On March 31, 2026, nine people died at the Sheetla Mata temple in Bihar’s Nalanda district. Eight of them were women. A large crowd had gathered, as they do every week on Tuesdays in the month of Chaitra. Severe overcrowding triggered a crush. Police and local villagers rushed in to begin rescue operations.Within hours, the familiar sequence followed: expressions of condolence, an announcement of inquiry, a promise of action.India has been through this sequence many times. Between 2001 and 2022, 3,074 people died in stampedes across the country. Researchers caution that stampede deaths are frequently underreported, so the real number is likely higher. Per one estimate, at least 114 people died in stampede incidents in 2025 alone, the second-highest annual toll in recent years. Per one estimate, of the nine incidents that took place between July 2024 and July 2025, six of them were at religious gatherings.The Nalanda temple incident was preceded by a string of prominent cases: The Karur TVK rally in September 2025. The Chinnaswamy Stadium in Bengaluru in June 2025. The New Delhi railway station in February 2025, where devotees scrambling to board trains to Prayagraj for the Maha Kumbh left 18 dead. The Maha Kumbh itself, where the BBC has estimated that at least 82 people were killed in four separate incidents on January 29, 2025. The Hathras Satsang in July 2024, where 121 people, mostly women, were crushed to death at an event that had permission for 80,000 people and drew 250,000. The list goes on.Contrary to popular belief, the causes are known and so are the solutions. However, there remains a large chasm between the two and that is exactly where the governance problem lies: an almost entirely self-inflicted one.What the science actually saysCrowd science is a well-developed field. Its core findings have been consistent for years and are accessible to any administrator who chooses to look.Crowd density above five persons per square metre makes free movement virtually impossible and creates the conditions for compressive asphyxia, which is the primary cause of death in crowd crushes. Compressive asphyxia kills. Trampling is the popular image; asphyxia is the medical reality. When people are compressed tightly enough that they cannot breathe, they die. The mechanism is simple. The prevention is equally simple: keep density below the threshold.Drones linked to ground-based monitoring systems can continuously track crowd density in real time. This technology exists, is commercially available and has been deployed at large Indian events before.Bottlenecks, counter-flows and slopes multiply pressure and destabilise movement. Bidirectional flows through shared gates are a known killer. A single combined entry and exit point at a crowded religious site reliably produces casualties when crowd numbers exceed a certain threshold. Calling it an oversight flatters the people responsible for it.Event management science also addresses the psychology of crowds. Trained stewards positioned at regular intervals can redirect movement before surges develop. Splitting crowds into manageable segments prevents the kind of massing that creates dangerous density. These are inexpensive measures and they work.The Hathras Satsang in 2024 had one makeshift tent, insufficient exits, no real-time monitoring and a crowd three times the permitted size. None of this was unknowable in advance. Disaster management expert Sanjay Srivastava noted after the incident that typically there should be eight to ten well-marked exits opening into open areas. There were not.India has the framework. What it severely lacks is enforcementThis is the critical point that each post-stampede inquiry tends to sidestep. India already has the guidance infrastructure to prevent most of what keeps happening.The National Disaster Management Authority has maintained its crowd management guide since 2020. The National Building Code prescribes standards for venue design. The National Institute of Disaster Management runs training programmes.The NDMA’s guide in particular recommends advance risk assessment, detailed site layout plans, calculated occupancy limits, predetermined ingress and egress routes, real-time monitoring, and communication protocols. The framework, in other words, is comprehensive.The problem is that almost none of it is statutory. The guidelines are advisory. Compliance is not a condition of permission for most events. Organisers face no credible legal consequences for ignoring them until people are already dead. And by that point, the inquiry that follows typically produces another set of recommendations that will also be advisory.The Srikakulam temple incident in November 2025, where nine people died in an unregistered private temple in Andhra Pradesh, illustrated the enforcement gap precisely. One combined entry and exit, use of an under-construction area for crowd movement, severe capacity overshoot, no qualified stewarding.Every failure had been documented in previous incidents and flagged in existing guidelines. The guidance existed. The enforcement was missing.The religious