometeorology advisories became the responsibility of the project managers with minimal supervision from the head office. This essentially moved the site of translation practices discussed above.

A responsibility of the community mobilizers was to organize regular farmer group meetings, separate ones for males and females. To mobilize people for these meetings, they would be in touch with local self-help group (SHG) members who would in turn invite others to join. During these meetings, community mobilizers together with village volunteers would facilitate discussions about projects undertaken by DRCSC and the male/female SHGs, as well as provide short trainings, demonstrations, and instructions on current agricultural concerns. They would also bring a printout of the agrometeorology advisories which they would discuss with the farmers’ groups and relate them to farmers’ everyday practices.

The majority of the community mobilizers recruited for the project were from the local farming households. This overlapping identity – both targeted beneficiary and implementer of the project – enabled them to draw upon community- and implementation-specific knowledge (Puri 2007)⁸ to balance the interests of DRCSC and the beneficiaries when they did not align.

In the villages

Apart from project managers and community mobilizers, we observed two other sets of actors in the villages: village volunteers and farmers, the intended beneficiaries of the system.

The meteorologist’s need for detailed, village-level data on weather, provided opportunity for the village volunteers to become involved in data collection and administration. For several volunteers, their work as volunteers also involved building relationships within the village. In case of one volunteer, she would regularly visit the village head’s (the majhi ⁹) house to take readings from the manual weather station installed there. In course of these visits, she became acquainted with the majhi and regularly discussed the data with him. In another case, a volunteer noted that he was now often invited to discuss weather and other issues with the elders of the village.

As a result of the regular work of updating the public displays of weather data – primarily blackboards and printed A4 sheets, the volunteers became publicly associated with the system in the village. Other villagers would therefore ask them questions, which they sometimes could answer themselves and sometimes use their connection to DRCSC to provide answers for.

Other village volunteers combined the data collected on behalf of the meteorologist with traditional means of weather forecasting – primarily based on wind speed and direction – to create their own, independent forecasts. In this way, the village volunteers could develop the skill of bringing together their intuitive understanding of weather with data, as well as the skill of articulating such information for sharing. As we followed some of these volunteers over a longer period of time, we observed that they became increasingly confident in their job and the way they handled their relationships with the rest of the village.

However, this process also gave rise to conflicts. In one village, the villagers felt that the volunteer kept information to himself. They observed his notebook and the fact that this information was not accessible to them – only the weekly updated agrometeorology advisories he wrote on the blackboard. Furthermore, the villages inferred that his preferential access to agrometeorology information would also mean that he received other resources or benefits from DRCSC. Probing further into this issue, we were told that while he was invited to meeting and events held by DRCSC through which he received additional training and information, he was not provided any material resources or support that others in the village would not have access to as well.

When considering the farmers’ use of the system, it was clear that the intended users or beneficiaries hardly formed a cohesive group with similar needs. Male farmers were primarily concerned with the main rice crop, as they migrate for work outside the region during other less profitable cropping seasons. Accordingly, they focused on using rainfall forecasts to know when to plant, when to transplant, when additional irrigation may be required and when to harvest. Female farmers’ responsibility, on the other hand, was for livestock, children, and kitchen gardens. This meant that their information needs differed from those of male farmers. For the women, temperature forecasts were important. Based on these forecasts, the women could decide whether to let their children walk to school or take livestock out for grazing.

Male and female farmers also differed in the way they accessed the information provided by the system. When it came to the blackboards these were relatively inaccessible for female farmers for two reasons. First of all, illiteracy was more prevalent among female farmers. Second, all of the blackboards were placed in public spaces which were male dominated. While men could “hang out” in such public places reading and discussing the information, women could only see the information in passing. For SMS, only a few of the male farmers acknowledged having received or seen them all. More than access to mobile phones – which was anyway a problem across all farmers – illiteracy was a major deterrent, particularly for female farmers. In fact, one interesting suggestion given by some of the female farmers was to transmit the agrometeorology advisory as an advertising jingle over the phone.

Other villagers, who worked as daily wage laborers, sought to use the system to determine whether there might be work in nearby larger towns or not – avoiding spending time and money to travel long distances to worksites if the weather conditions were adverse to the construction work they sought.

