By Venky Ramachandran
India is preparing to change the way its villages decide what to build with rural jobs funds – under what was MNREGA and is now the VB-G RAM G Act – including water works and disaster planning, both of which are assuming critical importance due to climate change. The biggest change lies in who decides what a village needs.
The Ministry of Rural Development, working with German development agency GIZ, published the Draft Framework for the Viksit Gram Panchayat Plan in June 2026 under the “Support to India’s Water Vision 2047″ project. (Viksit Bharat is the government’s vision to make India a fully developed nation by the year 2047, which marks 100 years of Independence.) The framework proposes a more scientific, geospatially enabled approach to planning works across India’s roughly 2.5 lakh Gram Panchayats.
At the heart of the proposal is Yuktdhara, a GIS-based geospatial planning portal launched in 2021 to facilitate village-level planning and asset creation under the Mahatma Gandhi National Rural Employment Guarantee Act (MGNREGA). It draws on datasets from platforms including PM Gati Shakti (for transport and logistics), India-WRIS (for water) and Bhuvan (satellite research).
The system will conduct a gap assessment and generate a shortlist of possible works, including proposed coordinates, site boundaries and dimensions. Planning teams will then take these recommendations to hamlets, where Gram Sabhas are expected to ‘ground truth’ them and validate, refine and prioritise interventions based on local needs and conditions. This sequence could fundamentally change the relationship between technology and local democracy. The algorithm identifies what could be built. The Gram Sabha reviews it.
A more scientific way to plan?
The case for better planning is difficult to dispute. The draft framework identifies 318 permissible works across 72 categories. Water security alone accounts for 107 options, while core rural infrastructure has 88, rural livelihoods 86 and disaster mitigation 37. The scale and specificity of the catalogue are part of the framework’s attempt to move away from what it describes as traditional demand-based or “wish-list” planning towards a system based on data and resource optimisation.
There is good reason to think the old system needed fixing. The last all-India CAG performance audit of MGNREGA, covering April 2007 to April 2013 – over a decade ago – examined 3,848 Gram Panchayats across 182 districts, 458 blocks, 28 states and four Union Territories. It found that in 1,201 sampled Gram Panchayats across 11 states, about 31% had either not prepared an annual plan or had prepared one incompletely. In three states, including Andhra Pradesh, none of the sampled Gram Panchayats had prepared a plan.
District labour budgets were missing in 49 districts. Where plans existed, projected employment generation was sometimes absent, while many plans contained neither a shelf of works nor a list of assets to be created. The time taken for a Gram Panchayat plan to reach the block also varied dramatically, from one to 21 months. And plans often bore little relationship to what eventually happened on the ground, the audit found. Across 14 states and one Union Territory, 129.22 lakh works worth Rs 1,26,961 crore were approved in annual plans during the period of the audit. Only 38.65 lakh, about 30%, were completed. Meanwhile, nearly 5,000 works were executed outside the annual plans in 25 districts across nine states.
There was another problem: the people and institutions responsible for implementing MGNREGA did not always have the capacity to do so. The Gram Rozgar Sevak, or GRS, is the village-level functionary responsible for functions including registration, job cards, work allocation, wage payments and social audits. The CAG found vacancy rates for GRS posts ranging from 20% in Uttar Pradesh to 93% in Punjab.
Four states, including Tamil Nadu and Kerala, had no dedicated GRS at all. In Uttar Pradesh, 74% of the money allocated for training these staff remained unspent. The democratic mechanism at the centre of the planning process has its own weaknesses. Studies of Gram Sabha participation have routinely found attendance in the low double digits. This creates a powerful argument for bringing better information into village planning. But it also creates a danger. If Gram Sabhas are already functioning imperfectly, will an algorithm strengthen their ability to make decisions, or simply make their role more ceremonial?
Where data could help
There are areas where this could be a significant improvement. Groundwater is an example. The framework proposes minimum expenditure on water-related works based on groundwater status: 65% in over-exploited and critical blocks, 40% in semi-critical blocks and 30% in safe blocks. A farmer may know that a borewell is failing. But groundwater systems do not follow Gram Panchayat boundaries. An aquifer can extend across several villages, making a wider spatial picture essential to planning. GIS could also help address duplication. GeoMGNREGA has geotagged more than three crore assets since the exercise began in 2016. If Yuktidhara can identify an existing percolation tank a few hundred metres from a proposed new one, that is useful information that a local planning meeting may not have.
The framework’s treatment of maintenance is another potentially important correction. Of the 318 permissible works, 97 are repair and maintenance works. This could help address a longstanding incentive problem in rural public works: building something new creates employment, photographs and political visibility. Repairing an existing check dam or restoring a neglected water structure produces less visible credit. Giving maintenance its own place in the planning framework could change that incentive.
The technology, therefore, is not the problem. In some cases, it may be precisely what the planning process needs.
But can a map see the village?
The problem begins with the data. The draft framework’s list of data layers reveals a striking mismatch between the age of some of the underlying information and the ambition of the planning system. The framework cites relatively recent satellite imagery, including high-resolution one-metre data from 2022 to 2024 and LISS-IV data from 2025 (LISS-IV refers to ISRO’s Linear Imaging Self-Scanning Sensor-4), a high-resolution multispectral optical camera system for detailed Earth observation and land mapping). But several interpretation layers are much older. The Land Use Land Cover layer is from 2015-16. Land Degradation data is also from 2015-16. Geomorphology data dates to 2005-06. Wetland data is from 2005. The Waste Land layer dates to 2008-09.
What happens when Yuktdhara tells a Gram Panchayat in 2026 that degraded wasteland is available for plantation, based on a land classification generated in 2008-09? The problem is not merely that the data may be old. It is that some categories are themselves difficult to interpret from a satellite. “Wasteland” is a particularly loaded example.
Land that appears unused on a map may be grazing land, fallow land or a community resource that supports livelihoods. A satellite can see the land, but not necessarily who uses it. A 2024 field study of digital planning tools for rural water security found that Bhuvan’s thematic layers could be spatially coarse and insufficiently updated. It also found that government staff were not always trained to use the available tools, while civil society field staff sometimes had to source higher-resolution data elsewhere.
The Gram Sabha must not become a rubber stamp
The new framework is trying to solve a real problem: rural planning has often been fragmented, poorly documented and vulnerable to local capacity constraints. But there is a difference between using technology to improve democratic decision-making and using technology to make the decision before democracy has a say. The new system gives the algorithm a privileged position. Its recommendations become the starting point for discussion, while the Gram Sabha is asked to validate, refine and prioritise them. This can work, but only if “ground truthing” means real scrutiny. A Gram Sabha must be able to reject a recommendation because the data is wrong, incomplete or blind to a local reality. It must be able to add something that does not exist in the database. And it must be able to propose something that the algorithm did not identify. Otherwise, the process risks producing a peculiar inversion of decentralised planning: a system in which the State’s datasets decide what a village needs and villagers are invited to correct the dataset.
India does not need to choose between algorithms and Gram Sabhas. It needs to decide what each is good at. An algorithm can identify patterns that are difficult for people to see. It can map groundwater systems, identify duplication, analyse spatial gaps and bring consistency to planning across thousands of villages. A Gram Sabha can know that the “wasteland” on a map is where cattle graze. It can know that a proposed water structure will fail because villagers tried it before. It can know which hamlet has been repeatedly overlooked or which asset matters more than the model suggests. The success of India’s new rural planning system should, therefore, not be measured by how much decision-making it takes away from the village. It should be measured by whether technology gives villagers better information with which to make decisions.
The piece is published in arrangement with Climate Action Live https://www.climateaction.live/