On 11 December 2006, former Mexican President Felipe Calderón deployed troops to fight the country’s increasingly powerful drug cartels, plunging Mexico into a war in which more than 100,000 people have been killed or disappeared. Now, a new study uses statistics and complex networks analysis to reveal the patterns by which violence spread across the country between 2007 and 2011—the last year for which records are available. The results may contribute to the debate about how effective the government’s policy of attacking cartel leaders has been in reducing violence, experts say.
This approach “represents an attempt to reveal the actual dynamics of drug violence [and demonstrates] how the conflict actually unfolds and evolves,” says Michael Lawrence, a Ph.D. candidate at the University of Waterloo in Canada, whose work has focused on the application of complexity science to issues of conflict and security and who was not involved in the research.
Jesús Espinal, a quantitative analyst at the National Institute of Genomic Medicine in Mexico City, remembers the night the drug war began to hit home for him. In 2009, he was a Ph.D. student at the National Autonomous University of Mexico (UNAM), Cuernavaca. He was on his way home when officers at a military roadblock told him they were running an operation and sent him back to his lab with instructions to stay put. He found out later that the Mexican Navy had shot down Arturo Beltrán Leyva, one of the most powerful drug lords in Mexico, just 6 kilometers from his lab. Like many in Cuernavaca, Espinal hoped that violence would recede after the kingpin’s death. Instead, it spiraled out of control. Cuernavaca “started to get violent as hell, to the point that people I knew were murdered,” Espinal recalls.
So Espinal teamed up with UNAM statistical physicist Hernán Larralde to try to find a pattern behind how and why the violence was spreading. A few months after Beltrán Leyva was killed, they dove into the official reports of drug-related homicides in Mexico, zeroing in on locations and dates and mapping each outbreak over time. Going month by month, they built a complex network that could illustrate correlations between cities where violence increased or decreased at the same time. If cities shared a death rate higher than 70 casualties per 100,000 inhabitants in a year and were less than 200 kilometers apart, Larralde and Espinal linked them together on their map to tease out broader geographical patterns.
The results were disheartening. In 2008, you could already travel from the northern state of Sonora to the Isthmus of Tehuantepec in the south by visiting only neighboring violent cities. In 2010, the northeastern states of Tamaulipas and Nuevo León suddenly became war zones. And by 2011 you could cross Mexico from north to south and from east to west driving only through places with a higher death rate than the one reported in Iraq during the U.S. invasion, Espinal and Larralde reported on 28 May in PLOS ONE. Still, carnage was not spread evenly over the country. “There are a bunch of cities with lots of killings and many cities with low drug-related murders,” Larralde explains.
The analysis also reveals that narco-related violence does not spread by geographical proximity, as an epidemic would. “It is not a medieval war that slowly expands across the country,” says Carlos Gershenson, a computer scientist at UNAM Mexico City, who wasn’t involved in the research. “Rather, the violence synchronizes in cities that are thousands of miles apart without affecting some in the middle.” The reasons for why far-away cities become violent at the same time are unclear, since the study doesn’t address causation. But the bloodshed does appear to cluster around a small core of cities, including Juárez, Acapulco, Cancún, Culiacán, Monterrey, Tampico, and Tijuana. Thinking of these places as central nodes in the web of narco violence may help researchers and politicians “generate better violence-reduction strategies amidst the drug war,” Lawrence says.
So should Mexican authorities focus on reducing violence in the central nodes, in the hopes that peace would ripple out from there? Not so fast, Lawrence and others say. “Attacking the central cities [of the network] could either be a great idea or a terrible idea,” warns Andrés Monroy-Hernández, a social computing researcher from Microsoft Research in Seattle, Washington, who has studied social media’s role in the Mexican drug war. Sure, it may temporarily stem the bloodshed, but it may also cause once centralized cartels to splinter into smaller networks, making the violence even more difficult to control. That’s what happened in 2010 after the head of the Gulf Cartel was imprisoned and its former armed wing, Los Zetas, emerged as one of the most aggressive drug trafficking gangs.
Mapping past violence alone doesn’t help predict where the drug war will spread in the future, Monroy-Hernández says. In order to do that, researchers need more data on daily death rates and geographical variations in violence as they relate to close mayoral elections, government crackdowns on crime, diversions of drug traffic, and changing alliances and rivalries between cartels. “These factors [may] enable better prediction of where and why drug violence is likely to erupt,” Lawrence agrees.
That’s easier said than done in Mexico, “where official records disappear and local governments try to hide information to minimize violence reports,” Monroy-Hernández says. Espinal acknowledges they have already run up against a troubling dead end: “The data simply disappears from every official record after September 2011.” He sees that as a counterproductive effect of the “silence policy” about drug violence, adopted by current President Enrique Peña Nieto. Until more information comes to light, the current study is “as good as it gets,” Monroy-Hernández says.