arxiv: 2604.27737 · v1 · submitted 2026-04-30 · ⚛️ physics.soc-ph

Recognition: unknown

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

Ezel \"Usten, Krisztina Konya, Mohcine Chraibi

Authors on Pith no claims yet

Pith reviewed 2026-05-07 06:00 UTC · model grok-4.3

classification ⚛️ physics.soc-ph

keywords agent-based modelingcrowd dynamicsJallianwala Bagh massacreAmritsar 1919social psychologycrowd psychologyhistorical simulationpolitical violence

0 comments

The pith

Agent-based simulations of the 1919 Amritsar massacre show death tolls well above the official count of 379, even under conservative assumptions about shooting and crowd movement.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper applies agent-based modeling to reconstruct the physical dynamics of the crowd during the Jallianwala Bagh shooting in Amritsar. It demonstrates that with moderate rates of fire, some bodies shielding others, and limited ways to escape, the number of fatalities consistently exceeds the recorded 379. The authors also examine how theories of crowd psychology from the era portrayed gatherings as irrational and dangerous, which justified the level of force used. If accurate, this reframing calls for a reassessment of the historical event and highlights risks in how authorities perceive and respond to crowds today.

Core claim

Through agent-based simulations incorporating physical constraints such as shooting cadence, crowd shielding, and escape routes, the model produces fatality estimates significantly higher than the official figure. Concurrently, the study traces how contemporary discourses on crowd psychology shaped the decision to use deadly force on the gathering.

What carries the argument

Agent-based modelling of individual trajectories and interactions within the crowd under gunfire, integrated with analysis of historical crowd psychology discourses.

If this is right

Revised estimates of the massacre's scale suggest official records are incomplete.
Similar physical and psychological factors may apply to understanding other historical crowd events.
The socio-psychological construction of crowds influences state responses to collective action.
Interdisciplinary modeling aids both historical analysis and modern crowd safety planning.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

Contemporary crowd control strategies could incorporate these simulation techniques to assess risks more accurately.
The persistence of viewing crowds as irrational may still affect policing of protests and public gatherings.
Applying the model to other documented crowd incidents could test its broader validity.
Physical simulations alone do not address intent but can quantify plausible outcomes under given conditions.

Load-bearing premise

The model assumes moderate shooting cadence, effective crowd shielding by bodies, and highly constrained escape routes accurately reflect the actual conditions at the time.

What would settle it

Archival ballistic evidence or survivor accounts that establish a much lower effective firing rate or more open escape paths than those used in the conservative simulation scenarios.

read the original abstract

Crowds have long held a paradoxical place in the human imagination, feared for their destructive potential yet essential for collective expression. This tension was tragically manifested in the 1919 Jallianwala Bagh massacre, when British colonial troops opened fire on a peaceful gathering in Amritsar, India. Although officially 379 deaths were recorded, eyewitnesses and historians have long challenged this figure. With this study, we critically revisit the events through the lens of the specific role of the crowd as a phenomenon, both regarding the physical and the socio-psychological dynamics. We show that even under conservative physical assumptions - moderate shooting cadence, crowd-shielding, and constrained escape routes - our agent-based simulations consistently yield fatality estimates well above the official death count. On the socio-psychological front, we explore how early 20th-century discourses, influenced by Le Bon's theory of crowd psychology, constructed the crowd as an inherently irrational and threatening entity, thus providing a rationale for the application of excessive force. Our findings show that acknowledging the socio-cultural construction of crowds as a relevant factor in how state power engage with and respond to collective gatherings brings to light contemporary parallels and the risks posed by their rhetorical framing. Furthermore, this study highlights the importance of interdisciplinary modelling for both historical accountability and current crowd safety, particularly in an era of growing political unrest, surveillance, and militarised crowd policing.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

ABM simulations claim higher Amritsar deaths under conservative assumptions, but lack the details to evaluate the implementation.

