Why India’s emergencies end the way they do.
Working notes on crowd control, high-rise fires, disaster rescue, perimeter security and heavy carriage — the figures, the mechanics behind them, and what they mean for the machines India needs next.
Each note takes one domain and asks the same question: when this goes wrong, what exactly is the thing that could not be done in time? The answers are more specific than “not enough resources”, and the specifics are what a machine has to be designed against.
Crowd control in India is a distance problem wearing a staffing costume
The standard reading of a crowd disaster is that there were not enough police. The staffing numbers support it: India polices at 152.8 officers per 100,000 people against a UN benchmark of 222, roughly a fifth of sanctioned posts sit vacant, and 86% of the strength is constabulary — the rank that actually stands in front of a gathering. Recruitment has run for a decade and the absolute gap has grown.
But look at what the officers who are present are physically doing, and staffing stops being the whole answer. To see a crowd at night, somebody drives a light tower to it. To be heard over it, somebody walks in until shouting works. To pull an injured person out of the middle, a line of officers goes in and makes a corridor. Every one of those actions is a person closing distance, and the cost lands on both sides of the line: 11,566 security personnel injured in 4,736 stone-pelting incidents, 2,744 central police personnel hurt in a single year, and more than 120 lives lost to stampedes in one year of national reporting.
Doubling the number of officers would put more people inside the same contact range. It would change the arithmetic of the line and not the mechanic of the injury.
152.8
police per 100,000 people, against a UN benchmark of 222
BPR&D, Data on Police Organisations
86%
of state police strength is constabulary — the rank that stands in front of a crowd
PRS Legislative Research, on BPR&D data
120+
lives lost in Indian stampedes during 2025
Compiled national reporting
The thing to design against is not the shortage. It is the fact that seeing, being heard, and reaching a person have all historically required a body at contact range.
High-rise fires outrun the equipment that answers them
India runs a 96.3% shortfall in fire personnel and a 97.5% shortfall in fire stations against its own assessed requirement — 3,377 stations against 8,559 needed — with an 80% shortfall in firefighting and rescue vehicles behind it. Those are the numbers usually quoted, and they describe a capacity problem.
The geometry problem underneath them is less discussed. A hydraulic ladder has a maximum height and needs a road wide enough to park and outrigger on. Cities have added floors above that height and narrowed the lanes below it at the same time. When a fire starts above the ladder line in a lane a tender cannot enter, the shortfall in vehicles stops being the binding constraint, because another vehicle of the same kind would not reach either.
The consequences show up in the incident record rather than the capacity tables: 5,888 fire deaths in a single year, and repeated events — 27 dead in a game-zone fire at a venue with no valid fire clearance, 25 in a single nightclub fire — where the people who died were on floors that help could see.
96.3%
national shortfall in fire personnel, and 97.5% in fire stations
NDMA / Standing Fire Advisory Council
3,377
fire stations in the country, against 8,559 assessed as required
Ministry of Home Affairs, Rajya Sabha reply
5,888
fire-related deaths in 2024
NCRB
What is missing is not another ladder. It is any way at all to work the face of a building from outside it, at the floor that matters, without a road underneath.
Most of the rescue window is spent travelling, not rescuing
Survival after a structural collapse is a function of time, and the curve is steep: international time-to-rescue analysis puts survival at roughly 74% for people reached within 24 hours and around 22% after 72. India records 1,604 deaths from the collapse of structures in a single year, and 2,803 lives lost to floods, landslides and cyclonic storms in another.
The part of that window a response actually controls is smaller than it looks. Rubble, standing water and unstable structure stop vehicles outright and slow people to a walk. Teams are well trained and correctly equipped; what they cannot do is arrive faster than the ground allows. So the hours that decide the outcome are consumed getting to the person rather than working on them.
Drones already fly at these incidents. They map damage and they find people, which is the sensing half of the job, and it is genuinely useful. What no standard machine does is carry weight the other way — water, a radio, a splint, an oxygen mask — or bring a person back out.