The provision of not just agricultural recommendations, which were aligned with the goals and DRCSC’s mission, but also forecasts of temperature, rainfall and wind speed, allowed the system to serve the information needs of different groups of farmers, not just those who practice organic agriculture. For example, many farmers would use the information and pair it with conventional agricultural practices, such as knowing when to apply fertilizer or spray pesticides. When asked about the advice provided, these farmers would often note that they lacked the necessary knowledge or organic inputs (such as vermicompost or neem oil) to implement the agroadvisory. We also observed that even if they theoretically had the resources or the knowledge to apply some of the recommendations given, farmers reported that the recommendations were not relevant to them as they were not “organic” farmers. Despite this, they would still value the system because they could use the provided data to enhance their own practice. The farmers who reported that they applied the agroadvisory often had been in contact with DRCSC for a longer time – they recognized the type of advice given and were open to applying it as they identified themselves as its beneficiaries.

Interpreting the information system

There were both overlaps and diversity in the meanings, purposes, topics, and questions through which the actors engaged with the system. The relationship of actors and the meanings they attached to the IS could best be described as many-to-many.

For the leadership of DRCSC, the goal of the information system was climate change adaptation and promotion of sustainable agricultural practices. Specifically, they sought to promote agroecological practices that emphasized use of bio-pesticides, natural fertilizers and drought tolerant crops. To implement the project, DRCSC set-up collaborations with a meteorologist who in turn consulted other experts. These actors all broadly shared the view that climate change adaptation was important and that the region was especially vulnerable.

However, the way they interpreted the role of IS in achieving this goal differed. The meteorologist sought improved weather modeling on a micro scale – seeing IS’s role to be generation of more accurate data that could help tackle the problem. He also collated agricultural advisory based on the conventional paradigm of agriculture, which viewed the IS as a means to improve agricultural productivity with reduced and more appropriate application of fertilizer and pesticides. In contrast, DRCSC’s expressed aim was to entirely eliminate use of synthetic pesticides and fertilizers, and have farmers adopt entirely organic farming practices. In their view, productivity was not the main contributor to resilience to climate change, but rather the adoption of long-term sustainable practices. This is not to say that DRCSC had only a single voice, as we have noted earlier, there was serious debate about how to promote sustainable agricultural practice. For many in DRCSC, immediate livelihood concerns – especially increases in short-term productivity – were their primary responsibility, as opposed to long-term natural resource management. This dilemma manifested in the way the agrometeorology advisories were interpreted as well as edited in the head office. Similarly, in order to realize the goals of the IS, some of the staff members would say that the farmers’ participation was of greater importance, whereas others would emphasize the need for expert advice.

This was a continuous debate within DRCSC – connected to much larger issues. Since its foundation, DRCSC’s primary concern has been sustainable management of natural resources and sustainable livelihoods for marginal farmers. However, increasingly the funding ecosystem was inclining toward issues of global climate change, which demanded a different approach to natural resource management. While they retained most of their regular practices within this new approach, they nonetheless were compelled to add interventions that directly addressed climate change. The weather information system project was a product of this new approach. It introduced additional complexity to the dilemma of balancing environmental concerns and local livelihoods.

On the village and field office level, the system’s meaning and use became even more varied. The field staff, being more deeply involved in the organization, were more likely to address it in terms of climate change adaptation than the village volunteers. Among them, the social and educational background shaped what the information system meant to them. For example, for one of the project managers (recruited from an urban area and had multiple professional degrees) maintenance of the information system was his primary job task. Since he was recruited to maintain the system at a local level, his main concern was with the smooth operation of the information processes of the system. In contrast, for the community mobilizers recruited locally, they attached multiple meanings to the system. On one hand, they were trying to manage the system and achieve DRCSC’s goals as required by their job. On the other hand, coming from farmer families, they saw the potential of the IS in influencing their lives beyond this project, e.g., the system provided a way for them to find a role and build status within their community. This understanding of the IS was shared by village volunteers. By managing the system and speaking for it, they could gain status and be considered as knowledgeable, having access to expert information. Their overlapping roles in relation to the information system meant they had a greater stake in its success. This lead them to undertake discretionary practices as and when needed. For example, they would try different methods of dissemination beyond those imagined by the system. They would also understand the varied information needs in the village and translate the information accordingly so as to make it meaningful to the villagers.

