OPEN Universal bursty behaviour in human

SUBJECT AREAS:
violent conflicts
APPLIED PHYSICS S. Picoli1, M. del Castillo-Mussot2, H. V. Ribeiro1, E. K. Lenzi1 & R. S. Mendes1
STATISTICAL PHYSICS
1
Departamento de Fı́sica and National Institute of Science and Technology for Complex Systems, Universidade Estadual de
Received Maringá, 87020-900 Maringá, Paraná, Brazil, 2Instituto de Fı́sica, Universidad Nacional Autónoma de México, 04510 México
D.F., Mexico.
23 December 2013
Accepted
2 April 2014 Understanding the mechanisms and processes underlying the dynamics of collective violence is of
considerable current interest. Recent studies indicated the presence of robust patterns characterizing the
Published size and timing of violent events in human conflicts. Since the size and timing of violent events arises as the
24 April 2014 result of a dynamical process, we explore the possibility of unifying these observations. By analyzing
available catalogs on violent events in Iraq (2003–2005), Afghanistan (2008–2010) and Northern Ireland
(1969–2001), we show that the inter-event time distributions (calculated for a range of minimum sizes)
obeys approximately a simple scaling law which holds for more than three orders of magnitude. This robust
Correspondence and pattern suggests a hierarchical organization in size and time providing a unified picture of the dynamics of
requests for materials violent conflicts.
should be addressed to
S.P. (spjunior@dfi.

I
ndividual and collective violence has a large impact on society1–3–implications for health, for instance, range
uem.br)
from psychological distress to injuries and mortality. Recent estimates of mortality from war and interpersonal
violence reported values around 0.5–1 million deaths annually1,2, while the non-lethal impacts may affect
millions more. In the light of the current status of violence and their impact on society, the knowledge of
dynamical properties of violent conflicts is of interest not only for theoretical reasons.
Fundamental questions in the study of collective violence concern to the estimation of the number of conflict-
related deaths in a given attack and how often the attacks will take place. In this direction, it has been reported a
robust power-law distribution describing the number of conflict-related deaths which holds for wars, insurgent
conflicts and acts of terrorism4–7. Robust patterns have also been reported for the timing of events7–9, like the
evidence for heterogeneous temporal behaviour characterizing violent conflicts7. More recently, some of these
findings have been extended to include a wide range of human confrontations10.
The heterogeneous temporal behaviour of violent events and the statistics of event sizes are both the result of a
dynamical process. Despite this fact, sizes and times of violent events have been studied separately. Here we
explore the possibility that both sizes and times in human conflicts can be characterized by a common approach.
By analyzing available catalogs of violent events in distinct parts of the world, we show that the inter-event time
distributions calculated for a range of values of minimum sizes can be approximately described by a single master
curve which holds for more than three orders of magnitude. This result signalizes the existence of a scaling law
expressing a hierarchical organization of times and sizes in violent conflicts.

Results
Our findings are based in the analysis of available catalogs on violent events occurred in Iraq7 (2003–2005),
Afghanistan11 (2008–2010) and Northern Ireland12 (1969–2001) which gives the estimated number of deaths s in
events occurred at time t (for details on the datasets see Methods). We defined the inter-event time t as the time
interval between the beginning of two successive events, tj 5 tj 2 tj21. Here we focus on the inter-event times tj
with j labelling only the events with size s $ smin, where smin is a minimum size representing a lower threshold.
Figure 1 shows the empirical raw data s versus t (Fig. 1a–1c) and their correspondent inter-event times t (Fig. 1d)
Ðt
for particular values of smin. The growth curves represent the cumulative number of events, n(t)~ 0 r(t 0 )dt 0
where r is the activity rate defined as the number of events greater than smin per unit of time. For a given smin we
also define the mean activity rate Æræ which is the averaged rate over the whole time period.
We then obtained P(t), the inter-event time distribution covering events with s $ smin (see Methods). Figure 2a
shows P(t) for the distinct conflicts and different values of smin. All curves display fat tails which can be viewed by
the broad range of values of the inter-event times. As expected, the curves are scattered reflecting quantitative

