How Three Nerdy Real-Life Superheroes Are Teaming Up To Stop Crime

Take all the crime data in a city, put it through PredPol’s system, and you will get a good idea of the places police should allocate their resources. It’s not quite Minority Report, but the results are impressive, especially for cash-strapped cities where police presence isn’t what it used to be.

How Three Nerdy Real-Life Superheroes Are Teaming Up To Stop Crime

Bring together the former mayor of a laid-back surf town, a criminologist, and a math professor, and what do you get? This isn’t the beginning of a bad joke. Former Santa Cruz, California, mayor Ryan Coonerty, math professor George Mohler, and criminologist Dr. Jeffery Brantingham have banded together to launch a startup–PredPol–that will help police departments around the world more efficiently fight crime.


The seeds of PredPol, which uses cloud-based analytics to tell police departments where they should focus their resources, have been germinating for years. Last summer, Santa Cruz started testing the software as a way to cut down on property crime, car theft, and home burglaries (it can be used for violent crime, but only if there’s enough to create a pattern, like with large-scale gang violence). The results were impressive. Before the program was implemented, property crimes in the city had increased by 20%–and after police started using the program, property crimes decreased 20%. Coonerty admits that there isn’t a direct correlation, but the software made a difference.

Los Angeles has also seen dramatic results since it began testing the software six months ago. Says Coonerty: “The LAPD has 100 analysts and some of the most sophisticated technology of any police department. The algorithm correctly predicted twice as much crime as analysts using best practices [in a double-blind study].” In one LA division (there are 300,000 people in a single division, and 21 divisions in the city), there was a 13% decrease in crime.

The whole PredPol system can be set up over a single afternoon. A piece of plug-in software taps into the police data stream for information on crime, including type, location, and time (all identifying details are omitted). The information is sent to the cloud, where it runs through an algorithm that spits out a number of 500-square-foot hotspots for police to patrol.

Cities use the information in a variety of ways. Santa Cruz has its police check into these hotspots (or “boxes”) between calls. Los Angeles has a larger police force, so cops in the city have to hit the boxes every hour and check in.

It makes sense in a huge city like LA, sure, but wouldn’t smaller cities easily be able to rattle off crime hotspots without needing an algorithm? They can–to a point. “It’s kind of like picking out your favorite breakfast cereal. You might be able to say your favorite is Captain Crunch and then [your second favorite is] Wheaties, but by time you get to pick your 10th favorite breakfast cereal, you’re probably just picking out random names,” explains Mohler. “The same is true picking out crime hotspots. An officer or crime analyst could tell you the top five but not the top 20 accurately.”

There are now a handful of California cities using the software, and 150 police agencies around the world that have already expressed interest. Coonerty says that the decision to launch as a startup was inspired by the obvious need for the technology, which couldn’t be scaled without a real sales team and developers.


“This is a big underserved market. Little cities have maybe one analyst and no technology,” says Coonerty. “These days, budgets are so tight that even the smallest expenditures matter a lot, but the idea with this is that you’re not going to get any more police officers anytime soon.” And if Predictive Policing can help cities keep police costs down, there might be more money for things like parks and youth programs, which are often the first to be cut when times are tough.

Predictive Policing is charging cities based on population, with costs ranging from $25,000 to $250,000 per year for the largest cities. Coonerty contends that the startup doesn’t have any real competition–IBM offers predictive analytics software that can be used for policing, but it’s more expensive. “Most cities can’t even begin to consider that kind of expenditure. Because ours is cloud-based, we think it’s accessible. It doesn’t require any new equipment,” he says.

PredPol is already looking at a new application for the software: Mohler says that the model “works for any kind of random event pattern where an event increases the likelihood of more events.” Predictive Policing is currently working on an algorithm to detect improvised explosive devices. Next steps might include emergency response and traffic predictions.

The startup is about to complete a $1.5 million seed round, and CEO Caleb Baskin believes that there are 14,000 to 25,000 law enforcement agencies in the U.S. that could use the Predictive Policing tool, generating upwards of $50 million each year just from a reasonable market share. “There are so many agencies in need of a tool like this,” he says.

About the author

Ariel Schwartz is a Senior Editor at Co.Exist. She has contributed to SF Weekly, Popular Science, Inhabitat, Greenbiz, NBC Bay Area, GOOD Magazine and more.