While, for DRCSC, the system was embedded in a context of climate change and sustainable agricultural adaptation, for many farmers their needs and concerns were more practical. There was widespread recognition among farmers that the climate was changing and that traditional means of forecasting weather had become less relevant. Additionally, they recognized that they were in need for government or nonprofit support to adapt and develop their farming practices. As we observed earlier, the introduction of the system did not lead to major shifts in agricultural practices, rather farmers would find ways to re-intrepret the agrometeorology advisories into their existing practice. For example, if heavy rain was forecasted, farmers would choose not to irrigate – saving the considerable cost of fuel for the pump, as well as their limited freshwater resources. For the farmer, this was primarily an economic decision whereas for DRCSC it not only contributed to improved livelihoods but also sustainable resource management. In another instance, using the temperature forecast to decide whether to graze livestock or not, helped protect a critical asset for the farmer. Again, this is an economically motivated decision and not at all viewed through the lens of sustainability.

Furthermore, DRCSC’s conceptualization of the information processes reflects a masculine approach to agriculture – emphasizing the male practices around farming. These male practices centered on the annual rice crop. This bias got embedded in the system and was reflected in the kind of information that was considered important and the modalities of dissemination. We can see this in the relative difficulty the female farmers faced while trying to access the information system. They had to rely on their male counterparts for access to the information and on their own ability to reinterpret the information to make it relevant to them.

For both male and female farmers, the IS served to improve their existing practices – whether within agriculture or outside of it. Hence, through their own agency and in collaboration with the help of community mobilizers and village volunteers, rather than adapting their practices in accordance with the IS, they adapted the IS in accordance to their needs.

Collaboration in the system

The detailed understanding of the information system presented above now allows us to return to the two questions we are pursuing in this paper: firstly, whether all the actors, albeit with different stakes in and understandings of the information system, could forge collaboration through their involvement with the system, and, secondly, what was the nature of these collaborative practices. Based on our observations, we argue that these collaborative practices were critical for community level capacity and resilience building for climate change adaptation.

Collaboration between DRCSC and the experts

DRCSC’s incorporation of agrometeorology advisories into their activities originated from their recognition that smallholding farmers, who are affected by changing climatic conditions, could build resilience by using weather forecasts in their everyday agricultural practices. Moreover, with climate change emerging as a central concern in the development funding ecosystem, DRCSC was motivated to interweave most of their interventions on sustainable agriculture and livelihood into the bigger problem of climate change. Accordingly, they sought partnerships with experts, who would help them reorient their work to climate change adaptation. Even though these experts were not always entirely aligned with DRCSC’s mission and approach to sustainable agriculture, it nonetheless saw value in collaborating with them, as the scientific knowledge that they could provide was not available internally.

In case of the weather information system, the meteorologist provided DRCSC staff with training in how to collect weather data and also how to use forecasts in their work. Furthermore, he customized a weather model specifically for this project focusing on the two geographical areas in which the project was being implemented. Through this collaboration DRCSC could build capacity within the organization to work with weather data and forecasts in a systematic way. Similarly, in the process of translating the raw output of the weather model into locally intelligible terms, DRCSC built capacity both in the head office as well as in the field offices.

When it came to the agricultural recommendations, the challenge of translation was two-fold: first, from English to Bangla and from scientific terminology to local idiom, and second, problem-solution mapping from conventional practices to agroecological practices, e.g., when the meteorologist flagged potential for a specific pest attack and recommended application of a certain pesticide, DRCSC staff would replace the pesticide application with an organic remedy, if such a remedy existed at all.

The information system provided DRCSC opportunity to leverage the experts’ knowledge and modify it in the context of their work without compromising the value of anyone’s expertise. A critical enabling factor for this, was that the IS was under the control of DRCSC, without which it could not have set such terms of the partnership. This, however, was not a smooth process. DRCSC’s agroecology specialist was skeptical of the agricultural advisory that came from the external experts and voiced his opinion quite strongly. His critique of the conventional recommendations was perceived to be threatening for the collaboration and smooth operation of the process that generated agrometeorology advisories. Consequently, he was eventually dropped off from the communication and feedback loop between DRCSC and the expert.