74% → 22%
survival for people trapped in a collapsed structure, reached inside 24 hours against after 72
Macintyre et al., Prehospital and Disaster Medicine — international study, not India-specific
1,604
deaths from the collapse of structures in a single year
NCRB, Accidental Deaths and Suicides in India
2,803
lives lost to floods, landslides and cyclonic storms in a single year
Ministry of Home Affairs, Lok Sabha reply, 2024
The design target is the ground between the team and the person, and the question of what can cross it carrying something useful.
A perimeter is continuous. The people guarding it are not
India guards a 15,106 km land border across 92 districts, and 72 airports under central industrial security cover, alongside refineries, plants, dams and rail. Every one of those assets has a perimeter that exists at every point simultaneously, watched by guards who exist at one point at a time.
The consequence is structural rather than a failure of diligence. A guard walking a stretch of fence leaves the rest of the line unobserved, and the gap travels with him on a schedule anybody watching can learn. Fixed cameras narrow the gap without closing it, because they see where they are pointed and a perimeter is mostly where they are not.
The threat profile has also moved upward. Roughly 294 drones were seized crossing one stretch of the western land border in a year, and anti-drone systems at civil airports have moved to approved phased rollout. Perimeter security is now partly an airspace problem, and airspace is the one dimension a foot patrol has never covered.
15,106 km
of land border, running through 92 districts
Ministry of Home Affairs, border management records
72
airports under central industrial security protection
CISF operational review, 2026
~294
drones seized crossing one stretch of the western land border in a year
MHA, Lok Sabha reply
The requirement is coverage that is continuous in the same way the perimeter is, and that can be over an intrusion before anyone reaches a gate.
Supplying a roadless outpost costs lives before anything else does
The hardest logistics in India are vertical. Posts sit between 3,000 and 5,500 metres, supplied by mules and porters that take six to ten hours to cover a short distance above 15,000 feet, on routes that a single pass closure can shut for nearly six months of the year. Where vehicles do run, the roads themselves are the hazard: sixteen personnel died when one truck of a three-vehicle convoy went off a high-altitude road.
The starkest figure is older and larger: 869 personnel lost on one high-altitude glacier over three decades — the majority to weather and terrain, not to any hostile act. The environment, not the adversary, is the thing doing most of the killing, and every journey made to carry supplies through it is exposure accepted because there was no alternative.
This is a carriage problem before it is anything else. The load is known, the route is known, and the cost of moving it is currently paid in people.
869
personnel lost on one high-altitude glacier over three decades — the majority to weather and terrain, not to any hostile act
Ministry of Defence, data placed before the Lok Sabha
16
killed when one vehicle of a three-vehicle convoy went off a high-altitude road
PTI and national reporting, 2022
The question is what can take a known load over known terrain, on schedule, without a person on the route.
The last kilometre of heavy lifting is the one that injures people
India records 26,437 deaths from falls in a single year, most of them from height, and construction-sector studies attribute a large share of site deaths to exactly that. Behind the number is a routine act: somebody carrying weight up a structure by hand, because nothing else on site could put it there.
The economics repeat across sectors. Over 5 lakh circuit kilometres of transmission network run largely through terrain without roads; 25,000 rural habitations are still waiting for all-weather road connectivity; mines, offshore platforms and high-rise sites all share the same last stretch — no road, no crane that reaches, no safe way to carry it up. Moving a hundred kilograms the final kilometre routinely costs more than moving it the first thousand, and the cost is not only money.
A crane needs ground, reach and time to erect. A road needs building. A person carrying a load at height is the cheapest option available, and the most dangerous.
26,437
deaths from falls in a single year, most of them from height
NCRB, Accidental Deaths and Suicides in India
5 lakh+
circuit kilometres of transmission network, much of it where no road runs
Ministry of Power
25,000
rural habitations still awaiting an all-weather road
Ministry of Rural Development
The design target is the point of work itself — the tower top, the platform, the unfinished floor — and what can put weight there with nobody climbing under it.
More notes as the machine meets the field.
These notes are revised as we learn from the people who work these incidents. Where something we assumed turns out to be wrong, the note changes and says so.