SCIENTIFIC REPORTS | 4 : 4773 | DOI: 10.1038/srep04773 1

Figure 1 | Event sizes and inter-event times. The estimated number of deaths s (in logarithmic scale) in a given event occurred at time t (in days, counted
from a given initial day) for conflicts occurred in (a) Iraq, with s $ 8 in the period 2003–2005; (b) Northern Ireland, s $ 4 (1969–2001); and (c)
Afghanistan, s $ 1 (2008–2010). The growth curves represent the cumulative number of events, n(t). For comparison, Æræ 5 0.30 (Iraq); Æræ 5 0.01
(Northern Ireland); and Æræ 5 0.12 (Afghanistan), where Æræ is the mean activity rate for the given values of minimum size. (d) Inter-event times t obtained
from the data shown in (a), (b) and (c) for the corresponding values of minimum size.

differences in the timing of events for distinct conflicts and minimum Discussion
sizes. In order to compare the shape of the different curves we The broad distributions shown in Fig. 2a reflect a large variability of
rescaled the data using the mean activity rate Æræ and plot Æræ21P(t) the inter-event times characterizing heterogeneous temporal beha-
versus Æræt. The rescaling causes a s-dependent shift of the curves viour. This result is consistent with previous findings7, pointing to a
since Æræ is a function of smin. Figure 2b shows that the different curves scenario where high-activity intervals alternate with long low-activ-
approximately collapsed into one master curve, which holds for more ity periods. Bursty behaviour has also been reported for non-violent
than three orders of magnitude. The data collapse signalizes the human activities and different explanations about the origin of these
existence of a scaling law in the form P(t) 5 ÆræF(Æræt), where the bursts have been proposed13–19.
scaling function F displayed distinct behaviours for small and large The data collapse shown in Fig. 2b signalizes the existence of a
values of the scaled variable. For comparison, we show a power-law scaling law governing the inter-event time distributions at distinct
decay with exponent 0.75 for small times and another power-law minimum sizes in violent conflicts. In the range considered, the
with exponent 2.6 for larger times (Fig. 2b). dependence on the sizes are approximately governed by the mean

Figure 2 | Inter-event time distributions with and without data rescaling. (a) Inter-event time distribution, P(t), calculated for the distinct conflicts
(Iraq, N. Ireland and Afghanistan) and different minimum sizes. (b) The same data, but using the scaled variables Æræt and Æræ21P(t). The dashed lines are
power-law decays–with exponent 0.75 for small times and with exponent 2.6 for large times.

