The Center for the Analysis and Modeling of Security (CEAMOS), active since 2007, specializes in developing models aimed at helping Chile's national police force and its prison guard unit, known as the Gendarmerie, develop more effective crime fighting strategies. 

"There aren't many mathematicians or engineers who are working on crime issues," CEAMOS engineer Fernando Ordoñez told InSight Crime. "It tends to be something dealt with in public policy, anthropology, sociology, but you don't see a lot of engineers [in Chile] working on the problem. In that sense, we're pretty unique." 

The center, made up of a team of 10 researchers -- the majority of them engineers and mathematicians -- developed one such model that showed the national police force, the Carabiniers (Carabineros), that there is a better way to deploy crime-fighting resources across the Chilean capital, Santiago. Using Carabinier statistics on the number and type of crime in certain neighborhoods -- as well as data on the geography of city neighborhoods and census data -- CEAMOS researchers were able to map crime "hotspots" in Santiago. 

The Carabiniers' current law enforcement strategy, known as the Cuadrantes Plan, involves dividing up major cities like Santiago into smaller sections monitored by police patrols. The CEAMOS model showed that there are better ways "cuadrantes," or police beats, could be designed, in order to more efficiently target the crime-prone zones.

As Ordoñez explained to InSight Crime, the cuadrantes patrolled by beat officers are often designed around neighborhood districts or roads. But as Ordoñez points out, "Crime doesn't respect these frontiers. The fact that the cuadrantes need to follow these political and neighborhood borders, it doesn't make sense." Via its mathematical model, CEAMOS aims to prove that police resources could be much better used if the cuadrantes are not developed in such a way.

According to CEAMOS director Raul Manasevich, currently there are no other comparable research centers in Latin America. Outside the region, the academic institutions best known for blending criminology with mathematical modeling include the University of California-Los Angeles (UCLA) and the University College of London. One crime-predicting model developed by UCLA was able to demonstrate which of Los Angeles' crime hotspots would most likely disappear if hit hard by police, and which ones would simply shift to other parts of the city. Research teams at two Spanish universities in Madrid have also done similar work. 

However, even in Latin American countries badly affected by dramatic increases in crime and violence, the use of mathematical modeling to aid law enforcement remains rare. In that sense, CEAMOS is something like a pioneer in the region, with projects that could yet be applied in other nations struggling to manage insecurity. 

The research center has also worked with the national prison guard, the Chilean Gendarmerie, on developing models predicting which inmates are most likely to become repeat offenders. CEAMOS also helped design a more efficient system for assigning and tracking prison guard shifts, as guards would frequently skip work or else be forced to work multiple shifts over short periods of time. 

 Another project involved analyzing police emergency calls, 80 percent of which were found to be pranks. Researchers found that over half of those who called the 133 emergency number hung up the phone before being connected with a call-taker, because they had to wait for so long. Such findings formed the basis of a CEAMOS researcher's recommendations that authorities should do a better job at fielding non-emergency calls to the general information number, 139, in order to free up the phone lines. 

InSight Crime Analysis

While Chile suffers some of the same security ills seen elsewhere in Latin America -- human trafficking, street gangs, and drug trafficking -- it remains one of the safest countries in the region. Nevertheless, the crime modeling practiced by CEAMOS could easily be applied to criminal phenomena elsewhere. Police in Colombia's major cities, including Bogota, follow their own Cuadrantes Plan, while Mexico has applied a similar model to track crime and accidents along its highways. Authorities in Colombia's second-largest city report that 70 percent of all emergency calls are prank or dropped calls, while predicting recidivism remains a challenge for prison systems from Central America to the US. 

There is also the possibility of developing mathematical models that predict which areas along a national frontier are most prone to crime or illegal crossings, research that could have significant implications for nations struggling to impose ambitious border security plans, such as Brazil. CEAMOS is currently developing one such proposal for Chile's Carabiniers, which would involve designing a model that predicts which border crossings are most likely to be used by undocumented migrants or smugglers. 

The possibility that mathematical models could help "predict" crime has echoes of a science fiction novel. But as Ordoñez says, such models "are no silver bullet."

"If these same models use bad data, they're going to produce bad results," he said.