read the letter

The key takeaway is that the authors ran agent-based simulations of the Jallianwala Bagh crowd and concluded that even moderate shooting rates, with some shielding and limited exits, produce more deaths than the recorded 379. They pair this with a discussion of how Le Bon's ideas about irrational crowds helped justify the British response. What works is the attempt to quantify a historical event using modern crowd modeling tools while connecting it to the psychological and rhetorical framing at the time. This kind of mixed physical and social analysis is not common for events like this, and the modern parallels they draw about crowd rhetoric and policing are relevant. The main weakness is that none of the modeling details are provided. There are no equations, no listed parameter values beyond mentioning shooting cadence and crowd size, no tests for how sensitive the results are to those choices, and no comparison to actual shooting data or post-event accounts. The fatality claim stands or falls on the specific rules for how agents block each other from bullets and how they try to escape the space. Without seeing the code or the exact geometry used, it's impossible to know if the higher numbers are a real finding or just a result of how the model was set up. The concern about unverified shielding and escape routes looks like it needs addressing. This paper would be of interest to people in social physics who do historical case studies, or to historians open to quantitative methods. It could also appeal to those studying crowd safety and state responses to protests. I would send it to peer review. The core idea is worth exploring, but the authors need to supply the missing technical information and run the checks that are standard in ABM work.

Referee Report

2 major / 2 minor

Summary. The paper claims that agent-based simulations of the 1919 Jallianwala Bagh massacre, incorporating conservative physical assumptions (moderate shooting cadence, crowd-shielding, and constrained escape routes), produce fatality estimates consistently above the official figure of 379. It combines this quantitative modeling with a socio-psychological analysis drawing on Le Bon's crowd psychology to explain the rationale for excessive force, and draws parallels to contemporary crowd management and political unrest.

Significance. If the ABM results prove robust upon full specification and validation, the work would provide a quantitative reframing of a key historical event through crowd dynamics, potentially supporting revised fatality estimates and illustrating how physical constraints interact with socio-psychological perceptions of crowds. The interdisciplinary integration of agent-based modeling with historical and psychological analysis offers a template for similar studies, with relevance to modern crowd safety and policing in contexts of political unrest.

major comments (2)

[Abstract and model description] Abstract and model description (likely §3 or §4): The central claim that simulations 'consistently yield fatality estimates well above the official death count' even under conservative assumptions is unsupported because the manuscript supplies no model equations, no numerical values for free parameters such as shooting cadence or initial crowd size/density, no explicit implementation details for crowd-shielding (e.g., ray-tracing occlusion versus probabilistic density reduction), and no description of escape geometry matching the documented narrow gate and wall layout of Jallianwala Bagh. Without these, it is impossible to verify that the assumptions are conservative rather than tuned to inflate lethality.
[Results section] Results section (likely §5): No sensitivity analysis, calibration against ballistic data or post-event reconstructions, error bars, or validation metrics are reported for the fatality estimates. This omission directly undermines the claim of consistent results 'well above 379' and leaves open whether the higher estimates arise from the physical model or from unstated implementation choices.

minor comments (2)

[Discussion] The socio-psychological discussion would benefit from explicit section headings or a dedicated subsection separating the Le Bon analysis from the ABM results to improve readability.
[Figures] Figure captions (if present) should include the specific parameter values used in the displayed simulation runs to allow readers to connect visuals to the (currently missing) methods.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their valuable feedback on our manuscript. The comments highlight important areas for improvement in the presentation of our agent-based model and results. We have prepared a revised version that incorporates detailed model specifications, parameter values, sensitivity analyses, and validation procedures to address these concerns. We believe these revisions will enhance the transparency and verifiability of our findings while preserving the interdisciplinary approach of the study.

read point-by-point responses

Referee: Abstract and model description (likely §3 or §4): The central claim that simulations 'consistently yield fatality estimates well above the official death count' even under conservative assumptions is unsupported because the manuscript supplies no model equations, no numerical values for free parameters such as shooting cadence or initial crowd size/density, no explicit implementation details for crowd-shielding (e.g., ray-tracing occlusion versus probabilistic density reduction), and no description of escape geometry matching the documented narrow gate and wall layout of Jallianwala Bagh. Without these, it is impossible to verify that the assumptions are conservative rather than tuned to inflate lethality.