SCIENTIFIC REPORTS | 4 : 4773 | DOI: 10.1038/srep04773 2

activity rate Æræ. Since the mean activity rate is linked to the distri- 2. Lozano, R. et al. Global and regional mortality from 235 causes of death for 20 age
bution of sizes, the collapse of the curves provides a unified picture of groups in 1990 and 2010: A systematic analysis for the Global Burden of Disease
Study 2010. Lancet 380, 2095–2128 (2012).
the size structure and temporal properties of the conflicts. Our find- 3. Krug, E. G., Mercy, J. A., Dahlberg, L. L. & Zwi, A. B. The world report on violence
ings suggest that the power-law type decay of the inter-event time and health. Lancet 360, 1083–1088 (2002).
distributions at small times (characterizing bursty behaviour) is just 4. Richardson, L. F. Variation of the frequency of fatal quarrels with magnitude. J.
the short time limit of a general hierarchical scaling phenomenon Am. Stat. Assoc. 43, 523–546 (1948).
which occurs in a wider range of time (and size) scales. 5. Cederman, L.-E. Modeling the size of wars: From billiard balls to sandpiles. Am.
Polit. Sci. Rev. 97, 135–150 (2003).
Our results contain some elements of similarity with previous find- 6. Clauset, A., Young, M. & Gleditsch, K. S. On the frequency of severe terrorist
ings in natural phenomena. For example, it has been shown that the events. J. Conflict Resolut. 51, 58–87 (2007).
inter-event time distributions for earthquakes whose magnitudes are 7. Bohorquez, J. C., Gourley, S., Dixon, A. R., Spagat, M. & Johnson, N. F. Common
greater than a lower threshold (taken over different spatial windows) ecology quantifies human insurgency. Nature 462, 911–914 (2009).
collapse onto a single curve, providing a basis for the proposed unified 8. Johnson, N. et al. Pattern in escalations in insurgent and terrorist activity. Science
333, 81–84 (2011).
scaling law20. This general scaling phenomenon captures elements of 9. Clauset, A. & Gleditsch, K. S. The developmental dynamics of terrorist
the Gutenberg-Richter law21,22 and the Omori law23 for earthquakes organizations. PLoS One 7, e48633 (2012).
unifying these observations. A similar scaling behaviour for the inter- 10. Johnson, N. et al. Simple mathematical law benchmarks human confrontations.
event time distribution has been reported in fracture experiments, Sci. Rep. 3, 3463 (2013).
even in the absence of spatial degrees of freedom24. The data collapse 11. Carran, S., Ravindar, A., Lau, S. Y. & Bohannon, J. Afghanistan Casualty Timeline
(2008–2010), 11 March 2011. https://www.sciencemag.org/content/331/6022/
shown in Fig. 2b for violent conflicts is remarkably similar those 1256/suppl/DC1. Date of access: 20/02/2012.
obtained in earthquakes and fracture experiments20,24. Specifically, 12. Sutton, M. An Index of Deaths from the Conflict in Ireland. http://cain.ulst.ac.uk/
the shape of the scaling function F and even the reported values of sutton. Date of access: 20/02/2012.
the power-law exponents for small and large times are in good agree- 13. Barabási, A. L. The origin of bursts and heavy tails in human dynamics. Nature
ment with the results shown in Fig. 2b for human conflicts. 435, 207–211 (2005).
14. Vazquez, A. et al. Modelling bursts and heavy tails in human dynamics. Phys. Rev.
Exponents around 1 characterizing the inter-event time distri- E 73, 036127 (2006).
bution at small times arises in seismic-like phenomena as a con- 15. Malmgren, R. D. et al. A poissonian explanation for heavy tails in e-mail
sequence of the Omori law, while exponents around 2.5 at large communication. Proc. Natl. Acad. Sci. 105, 18153–18158 (2008).
inter-event times have been explained in terms of temporal varia- 16. Stehlé, J., Barrat, A. & Bianconi, G. Dynamical and bursty interactions in social
tions of the activity rate20,24. The power-law exponent <0.75 for small networks. Phys. Rev. E 81, 035101 (2010).
17. Ratkiewicz, J., Flammini, A., Menczer, F. & Vespignani, A. Characterizing and
times (Fig. 2b) does not necessarily imply an underlying Omori law
modeling the dynamics of online popularity. Phys. Rev. Lett. 105, 158701 (2010).
governing the timing of violent conflicts but points to this possibility. 18. Jo, H.-H., Karsai, M., Kertész, J. & Kaski, K. Cicardian pattern and burstiness in
In addition, the power law exponent <2.6 for large times (Fig. 2b) mobile phone communication. New J. Phys. 14, 013055 (2012).
could be related to temporal variations of the activity rate of violent 19. Karsai, M., Kaski, K., Barabási, A. L. & Kertész, J. Universal features of correlated
events–which can be seem in the non-linear behaviour of the growth bursty behavior. Sci. Rep. 2, 397 (2012).
curves shown in Figs. 1a–1c. 20. Bak, P., Christensen, K., Danon, L. & Scanlon, T. Unified Scaling Law for
Earthquakes. Phys. Rev. Lett. 88, 178501 (2002).
Despite the fact that human activities and natural phenomena are 21. Gutenberg, B. & Richter, C. Seismicity of the Earth (Princeton University,
very different in nature, it has been suggested that both could be Princeton, 1949).
described by a common approach5,25. For example, the occurrence 22. Turcotte, D. L. Fractals and Chaos in Geology and Geophysics (Cambridge
of earthquakes has been related to the relaxation of accumulated University Press, Cambridge, 1997).
stress after reaching a threshold as in self-organized criticality 23. Omori, F. On the aftershocks of earthquakes. J. Coll. Sci. Imper. Univ. Tokyo. 7,
111–200 (1894).
(SOC)20,26. Analogously, violent events in human conflicts could be 24. Baró, J. et al. Statistical similarity between the compression of a porous material
associated with a threshold mechanism. In this scenario, a descrip- and earthquakes. Phys. Rev. Lett. 110, 088702 (2013).
tion of human conflicts in terms of SOC seems plausible. Our find- 25. Roberts, D. C. & Turcotte, D. L. Fractality and Self-Organized Criticality of Wars.
ings are consistent with this possibility, providing quantitative Fractals 6, 351–357 (1998).
support for the analogy between patterns in human conflicts and 26. Bak, P. How Nature Works: The Science of Self-Organized Criticality (Copernicus,
Springer-Verlag New York, 1996).
natural phenomena exhibiting SOC.