Beyond the head office, field staff also had direct collaboration with the expert. They were trained by the expert, they actively participated in the data collection, and, finally, some of them were included in the back-and-forth communication between the expert and the head office. This meant, for example, that the field project manager could actively take on the job of curating agrometeorology advisories as and when required. As the information system allowed many-to-many interactions across various group of actors, the project as a whole created scope for collaborations that were critical for capacity building across multiple levels within DRCSC.

Overall, however, the project was premised on the idea that DRCSC lacked specific scientific expertise that was required to build and run a weather information system. This led to the system incorporating a top-down approach to create agrometeorology advisories where scientific experts took the lead – undermining internal expertise. Rather than compelling the outside experts to accommodate the organization’s own approaches toward agroecology, DRCSC took it upon itself to adjust and modify the scientific expertise into what it believed was more locally relevant information. This way of organizing the IS allowed DRCSC to balance between top-down expert knowledge and its own expertise in sustainable agriculture. While this balancing act was beneficial for the collaboration between heterogeneous actors, the apparent power asymmetry between the outside experts and DRCSC’s experts limited the extent to which the agrometeorology advisories could reconcile the discrepancies between their two distinct approaches to agriculture.

Intra-organizational collaboration

DRCSC supported minimal centralized control in two ways. First, field project managers were made responsible for collecting and sending raw data from the AWS to the meteorologist without any interference from the head office. Second, the failure of SMS as a large-scale dissemination channel compelled DRCSC to rely on its human intermediaries and their discretionary practices (Lipsky 2010) for their community outreach.

The implementing team at the district level had a three-tier structure. It had a field project manager and field staff who relied on a group of community mobilizers, who in turn sought the help of village volunteers. Further, community mobilizers and the village volunteers often were also intended beneficiaries. All in all, these intertwining and overlapping relationships and relative autonomy afforded by the system engendered relationships that were more collaborative than hierarchical.

However, within the head office, the conditions were different. There was little consensus on how the IS should be designed and run. In absence of agreement, the decision-making authority over the IS was unilaterally held by the head of the team responsible for the project. Consequently, the field office managers were under the direct control of this manager and had little interaction with anyone else in the head office, which limited the exchange of ideas, best-practices, and feedback.

Collaboration between DRCSC and the villagers

To engage with the villagers, DRCSC deliberately involved them in the selection and appointment of village volunteers in each village. The community mobilizers alone could not manage household level dissemination of agrometeorology advisories, which entailed activities such selecting the spot for display boards, installing them, and then updating them regularly, and using various formal and informal gatherings to spread awareness. With its long-term presence in these areas, DRCSC understood that such tasks would be best performed by somebody from the village itself. Accordingly, DRCSC designed a role for village volunteers to complement the role of the community mobilizers. While the community mobilizers were the paid staff of DRCSC, volunteers worked without any monetary compensation. This distinction, we found, was crucial in creating a more collaborative relationship between these two groups in absence of a formal authority structure. The volunteers were recruited on the basis of their ability to read and write at a high school level, skills necessary to manage the manual weather stations, and keep appropriate records. DRCSC was also perceptive of the local power structure and therefore initially approached the traditional village head – the majhi – to consult where the blackboards and MWSs should be installed in the village as well as who should be recruited to manage them. Once the village volunteers had been recruited, they continuously maintained and developed this collaborative relationship. While DRCSC had co-opted the majhi in the initial decision making, thereafter it built an independent relationship with the volunteers via training, support, and provision of resources. On their part, the village volunteers, as local residents and operating relatively autonomously, were able to balance between the interests of the village and DRCSC. This helped build a collaborative relationship between DRCSC and the intended beneficiaries that not only made the IS effective but also helped create long-term ties between the two parties.

