In India, stampedes have become a routine tragedy, taking a heavy toll on lives and often being man-made. At one time, they were largely confined to religious events, triggered by the fervour of devotees. In recent times, they have also occurred at railway stations and celebrity events.
The Bangalore tragedy is a case in point. Meant to celebrate RCB's IPL T20 victory, the event at Chinnaswamy Stadium-which has a capacity of about 35,000-drew an estimated 200,000 to 300,000 people. The inevitable rush, jostling to get in and lack of discipline led to chaos. The resulting stampede left 11 dead and dozens injured. A senior police officer had reportedly flagged the risks before the event, but no action was taken.
Every such tragedy is unique. Yet, the reasons cited are almost always the same. Crowds are always 'unexpectedly large', sudden surges are caused by an 'unplanned event' (VIP or celebrity presence, late opening of a gate) and 'insufficient police presence', which result in poor crowd control.
Simple system improvements could prevent-or at least mitigate-such tragedies. Now, there is the additional tool of new technology. Tapping its potential, though, requires something uncommon in India: multidisciplinary teams. Such teams must include technologists, from computer programmers to data scientists to sensor and drone experts, and mathematicians. Crucially, they must also have social scientists-particularly sociologists and psychologists-a breed rarely seen on such teams.
To start with, such a team would collect as much data as possible from past stampedes and large gatherings to build a basic model of crowd dynamics. It would seek to model how and when crowds suddenly surge and events that lead to this getting out of control.
Analytics based on historical data would help create a model to answer these and other questions. Updates from new events, including large crowds and stampedes, would be used to continuously hone and refine the model.
Collecting old data may be difficult-but it is possible. It would have to depend on anecdotal data, social media posts, CCTV coverage and mobile video recordings from the scene.
Some events-from the past and in future-may have drone coverage, which would provide additional data. Collating and synthesising from all this, and integrating it with terrain data, local maps (where are the gates?) and weather information (sudden showers or high temperatures can cause stampedes) can best be done by using AI.
All that is mentioned above about crowds also applies to mobs-a more focused, often more volatile, form of crowd. Handling them poses similar challenges.
Augmenting data-based models with insights into human behaviour can make them significantly more accurate and useful. This is where sociologists-who understand group and crowd behaviour-and psychologists (who have insights on individual behaviour) play a vital role.
In addition to statistical correlation provided by data analytics and AI, social scientists would provide the cause-and-effect dynamics that are key to preventing such tragedies. In scientific terms, this is adding physics to mathematics. Also, social scientists can suggest what kind of social media messages, or 'nudges', might prevent crowd accumulation or ensure safe behaviour of crowds. With more data, social science, like data analytics, will be able to improve the model, do predictive modelling, even preventive modelling.
Four specific things follow from the above:
Data is key: Collecting and collating data from all sources is critical. Going forward, new technologies-drones, sensors, big data analytics, AI-must be used. While models based only on statistical data analytics may be fungible, these will not work well here since crowd behaviour is culture-specific.
Study crowd behaviour: There's a growing market for large, live events-sports, concerts, political rallies-involving tens of thousands of people. This makes such models vital. Their culture-specificity highlights the need to develop indigenous models, necessitating the involvement of social scientists.
Computational sociology: It requires sophisticated and multi-disciplinary research with a degree of mutual understanding between technologists and social scientists. This needs the development of a new discipline: computational sociology. GoI should support the roll out of dedicated courses and fund them. It is worth noting that any successful use of tech for people's benefit requires the involvement of social scientists. Isro demonstrated the advantages through its pioneering multi- and trans-disciplinary teams for the Satellite TV Experiment (SITE) in 1975.
Social media key: Given the role of social media in crowd collection and subsequent behaviour (often triggered by rumours or fake news), prompt corrective and pre-emptive posts would help. These can be created and sent out instantaneously by GenAI, which can be integrated into the model.
With our population size and culture, large crowds may be unavoidable. However, what the tech-based models can ensure is that crowds do not equal stampedes. That would be one tragedy we could-and should-finally prevent.