Methods
Data source and processing. To study the timing of violent conflicts we select three Acknowledgments
datasets containing time-stamped records–(a) Iraq; (b) Afghanistan; and (c) We thank CNPq and Fundação Araucária (Brazilian Agencies) and DGAPA-UNAM
Northern Ireland datasets. Iraq data were obtained from the supplementary (Mexican Agency) for financial support.
information of Ref. 7, covering 3000 events during 803 days within 2003–2005. Each
record gives the size of the event (the estimated number of deaths) and the day that the
event occurred or the day that the event started, in the case of multi-day attacks. Author contributions
Afghanistan data, from ref. 11, consist of media-reported civilian deaths occurred in a S.P., M.C.-M. and R.S.M. designed the research. S.P. and R.S.M. analyzed the data and wrote
period of 1056 days within January 2008–December 2010–a total of 123 events the main manuscript text. H.V.R. and E.K.L. contributed to the preparation of the figures.
containing at least one killed. Northern Ireland data are based on Ref. 12, a dataset All authors reviewed the manuscript.
covering 11,839 days within July 1969–December 2001. Unique events were obtained
by aggregating single deaths by day, giving 2143 events in the period considered. We
consider all deaths in each dataset, independently of the group that performed the Additional information
attack (e.g. non-state groups or state security organizations). Competing financial interests: The authors declare no competing financial interests.
How to cite this article: Picoli, S., del Castillo-Mussot, M., Ribeiro, H.V., Lenzi, E.K. &
Inter-event time distributions. For each dataset, we obtained the consecutive inter- Mendes, R.S. Universal bursty behaviour in human violent conflicts. Sci. Rep. 4, 4773;
event times at a given smin and removed null time intervals (t 5 0) in order to study DOI:10.1038/srep04773 (2014).
temporally separated events. The distributions P(t) for different conflicts and values
of smin correspond to the probability density function (pdf) of the data. The number of This work is licensed under a Creative Commons Attribution 3.0 Unported License.
bins was computed based on the length of data and the binning procedure was applied The images in this article are included in the article’s Creative Commons license,
on log-transformed data. unless indicated otherwise in the image credit; if the image is not included under
the Creative Commons license, users will need to obtain permission from the license
holder in order to reproduce the image. To view a copy of this license, visit
1. Mathers, C., Boerma, T. & Ma Fat, D. The Global Burden of Disease: 2004 Update
http://creativecommons.org/licenses/by/3.0/
(World Health Organization, 2008).

SCIENTIFIC REPORTS | 4 : 4773 | DOI: 10.1038/srep04773 3