gathering problem specificallyPer some estimates, roughly 80% of stampedes in India occur at religious gatherings or pilgrimages. An IIT-Delhi study analysing 137 stampedes worldwide between 1883 and 2017 found that religious events account for 64% of all stampede fatalities globally, more than sports, entertainment, or political events combined.In India, the proportion is even higher because of a structural governance failure that applies specifically to religious events: they frequently proceed without any licence that ties permission to a verifiable crowd safety plan.Secular events, stadiums, concerts and political rallies operate under at least nominal regulatory oversight. Religious gatherings, particularly at smaller temples and pilgrimage sites, often operate outside any adequate licensing framework.Authorities may infer venue capacity from the physical space available rather than from calculated egress options and evacuation times. Temporary barricades go up without certified load ratings. Under-construction areas get absorbed into crowd circulation routes because there is no one formally responsible for preventing it.When an event lacks a licensed safety plan, there is no document to test for compliance, no official to hold accountable for deviations and no mechanism for pre-event intervention. The result is a governance vacuum that gets filled by organisers whose incentive is to maximise attendance, sometimes for genuine devotional reasons, sometimes for prestige and sometimes for the revenue that large gatherings generate.The counterexample that proves the pointThe strongest argument that India’s stampede problem is a governance failure rather than an inevitability is Telangana’s Sammakka Saralamma Jathara at Medaram, Asia’s largest tribal religious congregation.The Jathara reportedly draws approximately one crore devotees over four days, held in a remote forest area in Mulugu district. It is a biennial event, deeply emotionally charged, drawing pilgrims from multiple states. Every variable that typically correlates with stampede risk in India is present at maximum intensity.Telangana seems to have recorded zero fatalities from stampedes at the Jathara since 2014. The state spends significant resources on the event, deploying over 42,000 personnel from 21 departments, operating a dedicated emergency call centre, constructing permanent crowd management infrastructure and coordinating real-time surveillance.The festival is treated as an engineered system requiring advance planning, licensed capacity and active management rather than as a spontaneous religious gathering that the administration manages reactively.The contrast with Nalanda, with Hathras, with Srikakulam, is produced entirely by administrative choices. The crowds at Medaram are larger. The fatalities are zero.The accountability gap that sustains the cycleThe persistence of India’s stampede record despite available solutions points to a specific governance failure: the absence of meaningful accountability for preventable deaths at mass gatherings.When 121 people die at Hathras, the organiser disappears. A few officials are suspended pending inquiry. An SIT is constituted. Reports are filed. And the next large gathering is likely left free to proceed without the licensing requirements, verified safety plans or certified infrastructure that would prevent a recurrence.Event organisers have a structural incentive to maximise crowd size. At religious events, larger attendance signals divine blessing and organisational importance. At celebrity and political events, crowd size is a measure of status. The economic incentive runs in the same direction: more people mean more revenue from ancillary activities.Without statutory obligations that create a countervailing legal risk, the incentive to manage crowds safely competes with the incentive to fill them.India already has a model for what statutory crowd management looks like in practice. The Integrated Command and Control Centre at Tirumala established months after a stampede at that temple, and Medaram’s proven track record all demonstrate that the operational ingredients are available and have been deployed at large scale.The remaining task is legislative and institutional. Mandatory licensing for all large gatherings, compliance with NDMA or National Building Code prescriptions as a condition of permission, calculated occupancy limits that are verified rather than assumed, certified structures that prevent bidirectional flows, and post-event reviews with genuine accountability for non-compliance: these measures are known, tested and feasible.What they require is a policy culture that treats religious events as engineered systems subject to safety law rather than as expressions of faith that exist outside the ordinary governance of public space. That cultural shift has not happened. Nine deaths in Nalanda on March 31, 2026, are the most recent data point confirming it.Anas Ahmad Tak is a policy researcher, writer and political analyst. He runs Politics to Policy, a weekly newsletter on Indian governance and electoral politics.