In the new world of interconnectivity and data analytics, examining computational data programs provides insights into how Big Data binds together an analytical story. By analyzing crime statistics, such as shootings, robberies, and sexual assaults, data computational tools can place pin points on a map to predict where and when crimes may happen. Although the approach at first seems to capture insignificant data, when large quantities of data are merged together from cell phones, social media, check-ins, and shopping histories, these data sets can paint a story of a criminal mind versus a law-abiding citizen. Big Data might even draw an illustration from a broad perspective and then focus to such precision to create a personal story. If Target can predict pregnancy and Google can predict flu epidemics, then Big Data has potential to prevent and predict crime. However, can analyzing Big Data yield positive outcomes regarding violent crime prevention? In recent years lawmakers have increasingly been turning to Big Data to predict and prevent crime; however, Big Data tools are not necessarily predictors and preventers.
By definition, Big Data refers to collections of data sets that cannot be analyzed by traditional computing processes. It spans data storage; analytic systems; and data capturing devices, such as cameras, microphones, phones, computers, and anything connected through the internet and stored on databases in the cloud or servers. “A popular definition of Big Data is that it involves (at least) three V’s—Volume (the amount of data), Velocity (the speed at which data is being added and processed) and Variety (the fact that data may come from multiple sources using different formats and structures)” (Chan 24). Volume refers to the range of data from terabytes (1024 gigabytes) to petabytes (1024 terabytes). Velocity refers to how quickly data sets are collected, sorted, and stored. Variety refers to social media posts, music, photos, financial transactions, and more. In the early days of Big Data, marketing companies used these sources to predict shopping habits and personal interests in order to target specialized advertisements. Over time data analysts realized that this data can also be used to predict violent crimes such as shootings, robbery, rape, and terrorism. When data gets so big, it can lead to “End of Theory,” an idea defined by Chris Anderson, editor of Wired magazine. When a considerable amount of data collects in data sets, computation from machines will outperform human predictions, leading towards precise predictability, which in turn will help predictive policing outcomes (Chan 24).
According to Chan, there are two primary areas that utilize Big Data to research crime and deviance. The first area, social media, allows criminologists to view real time data as people post subject matter online. The second area uses computer modeling/algorithms to predict police strategies organized with criminal justice decisions (Chan 25).
Aside from friendly interactions, social media helps to organize uprisings, protests, and flash mobs. Every post to Facebook, Instagram, and Twitter creates a trackable timestamp which data programs can analyze. Law enforcement can use such data to track real time crime, because each post generates immediate content. By analyzing and tracking posts from released felons, terrorist watch suspects, general suspicious persons, or even hashtags users; law enforcement use data to predict and prevent crime. In 2011, popular trends, such as violent flash mobs, were dampened by police who analyzed incriminating posts and hashtags. As social media companies increase their privacy policies, and ill-intentioned people send encrypted messages, these communications become harder to target. Social Network Analysis (SNA) allows criminologists to analyze data based on observations of information found on social media. The Department of Homeland Security (DHS) utilizes techniques from the National Security Agency (NSA) to gather and understand key characteristics of terrorist organizations and networks (Hassani 145-146).
Computer modeling and algorithms are the second major area involving Big Data research related to crime prediction and prevention. Even before the invention of Big Data, police used crime patterns to analyze hot spots to assist with predictive policing. Pinning crime data on a map gives police a geographic location to increase patrols. Today’s predictive policing goes beyond mapping; crime data sets now include the following: weather, time of year, time of day, neighborhoods, criminal records, cameras, video recording systems, gunshot monitors, and biometric behavioral collection. Analytics on crime prediction are drawn from computer queries, terrain analysis, and advanced data mining techniques. “Predictive systems can take a variety of forms, ranging from individualized predictions based on individual biometric cues, to profiling based on group attributes gleaned from past behaviors, to more generalized “high crime area” targeting.” (Miller 116). The Transportation Security Administration (TSA) pre-flight check list system represents one example of a program being used by law enforcement to prevent crime. By analyzing passenger data from the International Revenue Service (IRS), law enforcement agencies, frequent flyer programs, and credit scoring agencies, the TSA can prescreen travelers before they even enter the airport. (Miller 116)
The Future Attribute Screening Technology (FAST) project developed by DHS predicts future crime by collecting biometric behavioral data such as cardiovascular signals, pheromones, skin conductivity, eye blink rate, and respiratory patterns by using sensors, videos, and audio recordings. “According to the FAST privacy assessment, “The future time horizon can range from planning an event years in advance to planning to carry out the act immediately after passing through screening. The consequences to the actor (perceived as either positive or negative) can range from none to being temporarily detained to deportation, prison, or death” (Miller 116).
Programs like NSA’s Advanced Question Answering for Intelligence (AQUAINT), and Raytheon’s Rapid Information Overlay Technology (RIOT) promise to not only understand where to find a particular person, but also to understand how a person thinks. By gathering bundles of personal human data, these programs can predict behavior and probability to commit crimes (Miller 117).
The debate surrounding the effectiveness of using this technology to predict and prevent crime focuses on the rate of success. Data scientists question the effectiveness of the NSA eavesdropping programs and their prevention of terrorist attacks. The Obama Administration claimed that the NSA program had prevented 54 presumed terrorist attacks, but under testimony, NSA Director General Keith Alexander only admitted to one case. The “piece of mind” metric coined by James Clapper, Director of National Intelligence, explains that agencies were able to see a database to predict whether there were any threats to New York city following the Boston Marathon bombing. Since no other attacks took place in the days and weeks following the attack, Clapper concluded that the technology was a success. Other predictive systems, such as Suspicious Activity Reporting (SAR), are also questionable. SAR allows law enforcement officials to add people they believe are deemed suspicious to a database. “As of 2010, 161,948 SARs were in the database, of which 103 turned into full investigations leading to 5 arrests and no convictions” (Miller 118-119).
The success of these programs really lies in human behavior and acceptance. As data precision progresses, the systems will more than likely work. They have already shown promise for the future. Furthermore, people have a tendency to trust computation and computer-based systems. Jaron Lanier, computer scientist, has noted these common behaviors and expectations of how quickly humans will trust automated processes and systems (Miller 122).
By understanding how Big Data intersects with Social Media and computational modeling/algorithms, criminologists are finding new approaches to predict and prevent crime. However, aside from deterring violent flash mobs in Chicago, many of the case studies have not proven a positive outcome in predictive analysis. While social media data may tell the story of potential criminals, can potential criminals be arrested before they commit a crime? In today’s market place, giant corporations are spending millions of dollars to invest in predictive policing software. This demand has caused a rise in predictive analysis start-ups which are pushing the boundaries of criminology. In order for criminologists to create new predictive analytic tools, the must not only understand the importance of the technology, but also recognize the weaknesses. Technology and Big Data are not going away—only growing stronger over time.
Works Cited
Chan, Janet, and Lyria Bennett Moses. “Is Big Data Challenging Criminology?” Theoretical Criminology 20.1 (2016): 21-39. Web.
Hassani, Hossein, Xu Huang, Emmanuel S. Silva, and Mansi Ghodsi. “A Review of Data Mining Applications in Crime.” Statistical Analysis and Data Mining: The ASA Data Science Journal 9.3 (2016): 139-54. Web.
Miller, Kevin. “Total Surveillance, Big Data, and Predictive Crime Technology: Privacy’s Perfect Storm.” Journal of Technology Law & Policy 19.1 (2014): 105-46. Web.