The writer is chairperson, Indraprastha Institute of Information Technology Delhi
The Bangalore tragedy is a case in point. Meant to celebrate RCB's IPL T20 victory, the event at Chinnaswamy Stadium-which has a capacity of about 35,000-drew an estimated 200,000 to 300,000 people. The inevitable rush, jostling to get in and lack of discipline led to chaos. The resulting stampede left 11 dead and dozens injured. A senior police officer had reportedly flagged the risks before the event, but no action was taken.
Every such tragedy is unique. Yet, the reasons cited are almost always the same. Crowds are always 'unexpectedly large', sudden surges are caused by an 'unplanned event' (VIP or celebrity presence, late opening of a gate) and 'insufficient police presence', which result in poor crowd control.
Simple system improvements could prevent-or at least mitigate-such tragedies. Now, there is the additional tool of new technology. Tapping its potential, though, requires something uncommon in India: multidisciplinary teams. Such teams must include technologists, from computer programmers to data scientists to sensor and drone experts, and mathematicians. Crucially, they must also have social scientists-particularly sociologists and psychologists-a breed rarely seen on such teams.
To start with, such a team would collect as much data as possible from past stampedes and large gatherings to build a basic model of crowd dynamics. It would seek to model how and when crowds suddenly surge and events that lead to this getting out of control.
- Is there a critical mass or crowd density that sparks uncontrolled pushing and panic?
- Are problems due to choke points or sudden pressure releases (as when a gate is opened, or a celebrity arrives)?
- Are there patterns of crowd agglomeration and movement?
Analytics based on historical data would help create a model to answer these and other questions. Updates from new events, including large crowds and stampedes, would be used to continuously hone and refine the model.
Collecting old data may be difficult-but it is possible. It would have to depend on anecdotal data, social media posts, CCTV coverage and mobile video recordings from the scene.
Some events-from the past and in future-may have drone coverage, which would provide additional data. Collating and synthesising from all this, and integrating it with terrain data, local maps (where are the gates?) and weather information (sudden showers or high temperatures can cause stampedes) can best be done by using AI.
All that is mentioned above about crowds also applies to mobs-a more focused, often more volatile, form of crowd. Handling them poses similar challenges.
Augmenting data-based models with insights into human behaviour can make them significantly more accurate and useful. This is where sociologists-who understand group and crowd behaviour-and psychologists (who have insights on individual behaviour) play a vital role.
In addition to statistical correlation provided by data analytics and AI, social scientists would provide the cause-and-effect dynamics that are key to preventing such tragedies. In scientific terms, this is adding physics to mathematics. Also, social scientists can suggest what kind of social media messages, or 'nudges', might prevent crowd accumulation or ensure safe behaviour of crowds. With more data, social science, like data analytics, will be able to improve the model, do predictive modelling, even preventive modelling.
Four specific things follow from the above:
Data is key: Collecting and collating data from all sources is critical. Going forward, new technologies-drones, sensors, big data analytics, AI-must be used. While models based only on statistical data analytics may be fungible, these will not work well here since crowd behaviour is culture-specific.
Study crowd behaviour: There's a growing market for large, live events-sports, concerts, political rallies-involving tens of thousands of people. This makes such models vital. Their culture-specificity highlights the need to develop indigenous models, necessitating the involvement of social scientists.
Computational sociology: It requires sophisticated and multi-disciplinary research with a degree of mutual understanding between technologists and social scientists. This needs the development of a new discipline: computational sociology. GoI should support the roll out of dedicated courses and fund them. It is worth noting that any successful use of tech for people's benefit requires the involvement of social scientists. Isro demonstrated the advantages through its pioneering multi- and trans-disciplinary teams for the Satellite TV Experiment (SITE) in 1975.
Social media key: Given the role of social media in crowd collection and subsequent behaviour (often triggered by rumours or fake news), prompt corrective and pre-emptive posts would help. These can be created and sent out instantaneously by GenAI, which can be integrated into the model.
With our population size and culture, large crowds may be unavoidable. However, what the tech-based models can ensure is that crowds do not equal stampedes. That would be one tragedy we could-and should-finally prevent.
The writer is chairperson, Indraprastha Institute of Information Technology Delhi
(Disclaimer: The opinions expressed in this column are that of the writer. The facts and opinions expressed here do not reflect the views of www.economictimes.com.)
Kiran Karnik
The writer is former director, Development and Educational Communicational Unit, Isro, and former member, Scientific Advisory Council to the Prime Minister