🔬 physics

Crowd Dynamics in Historical Perspective: Reframing the Amritsar Massacre through Agent-Based Modelling and Social Psychology

This study employs agent-based modelling and social psychology to demonstrate that the 1919 Jallianwala Bagh massacre resulted in significantly higher fatalities than officially recorded due to physical constraints, while also revealing how early 20th-century crowd psychology theories rationalized excessive state violence, offering critical insights for historical accountability and modern crowd safety.

Original authors: Mohcine Chraibi, Krisztina Konya, Ezel Üsten

Published 2026-05-01

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Mohcine Chraibi, Krisztina Konya, Ezel Üsten

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine a crowded room where people are trying to escape a sudden, terrifying threat. Now, imagine that the people in the room are not just individuals, but a single, shifting organism where the behavior of one person changes the fate of everyone else. This is the core idea behind a new study that uses computer science and history to re-examine one of the darkest moments of the British Empire: the 1919 Amritsar Massacre.

Here is a simple breakdown of what the researchers did and what they found, using everyday analogies.

The Historical Mystery: "How Many?"

In 1919, British troops fired on a peaceful crowd of unarmed people in a walled garden in India. The British government officially said 379 people died. However, for over a century, historians and witnesses have argued this number is far too low, suggesting the real toll was in the thousands.

The paper asks: If we look at the physics of the situation, does the official number make sense?

The Computer Experiment: A Digital Time Machine

The researchers built a "digital twin" of the garden (Jallianwala Bagh) and filled it with thousands of virtual people (called "agents"). They didn't just guess; they programmed these people to run, bump into each other, and react to danger based on real-world physics.

Think of it like a very serious video game simulation where the goal isn't to win, but to see how many people would realistically get hurt given the layout of the room and the speed of the bullets.

The Setup:

The Room: A walled garden with only a few narrow exits (like a funnel).
The Threat: 50 soldiers firing continuously for 10 minutes.
The Crowd: Between 5,000 and 15,000 people, packed tightly together.

The Big Discovery: The "Shielding" Paradox

The most interesting part of the study is how the crowd itself acted. The researchers found that being in a crowd has a double-edged sword effect, which they call the "shielding" paradox.

The Shield Effect: If you are surrounded by people, your neighbors might physically block a bullet from hitting you. It's like standing behind a thick wall of people; you are safer because others are in the way.
The Trap Effect: However, if the soldiers are aiming at the densest parts of the crowd (which they likely were), being in a tight group makes you a bigger target. It's like a target on a dartboard; if the board is crowded with darts, you are more likely to get hit.

The computer simulation showed that when the crowd is very dense, the "Trap Effect" often wins. People get crushed, blocked from running, and hit by bullets meant for others. The crowd stops being a group of individuals and becomes a chaotic, moving obstacle course.

The Results: The Numbers Don't Add Up

When the researchers ran the simulation with "conservative" assumptions (meaning they didn't try to make the worst-case scenario; they used moderate numbers for how fast the soldiers shot and how fast people could run), the results were shocking.

Official Count: 379 deaths.
Simulation Count: Even with the most cautious settings, the computer predicted hundreds to thousands more deaths than the official report.
- With 5,000 people, the model predicted roughly 670 to 1,800 deaths.
- With 15,000 people, the model predicted up to 7,900 deaths.

The Takeaway: The official British number of 379 is physically impossible if you account for the size of the crowd, the narrow exits, and the duration of the shooting. The simulation suggests the tragedy was much larger than history books have admitted.

The "Mindset" Factor: Seeing a "Mob"

The paper also looks at why the soldiers acted the way they did. It explains that in the early 1900s, there was a popular theory (by a man named Gustave Le Bon) that crowds were like wild animals—irrational, dangerous, and not made of real human beings.

The British commander, General Dyer, didn't see a crowd of parents, children, and farmers. He saw a "mob." The paper argues that this dehumanizing mindset was like wearing blinders. Because he viewed the crowd as a single, threatening monster rather than a collection of individuals, he felt justified in using extreme force. The computer model helps prove that this mindset led to a disaster that was far worse than necessary.

Why This Matters Today

The authors say this isn't just about history. They argue that we still make the same mistake today: treating large groups of protesters as "mobs" rather than people. When leaders view crowds as irrational threats, they are more likely to use violence that causes unnecessary chaos and death.

Summary

This paper uses a computer simulation to show that:

Physics matters: In a crowded, walled space, a shooting lasts longer and kills more people than simple math suggests because people get stuck and block each other.
History was likely wrong: The official death count of 379 is too low; the simulation suggests the real number was much higher.
Perception is dangerous: Viewing a crowd as a "monster" rather than people can lead to decisions that turn a tragedy into a massacre.

The study acts like a forensic tool, using math to check the "story" of history and finding that the official story doesn't hold up under the weight of reality.

1. Problem Statement

The paper addresses two interconnected problems:

Historical Discrepancy: The official British colonial record of the 1919 Jallianwala Bagh (Amritsar) massacre cites 379 deaths. However, Indian witnesses and historians have long estimated the toll to be significantly higher (ranging from 1,000 to 1,500+). The authors argue that existing historical narratives often lack a quantitative, physics-based verification of these claims.
Methodological Gap: Contemporary crowd science primarily focuses on optimizing orderly movement (evacuation, ingress/egress). There is a lack of frameworks to model crowds as targets of intentional violence (e.g., active shooters, mass shootings), where dynamics differ fundamentally due to panic, obstruction by casualties, and targeted fire.
Socio-Psychological Context: The paper seeks to understand how early 20th-century crowd psychology (specifically Gustave Le Bon's theories) dehumanized crowds as "irrational mobs," providing a rhetorical justification for excessive state force, a framing that persists in modern political discourse.