Even though DRCSC was the main owner and manager of the IS it did not exercise strict control over how it was used. While it sought to promote its notion of sustainable agricultural practice, it viewed the IS as a relatively open space where the access, use, re-use, and reinterpretation of the information was multidimensional. For example, even though the expert advice was usually edited and curated at the head office, field project managers were not denied access to such information practices and eventually they took over this process almost entirely. Similarly, use of simple technologies, such as MWS and blackboards, for information generation and dissemination, also meant that fieldworkers and village volunteers could exercise their discretion in reinterpreting the advisories for local farmers and other intended beneficiaries. Rather than restricting its use to the farmers, they encouraged broad based adoption of the IS. They even highlighted – through their case stories – the various ways in which villagers had found the system useful. However, as we have already discussed, DRCSC’s initial framing of the IS as a way to promote improved agricultural practice did have some exclusionary effects. For instance, it meant that the roles played by the women in the villages and needs were poorly catered to in its design. In spite of this, thanks to the collaborative practices between DRCSC and the villagers, the IS could, to some extent, work around these limitations.

As we trace the IS through its designed processes, practices, and dynamic interpretations shared among all three set of actors – the expert, DRCSC staff, and the intended beneficiaries in the village – we understand that its ill-defined structure (which allowed these different actors, functions, and practices to intersect) helped build a shared meaning of the system and the project at large. This did not mean that the structure and meaning of the IS was separated from the power-relations between the actors. All the actors did not have the similar stake in the project, and, as the originator and manager of the system, DRCSC was the primary entity determining what kind of collaboration took place around the IS. As we have illustrated, however, their freedom to do so was circumscribed by other power-structures. The perception of the value of scientific expertise shaping the design of the IS, the village governance structures shaping the recruitment of climate volunteers, and gender hierarchies shaping the reception and use of the IS within the villages, are all instances of different power relations affecting the degree and nature of collaborations in the IS. While in this section we elaborate the degree and nature of collaboration enabled by the IS, in the next section we reflect on the ways such collaborative practices may have affected resilience building around the IS.

From collaboration to resilience

Resilience in climate change adaptation can be conceptualized in three broad terms. First, the ability of the system to withstand external disturbances or shocks; second, the ability of the system to recover from such shocks; and, third, the capacity of the system to change in response to such disturbances or shocks while still maintaining its core purpose and function (Ospina and Heeks 2010). We use the terms “withstanding”, “recovering,” and “changing” to refer to these three different dimensions of resilience.

In this section, we ask what impact the collaborative practices supported by the information system had on resilience among the multiple actors, in any of the above terms. This is not intended to be a comprehensive assessment of the impact of the IS, but rather a way to understand the way in which collaboration can support resilience. It is also important to not treat these different kinds of resilience as mutually exclusive, but as interrelated outcomes of the collaborative practices enabled by the information system. While we focus on actors who are considered members of the farming communities as targeted by the project proposal, we do not restrict ourselves to those labeled by DRCSC as its “farmer beneficiaries”. Rather, we also include community mobilizers, village volunteers, and those employed in jobs such as manual labor.

Withstanding

DRCSC’s collaborative mode engendered the development of a wide variety of uses of the IS in the villages. For instance, in a dry region (such as the one where the IS was installed) the question of whether or not to irrigate is especially important as it entails substantial depletion of stored water as well as cost of electricity and pump rental. The farmers could combine the IS provided rainfall data with their own experience and knowledge to decide whether to irrigate right away or wait in case it rains. For female farmers, who are largely responsible for livestock and home kitchen gardens, temperature forecasts helped them decide whether or not to take the livestock out to the fields on a specific day.

The relative low level of control enforced by DRCSC within the system allowed the weather forecasts and the agriculture advice to be decoupled from each other and used separately. This, combined with the system’s free availability in public space, provided for an opportunity to employ the information to uses outside of the agricultural domain. For example, people working at brick kilns or as construction labor used the information provided to gauge whether or not there would be work available that day. When heavy rain was predicted, they could stay home and not incur the cost in time and money to travel to their work sites only to find that there was no work that day. For women, the system enabled them to identify when there would be extreme temperatures, information they used to decide whether or not to let their children go out of the house.

These are all examples of where the system helped farmers and others in the villages to withstand and recover from external shocks – such as increased drought spells or higher average temperatures – brought about by climate change. The farmers could act on the information provided by IS and ensure the safety of themselves, their livestock, and their children under extreme temperatures. They could better manage their water resources and plant seeds, replant saplings, and time the harvesting of their crops taking into account unseasonal and intermittent rainfall and periods of drought.

