In this interview with The Policy Edge, Dr Gokarn reflects on how state capacity is built and exercised in practice: from decision-making under uncertainty and the limits of paper systems, to designing interventions that endure beyond individual tenures. Drawing on experiences across policing and administration, he offers insights into how governance operates on the ground, and what it takes to move from individual action to system-level impact.
Your early years in the IPS exposed you to coercive state power, while the IAS offers a broader developmental mandate. In practice, how do these roles differ in shaping outcomes?
In the IPS, the interface with society is largely at the point where something has already gone wrong. You intervene when tensions escalate, a crime occurs, or order breaks down.
During my time in Agra and surrounding areas, whether it was dealing with dacoit gangs along the Chambal belt or large-scale community mobilisations, I realised that the state’s presence is often tested at the edge. The question is whether the state can assert credibility at that moment. The toolkit is largely coercive: law, force, deterrence.
But over time, I realised that responding at the point of breakdown is not sufficient; it is more important to address the underlying conditions. When I moved to the IAS, the nature of engagement changed. One begins to see how those same regions function outside moments of crisis: how livelihoods are structured, where infrastructure gaps persist, and how institutions function or fail. I was then in a position to influence those conditions.
This, however, requires remaining connected to on-ground realities. Files move, reports come in, and it is possible to assume that things are working when they are not.
So the distinction is this: policing gives you immediacy and clarity under pressure, but limited scope to change structural conditions. Administration gives you scope to shape those conditions, but only if you remain grounded in reality. If the two don’t inform each other, the state either becomes overly coercive or overly procedural.
You served in the Kumbh Mela bandobast at a time when decisions had to be taken without reliable data or technology-enabled oversight. How did this experience shape your approach to effectiveness and accountability under uncertainty?
This was a time when no CCTV, no drones, no centralised live dashboards existed. What we had was a walkie-talkie, and even that would run out of battery in a few hours. Beyond that, decisions relied entirely on on-ground judgment.
We were stationed inside the Mela area, living in tents, moving largely on horseback. On the main bathing days, we would be on horseback for continuous stretches, day and night, with only brief breaks. We were responsible for a specific zone, and whatever happened there, we had to respond to it in real time.
Decision-making, therefore, depended on understanding the system in real time. In a crowd, for example, as long as it is moving, it is manageable. The moment it stops and pressure builds at a point, that is where risk escalates.
So the decisions are not about reacting to isolated events, but about constantly managing flow: opening barricades at the right time, diverting movement, anticipating where pressure might build next. These are decisions one takes without consulting anyone, because delay itself becomes a risk.
We were accountable for action, not waiting. If something goes wrong because one hesitated waiting for instructions, that is equally a failure. This required confidence in taking decisions with incomplete information.
Errors are inevitable in such conditions; the challenge is to prevent them from becoming catastrophic.
In several early postings, you encountered administrative systems that existed on paper but lacked feedback and verification on the ground. What does this gap imply for policy design?
When I went for my district training in areas like Jalalabad, what struck me was not the absence of the state, but the nature of its presence. Large parts of the region were physically cut off for months due to flooding, lack of bridges, and poor connectivity. In many such areas, people did not know who their local officials were. The administration would appear intermittently, and then disappear again.
Take the public distribution system. Supplies would be issued from the depot, and records would show that they had been delivered. But whether they actually reached the village was unknown. There was no feedback mechanism, no verification loop.
This makes the system self-referential. For policy design, this is critical. We often assume that once a scheme is designed and funds are allocated, delivery will follow. But unless there is a mechanism to know what is happening at the last mile, the system can drift.
It also explains why systems become person-dependent. If an officer takes the effort to go out, to verify, to engage, things improve. If not, the system continues to run on paper.
Your work in Almora relied on intensive, face-to-face engagement with local institutions rather than top-down programme delivery. When does decentralised, interaction-heavy governance outperform standard bureaucratic processes?
Almora was a very different kind of experience. The terrain was hilly, populations were dispersed, and administrative reach was limited. If we tried to run everything through conventional file-based processes, very little would move.
We therefore inverted the process. Instead of expecting proposals to come up through layers, we took the administration to the people. We would camp at the block level for three days at a time, with the entire team of engineers, accounts staff, development officers.
The first day was for listening. This often meant hearing grievances about what had not been done or delivered. This mattered, because without it, the engagement would stay superficial.
The second day focused on feasibility. Out of everything that was said, what can actually be done? What are the constraints: technical, financial, logistical?
The third day was for decisions. Projects were sanctioned there itself, funds were released, and responsibilities were assigned. This removed delays between proposal and approval.
This reduced the distance between problem and solution. It also created ownership. When a Gram Pradhan knows that the project has been sanctioned in front of the community, and that progress will be reviewed in a few months, accountability improves.
This works when local institutions are functional and willing to engage. It also requires sustained administrative presence and follow-up. Where problems are local, visible, and solvable with limited resources, bringing the state closer to the people can be far more effective than standard processes.
Given limited administrative bandwidth, how should the state prioritise between incremental and high-impact, long-gestation interventions?
In Lalitpur, the Rajghat dam project had been under development for nearly 25 years. Most of the work had been done, but the last mile including clearances, land acquisition, canal energisation had remained pending. In such cases, the system often becomes accustomed to the project remaining incomplete; it fades into the background. But if you look closely, the potential impact is enormous.
Once the final clearances were obtained and the system was operationalised, the change was immediate. Areas that were dependent on a single crop could now support two. Income levels improved, and the local economy shifted.
This highlights an important point. Often, high returns come from completing projects that are already near completion but have stalled
At the same time, smaller interventions remain important. In many places, a minor irrigation channel or a small bridge can significantly improve daily life.
Prioritisation is therefore not binary. It is identifying where effort will unlock disproportionate value, whether in completing a stalled project or addressing a daily constraint through a smaller intervention.
Your later work involved temple area expansion in Varanasi, acting within a socially sensitive environment and ahead of clear political endorsement. What legitimises administrative action in such contexts?
When I first went to Varanasi, my initial impression was of chaotic streets, congestion, and hygiene issues. But as I began to understand its history and spiritual significance, I realised that any intervention here is not just administrative; it is also cultural.
When we began thinking about expanding the temple area, there was no clear mandate. There was also significant opposition: local, political, social. And the execution itself was complex. We had only a small window of a few hours at night to work. No heavy machinery could be brought in because of the narrow lanes.
In such situations, authority allows you to initiate action, but does not guarantee acceptance; that must be built.
What helped was that the intervention did not alter the fundamental character of the place. The architectural style was preserved. And existing structures were respected. As the work progressed and became visible, resistance gradually reduced.
Outcomes matter, but they are not immediate. Acceptance builds over time, through engagement and visible outcomes. Legitimacy in such contexts is layered. Authority allows you to act, but acceptance sustains the action.
In services, administrative tenures are inherently short. What distinguishes interventions that endure from those that dissipate, and how should officers design for continuity?
Early in service, it becomes clear that if an intervention depends entirely on an individual , it will not last. The question, then, is whether one is addressing a problem directly, or building a system that continues to address it.
For example, if you personally monitor something every day, it may work while you are there. But the moment you leave, it stops. On the other hand, if responsibilities are defined, feedback loops built in, and accountability established, the system has a greater chance of continuing.
This is where systems thinking becomes important. It is not about doing more, but about designing in a way that reduces dependence on individual effort.