Authors: We accept that the manuscript as submitted lacks the level of detail necessary for full verification of the model. In the revised manuscript, we will expand Section 3 to include the complete model equations for agent navigation, interaction, and ballistic simulation. Specific parameter values will be provided, including a shooting cadence of 10-15 rounds per minute per soldier (drawn from historical accounts of Lee-Enfield rifle usage), an initial crowd size of 10,000-15,000 agents calibrated to contemporary estimates, and a density of 4-6 persons per square meter in the central area. Crowd-shielding is implemented via a line-of-sight ray-tracing method where each bullet trajectory is checked for occlusion by intervening agents, with a density-dependent probability of hit reduction. The escape geometry replicates the historical Jallianwala Bagh layout, featuring a single 3-meter-wide gate and 3-meter-high walls enclosing the garden. These additions will allow readers to confirm the conservative nature of the assumptions. We have also added a supplementary file with the full model code for reproducibility. revision: yes
Referee: Results section (likely §5): No sensitivity analysis, calibration against ballistic data or post-event reconstructions, error bars, or validation metrics are reported for the fatality estimates. This omission directly undermines the claim of consistent results 'well above 379' and leaves open whether the higher estimates arise from the physical model or from unstated implementation choices.

Authors: We agree with the referee that additional analyses are required to substantiate the robustness of the fatality estimates. The revised results section will include a sensitivity analysis varying shooting cadence by ±30%, crowd density by ±20%, and escape route capacity, with results presented as mean fatalities with standard error bars from 500 simulation runs. We will calibrate the model using ballistic data from period weapons and compare simulated outcomes to documented crowd dynamics in similar historical events. Validation will involve metrics such as the proportion of agents escaping versus historical accounts and a discussion of model assumptions. These will be added as new figures and tables in Section 5, along with a limitations subsection. While we maintain that the qualitative finding of higher fatalities holds, these quantitative enhancements will address the potential for implementation-specific biases. revision: yes

Circularity Check

0 steps flagged

No circularity: simulation outputs rest on independent modeling assumptions rather than definitional reduction or self-citation chains

full rationale

The paper's central claim—that agent-based simulations produce fatality estimates above the official count even under stated conservative assumptions—is presented as an empirical result of the ABM runs. No equations, parameter-fitting procedures, or self-citations appear in the provided text that would reduce the output by construction to the inputs (e.g., no fitted shooting cadence renamed as a prediction, no uniqueness theorem imported from prior author work, and no ansatz smuggled via citation). The derivation therefore remains self-contained: the model is run with explicit (if undetailed here) physical rules for cadence, shielding, and geometry, and the higher estimate is reported as a consequence rather than a tautology. Absent load-bearing self-referential steps, the circularity score is zero.

Axiom & Free-Parameter Ledger

2 free parameters · 2 axioms · 0 invented entities

The claims rest on unstated model parameters for shooting cadence and crowd geometry plus interpretive assumptions about historical psychological influences; these are not independently derived or validated in the provided abstract.

free parameters (2)

shooting cadence
Described as moderate and conservative but no numerical value or derivation from data is given in the abstract.
crowd initial size and density
Must be assumed from historical estimates to run the simulation; choice directly affects fatality output.

axioms (2)

domain assumption Simple agent-based rules for movement, shielding, and escape can faithfully reproduce physical fatality counts in a dense crowd under gunfire.
Invoked by the use of ABM to challenge the official death count.
domain assumption Le Bon's crowd psychology accurately captures the early 20th-century discourses that shaped British colonial responses to gatherings.
Used to explain the rationale for applying excessive force.

pith-pipeline@v0.9.0 · 5563 in / 1749 out tokens · 75773 ms · 2026-05-07T06:00:17.803935+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

27 extracted references · 12 canonical work pages

[1]

Hunter, G

W. Hunter, G. C. Rankin, W. E. Rice,et al., Disorders Inquiry Committee Report 1919–1920 (1920), official colonial report investigating unrest in India

1919
[2]