Recovering

The ability to manage resources – whether to deplete stored water and spend money on irrigation or wait in case it rains – supported the farmers’ capacity to recover from shocks. By being able to use resources efficiently they could grow additional summer crops that could serve as an economic buffer in case other crops failed. This included the kitchen gardens which ensured food security and could provide some additional income for the family when needed.

Collaborations engendered by the introduction of the IS enabled new relationships to form in the village. For instance, village volunteers would build relationships with others within the village – in particularly the elders and the majhis – which provided them greater access social resources and also enhanced their social standing. This is especially notable for those volunteers who were young and/or female – their community’s patriarchal norms would otherwise have limited their integration into the village power structure. While we could not directly observe this capacity being employed to recover from adverse climate events during the period of the project, the villagers were aware of their enhanced capacity, as the following interview transcript shows:

P: “Have you noticed any changes since becoming the village Kiosk?”

M: “Now I get to talk to people in the village that I would not before. Since the manual collection rain guage has been installed in the local village leader’s plot [majhi] I now regularly visit his house. When it has rained I go there to collect the data and talk to [the majhi].”

P: “Did he know who you were before this?”

M: “No, my father is [a relative] of his, but I do not think he knew my name before.”

P: “Do other villagers ask question about the weather information?”

M: “Yes, sometimes other villagers ask question about the weather data…”

P: “Can you remember for which data they ask for?”

M: “Mainly rainfall data.”

P: “As they are now asking you questions, do you think you are getting more importance in the village?”

M (with a big smile): “Yes….”

P: “Do you think you get better treatment from the villager after becoming the Kiosk?”

M: “Yes, I think my friends become jealous of me and aged people seek solution regarding the weather information from me.”

Changing

DRCSC’s stated objective for the system was to help farmers change their agricultural practice to ones that are more sustainable in the face of climate change. Specifically, it wanted farmers to adopt less input-driven and more agroecologically sustainable practices that increase local self-sufficiency. However, we did not see a significant change in the way individual farmers approached agriculture, e.g., move from synthetic fertilizer and pesticide use to organic farming. Rather, we saw the emergence of IS enabled “hybridised knowledge” (Puri 2007). Farmers – rather than directly adopt the given advice – would selectively combine information and knowledge from different sources. On the one hand, they could use their indigenous practices and knowledge with regard to weather, crops, local pests, and other elements of farming. On the other hand, they would also rely on the agricultural extension services and private dealers of pesticides and fertilizers for crop disease and pest diagnosis as well as recommendations on choice and application of chemical products. The fact that DRCSC neither attempted to replace any of the other sources nor make the system in any way exclusive made room for comingling of different knowledge systems. This kind of collaboration worked at a collective level where the farmer groups would discuss and evaluate how to combine the new information made available by the IS with other sources to understand how to respond to challenges or adverse conditions. We identify this as a collective shift in farming practices and correspondingly a shift in farmers’ groups as communities of practice (Lave 1991). In a way, such hybrid implementation enabled by the IS could be viewed as detrimental for the capacity of agricultural system to adapt in response to climate shocks. However, we would argue that this is only in the short-term. If DRCSC had tried to push a shift toward different practices and not accommodated these parallel knowledge systems, there might have been an initial adoption of the practices it promoted but it would not have been sustainable or resilient in the face of future adverse conditions.

Conclusion

Vulnerabilities with regards to livelihood, access to water, food security, habitat, and health are all intertwined with and exacerbated by climate change. It is now imperative that farmers be able to withstand and recover from climactic shocks as well as adapt their practices to new agroclimatic conditions.

While considerable effort has been dedicated to the way in which ICTs can support agricultural development, so far there is limited work specifically addressing how ICTs may contribute to climate change adaptation among small-holding and marginal farmers. In this article, we show how collaborative practices between different actors can improve local capacities required for such adaptation. These practices and capacities are critical elements for a resilient agricultural system.