2. Methodology

The study employs an interdisciplinary framework combining computational physics (Agent-Based Modelling) with critical social psychology.

A. Agent-Based Simulation (Physical Model)

The authors developed a physically grounded simulation using the JuPedSim open-source framework.

Agent Dynamics: Pedestrians are modeled using a first-order collision-free speed model. Agents have an initial preferred escape speed ( $v_0 = 3$ m/s).
Geometry: The simulation reconstructs the Jallianwala Bagh layout (approx. 182m x 137m), including five narrow exits, internal obstacles (trees, a well, a stage), and a firing line of 50 soldiers along one boundary.
Shooting Mechanics:
- Fire Rate: Based on historical records, 1,650 rounds were fired over 10 minutes (approx. 3 rounds/soldier/min), modeled as a continuous, sustained fire rather than rapid volleys.
- Spatial Risk ( $r_{space}$ ): Calculated based on distance from the firing line using an inverse-square decay function.
- Temporal Risk ( $r_{time}$ ): Modeled via exponential decay over the 600-second duration, accounting for cumulative exposure.
Collapse Probability: An agent collapses (becomes a fatality/incapacitated obstacle) based on a probability function $P_{collapse}$ $P_{co l l a p se}$ that integrates spatial risk, temporal exposure, and crowd shielding.
- Shielding Parameter ( $\alpha$ ): A critical variable ( $0 \le \alpha \le 1$ $0 \leq α \leq 1$ ) governing the dual nature of density.
  - $\alpha = 1$ : Crowds provide protection (agents behind others are shielded).
  - $\alpha = 0$ : Crowds act as a liability (dense clusters attract concentrated fire, increasing secondary victimization).
Exit Constraints: Exits are modeled with a flow limiter (20% of theoretical capacity) to simulate bottlenecks and physical obstructions.

B. Social Psychological Analysis

The authors analyze historical texts (Hunter Committee Report, General Dyer's accounts) and early 20th-century literature (Le Bon, Taine) to demonstrate how the crowd was constructed as a "mindless, threatening mob." This dehumanization ("infrahumanisation") is presented as the cognitive framework that allowed Dyer to justify disproportionate force.

3. Key Contributions

Forensic Simulation of Historical Violence: The study introduces a novel application of agent-based modelling to reconstruct historical mass-casualty events where the crowd is the target, moving beyond standard evacuation models.
Quantitative Challenge to Official Records: The simulation provides a data-driven counter-narrative to the official death toll, demonstrating that even under conservative physical assumptions, the casualty count must be higher.
The "Dual Role" of Crowd Density: The model explicitly quantifies the tension between crowd density as a protective shield versus a liability that amplifies risk through cascading failures and targeted fire.
Interdisciplinary Synthesis: It bridges the gap between computational physics and social history, arguing that understanding crowd disasters requires analyzing both the physical mechanics of the event and the socio-cultural construction of the crowd.

4. Key Results

The authors ran 10 simulations for each scenario with varying crowd sizes ( $N = 5,000, 10,000, 15,000$ ) and shielding parameters ( $\alpha$ ).

Fatality Estimates:
- In all plausible scenarios, the simulated fatality counts significantly exceeded the official record of 379.
- Scenario ( $N=10,000$ , $\alpha=0.3$ - high liability): Mean fatalities $\approx 4,526$ (45% of the crowd).
- Scenario ( $N=10,000$ , $\alpha=0.7$ - high shielding): Mean fatalities $\approx 1,188$ (12% of the crowd).
- Even in the most "protective" scenarios, the death toll was 3–4 times higher than the official British estimate.
Spatial Distribution:
- Fatalities were not concentrated solely at the firing line. The highest density of casualties occurred deeper inside the garden (central and lower-right regions).
- This is due to bottleneck effects: As agents near the front collapsed or fled, they created blockages. Agents further back were trapped in high-density zones, exposed to prolonged fire, and unable to escape, leading to "cascading failures."
Non-Linearity of Density: Increasing crowd size led to a non-linear rise in fatalities. Higher density reduced the efficacy of shielding because the sheer volume of people and restricted movement created conditions where collapse propagated rapidly.
Sensitivity to Shielding: The parameter $\alpha$ drastically altered outcomes. When dense clusters attracted fire ( $\alpha \to 0$ ), the casualty rate skyrocketed, validating the hypothesis that targeted violence against dense crowds creates a "kill zone" effect beyond simple line-of-sight exposure.

5. Significance and Implications

Historical Accountability: The study provides a "scientific estimate" that supports long-standing claims that the British colonial record underreported the massacre's scale. It serves as a forensic tool to challenge narratives that minimize state violence.
Modern Crowd Safety: The findings are critical for contemporary security planning. Current models often fail to account for intentional external threats (active shooters). This research highlights that in such scenarios, crowd density can become a lethal factor due to congestion and secondary victimization, necessitating new safety protocols for public gatherings.
Societal Relevance: The paper draws a direct line from 1919 to modern political rhetoric (e.g., labeling protesters as "thugs," "vermin," or "mobs"). It argues that the dehumanizing language used by Dyer and the Hunter Committee is structurally similar to modern discourses that justify police brutality and state violence against dissent.
Ethical Modeling: The authors emphasize that while the simulation reduces human tragedy to data, its purpose is not sensationalism but to amplify the reality of the event and foster critical reflection on how crowds are perceived and treated by state power.

In conclusion, the paper demonstrates that the Jallianwala Bagh massacre was not merely a shooting but a complex systemic failure where architectural constraints, crowd dynamics, and dehumanizing ideology converged to produce a death toll far exceeding official records.