K. A. Wagner,Amritsar 1919: An Empire of Fear and the Making of a Massacre(Yale University Press) (2019), doi:10.2307/j.ctvcb5btz

work page doi:10.2307/j.ctvcb5btz 1919
[3]

euronews.com/my-europe/2024/12/21/timeline-deadly-attacks-on-christmas -markets-in-europe, accessed: 2025-07-17

Euronews, Timeline: Deadly attacks on Christmas markets in Europe (2024),https://www. euronews.com/my-europe/2024/12/21/timeline-deadly-attacks-on-christmas -markets-in-europe, accessed: 2025-07-17

2024
[4]

R. A. Nye,The Origins of crowd psychology: Gustave Le Bon and the crisis of mass democracy in the third republic, no. 2 in Sage studies in 20th century history (Sage publications) (1975)

1975
[5]

Le Bon,The crowd: A study of the popular mind(Internet Archive) (1895),https: //archive.org/details/the-crowd-a-study-of-the-popular-mind-by-gustave -le-bon/page/80/mode/2up

G. Le Bon,The crowd: A study of the popular mind(Internet Archive) (1895),https: //archive.org/details/the-crowd-a-study-of-the-popular-mind-by-gustave -le-bon/page/80/mode/2up
[6]

D. Palano, The Science of Crowds: A Genealogical Analysis of Gustave Le Bon’s Collective Psychology.Genealogy9(2), 38 (2025), doi:10.3390/genealogy9020038,https://www.md pi.com/2313-5778/9/2/38

work page doi:10.3390/genealogy9020038 2025
[7]

Borch, B

C. Borch, B. Schiermer, Crowd and Collective Behavior, inHandbook of Classical Sociological Theory, S. Abrutyn, Ed., Handbooks of Sociology and Social Research (Springer International Publishing AG), pp. 439–465 (2021), doi:10.1007/978-3-030-78205-420

work page doi:10.1007/978-3-030-78205-420 2021
[8]

Barrows,Distorting mirrors: visions of the crowd in late nineteenth-century France, no

S. Barrows,Distorting mirrors: visions of the crowd in late nineteenth-century France, no. 127 in Yale historical publications Miscellany (Yale Univ. Pr) (1981),https://archive.org/ details/distortingmirror0000barr/page/n5/mode/2up

1981
[9]

P. Jupp, E. Magennis, Introduction: Crowds in Ireland, c. 1720–1920, inCrowds in Ireland, c. 1720–1920, P. Jupp, E. Magennis, Eds. (Palgrave Macmillan UK), pp. 1–42 (2000), doi: 10.1057/9780230288058 1,https://doi.org/10.1057/9780230288058_1. 26

work page doi:10.1057/9780230288058 1920
[10]

Moscovici,The age of the crowd: a historical treatise on mass psychology(Cambridge University Press) (1985)

S. Moscovici,The age of the crowd: a historical treatise on mass psychology(Cambridge University Press) (1985)

1985
[11]

Thonhauser, A critique of the crowd psychological heritage in early sociology, classic phenomenology and recent social psychology.Continental philosophy reviewpp

G. Thonhauser, A critique of the crowd psychological heritage in early sociology, classic phenomenology and recent social psychology.Continental philosophy reviewpp. 1–19 (2022), doi:10.1007/s11007-022-09566-z

work page doi:10.1007/s11007-022-09566-z 2022
[12]

Tordeux, M

A. Tordeux, M. Chraibi, A. Seyfried, Collision-Free Speed Model for Pedestrian Dynamics, inTraffic and Granular Flow ’15, V. L. Knoop, W. Daamen, Eds. (Springer International Publishing, Cham) (2016), pp. 225–232, doi:10/ggswz2

2016
[13]

K. D. Malaviya,Open Rebellion in the Punjab : With Special Reference to Amritsar(Allahabad : Abhyudaya Press) (1919)

1919
[14]

Haslam,Dehumanization and intergroup relations.(American Psychological Association), pp

N. Haslam,Dehumanization and intergroup relations.(American Psychological Association), pp. 295–314 (2015), doi:10.1037/14342-011,https://content.apa.org/books/1434 2-011

work page doi:10.1037/14342-011 2015
[15]