The IS was the catalyst for the emergence of valuable alliances between different actors – DRCSC, the expert, and the villagers, who work from different knowledge systems. As Puri (2007) argues, such alliances across such diverse knowledge systems are critical “for effective IS development and implementation.” Such alliance building role of information systems has been explored by different theoretical approaches in the past. For example, the “boundary objects” framework helps us understand how coherence across different social worlds with conflicting views can be built around an information system, even when there is an absence of consensus, as was the case for the IS we studied (Callon 1986; Oswick 2005; Star 2010; Star and Griesemer 1989). Recent literature on open development also highlights the potential of ICT-mediated practices to create flexible social structures by supporting spaces for participation and collaboration (Smith 2014). However, in order to make collaboration possible between diverse communities while retaining their specific interests, the stewards of an IS have to confront power relations and manage spaces of contestations and negotiations between actors; in our case, three sets of actors spread across head office, field office, and the villages (Chaudhuri and Kendall 2019; Cornwall 2003; Escobar 2011; Ferguson 1990; Singh and Gurumurthy 2005). Accordingly, we argue, while designing and implementing an IS, careful attention needs to be paid to such spaces of contestation rather than avoiding them. In our case, we witnessed conflicts within DRCSC where its agroecological expert’s discomfort with recommendations from an external expert became an issue¹⁰. Hence, instead of treating the weather information system as a benign intervention which should to be accessible, open and flexible, we take note of the conflict of interests surrounding it. We show, amidst these diverse and conflicting interests, a shared meaning of the IS can emerge, which allows for individual autonomy and mutual collaboration at the same time.

Furthermore, the emergence of a shared meaning of the IS allowed it to become embedded in the everyday practices of the villagers – as community mobilizers, volunteers, and intended beneficiaries. Use of simple technologies such as the manual weather stations, blackboards, and printouts allowed the villagers to easily make it part of their daily lives. This resulted in a form of “artful integration” (Oreglia 2013; Orlikowski 2002), wherein local capacities emerged from within, rather than being imposed from above, to achieve desired development goals.

Bringing fundamental changes in agricultural practices, especially among small-holding and marginal farmers, is an exceedingly difficult task. Their reliance on agriculture as a primary source of livelihood makes it difficult for them to risk making big changes. Nonetheless, given their socioeconomic and climatic vulnerability, fundamental shifts are required. In these circumstances, mere access to information or expert knowledge is insufficient. Instead, we need to ask what building capacity for adaptation and resilience through an IS means. We believe it entails the following: enabling collaboration across knowledge systems (such as scientific and indigenous knowledge), embedding information practices in the everyday lives of intended and unintended beneficiaries, and encouraging actors to assign their own meanings to these practices.

Notes

Notes

1 A block is an administrative subdivision of a district that sits above a village.

2 In recent times, IMD has begun started providing forecasts on a block level in a phased manor (PTI New Delhi 2019).

3 Scheduled Tribes and Scheduled Castes are groups recognised in the Indian constitution as historically discriminated against. They are provided special protection and access to various affirmative action programmes. Similarly, Other Backward Classes are groups deemed as “socially and educationally backward classes” and afforded reservations and access to affirmative action programmes.

4 DRCSC called them “nutrition gardens”, as they provided food security and nutrients in addition to the primary crop, paddy.

5 “Chinese” is a local term used to describe low cost imported smartphones which were not specifically customised for the Indian market, e.g. lacking ability to decode Indian scripts.

6 For more on the differences in the world view of agroecology see Pretty (1995).

7 Lead farmers were often farmers with long associations with DRCSC and had adopted its agricultural practices.

8 According to Puri (2007, 358), community specific knowledge is “acquired by local communities through the accumulation of experiences, informal experiments, and intimate understanding of the environment in a given culture” and implementation specific knowledge is “built upon bureaucratic rules, guidelines, and financial norms prescribed by government and/or international donor agencies.”

9 Majhis have been traditionally village-level leaders, acting as the patriarchs of their villages. However, after the introduction of the three-tier panchayat system in 1992, there is also an elected Gram Pradhan who heads the Gram Panchayat (“village council”).

10 In another example in the case of a weather IS in Bangladesh, the local imam strongly resisted the introduction of the forecast system on basis of religion (Christensen et al. 2019).