Garc ´ıa-Garc´ıa, After the Great War: Nationalism, Degenerationism and Mass Psychology

J. Garc ´ıa-Garc´ıa, After the Great War: Nationalism, Degenerationism and Mass Psychology. Journal of Social and Political Psychology3(1), 103–123 (2015), doi:10.5964/jspp.v3i1.371, https://jspp.psychopen.eu/index.php/jspp/article/view/4847

work page doi:10.5964/jspp.v3i1.371 2015
[16]

Wines, Trump Tweets ’When the Looting Starts, the Shooting Starts’ (2020),https: //www.nytimes.com/2020/05/29/us/looting-starts-shooting-starts.html, accessed June 24, 2025

M. Wines, Trump Tweets ’When the Looting Starts, the Shooting Starts’ (2020),https: //www.nytimes.com/2020/05/29/us/looting-starts-shooting-starts.html, accessed June 24, 2025

2020
[17]

Weak” Governors “Dom- inate

B. Levin, Trump, Who Spent the Weekend in a Bunker, Demands “Weak” Governors “Dom- inate” Protesters (2020),https://www.vanityfair.com/news/2020/06/man-who-spe nt-the-weekend-in-a-bunker-demands-weak-governors-dominate-protesters, accessed June 24, 2025

2020
[18]

A. Slisco, North Carolina Republican Lawmaker Calls Black Lives Matter Protesters ‘Vermin’ (2020),https://www.newsweek.com/north-carolina-republican-lawmaker-bla ck-lives-matter-protesters-vermin-1511381, accessed June 24, 2025. 27

2020
[19]

B. News, Here’s What China Is Telling Its People About Hong Kong Protests (2019),https: //web.archive.org/web/20191217053028/https://www.bnnbloomberg.ca/here-s -what-china-is-telling-its-people-about-hong-kong-protests-1.1347093, archived version, accessed June 24, 2025

work page arXiv 2019
[20]

E. Barry, Putin Defends Vote, but Offers Concession (2011),https://www.nytimes.com/ 2011/12/16/world/europe/putin-answers-questions-on-television-on-russi an-election.html, accessed June 24, 2025

2011
[21]

Wikipedia contributors, Chapulling (2013),https://en.wikipedia.org/wiki/Chapul ling, accessed June 24, 2025

2013
[22]

H. Tas ¸, Adding Insult to Injury: Vilification as Counter-Mobilization in Turkey’s Gezi Protests (2013),https://pomeps.org/adding-insult-to-injury-vilification-as-count er-mobilization-in-turkeys-gezi-protests, accessed June 24, 2025

2013
[23]

Haslam, S

N. Haslam, S. Loughnan, Dehumanization and Infrahumanization.Annual Review of Psychol- ogy65(1), 399–423 (2014), doi:10.1146/annurev-psych-010213-115045

work page doi:10.1146/annurev-psych-010213-115045 2014
[24]

Zlobina, M

A. Zlobina, M. L. Bettinsoli, M. P. Miranda, M. Formanowicz, Back to Basics: Human Rights Violations and Dehumanization.Current Opinion in Behavioral Sciences51, 101263 (2023), doi:10.1016/j.cobeha.2023.101263

work page doi:10.1016/j.cobeha.2023.101263 2023
[25]

J. I. Wani, ‘Using a Blacksmith’s Hammer to Crush a Fly’: Jallianwala Bagh, Public Order and Popular Protests in Late Colonial India.History and Sociology of South Asia14(1-2), 7–21 (2020), doi:10.1177/22308075211049835

work page doi:10.1177/22308075211049835 2020
[26]

M. Chraibi, Rethinking the Amritsar Massacre through Agent-Based Modelling and Social Psychology,https://doi.org/10.5281/zenodo.16264825(2025), doi:10.5281/zenodo .16264825, zenodo software release, version v1.0

work page doi:10.5281/zenodo.16264825(2025 2025
[27]

Chraibi, K

M. Chraibi, K. Kratz, T. Schr ¨odter, The JuPedSim Development Team, JuPedSim,https: //github.com/PedestrianDynamics/jupedsim. 28