What is big data policing?
Big data policing is a popular, yet controversial topic for pracademics working with police organizations, and is therefore of great interest to the Good Data Initiative. The trend of analyzing big data - as seen in industries such as banking, media, and healthcare - has now entered the field of policing. Big data refers to a) the vast amount of digitized data now available and b) the application of artificial intelligence (Joh, 2014). The purpose of incorporating big data into policing practices is to encourage “smart policing”, whereby efficiency is increased despite decreasing budgets. Many police agencies are currently working
with big data, which may be in the form of crime statistics, personal data, environmental data, social media (i.e. Facebook and Twitter feeds), and data from surveillance sources. The application of big data within the field of policing is demonstrated in the following examples:
1. Chicago Police Department: created a “heat list” that uses algorithms to assign individuals a “threat score” that predicts their likelihood of engaging in violent criminal behaviour.
2. New York Police Department: in conjunction with Microsoft, created a “Domain Awareness System” that collects information and indicates connections between various databases, such as CCTVs and automatic license plate readers.
3. Santa Cruz Police Department: uses computer algorithms to both predict areas where future burglaries are likely to occur and identify “crime hotspots.”
How is big data used by police agencies?
Currently, big data is used by police agencies for two main purposes; predictive policing and mass surveillance.
A) Predictive Policing: Police agencies most commonly use big data for the purposes of predictive policing, which applies computer-generated algorithms to historical crime data in order to predict future criminal acts (Joh, 2014). This is sometimes referred to as “crime forecasting”, and accounts for a variety of factors, such as past crimes, seasonal variation, and potential escape routes. The principle behind this technique is simple: police officers patrol areas that have been marked as having high levels of criminal activity, and therefore deter criminals from targeting these locations. For
example, research has demonstrated that 50% of crime in Seattle over a 14-year period occurred in only 4.5% of the city’s street segments (Joh, 2014). Alternatively, predictive policing can involve applying algorithms to various forms of social media. This allows police agencies to identify individuals within a criminal network that are referred to as “central nodes”, as they are highly connected within a criminal organization.
B) Mass Surveillance: While surveillance has always been an essential component of policing work, the advancement of technology and its ability to analyze big data has significantly transformed the nature of surveillance. An example that best illustrates the concept of mass surveillance is the N.Y.P.D.s “Domain Awareness System” (DAS). The DAS is collecting data 24/7 from multiple sources throughout New York City, including police databases, CCTV cameras, automatic license plate readers, and radiation sensors. The DAS is a beneficial tool for crime analysts, as it allows police to access real-time data and automatically indicates links between information gathered from different data sources. For example, the system can search for vehicles based on a physical description, conduct another search on these license plate numbers, and subsequently have access to the vehicle owners’ criminal histories. However, this example only scratches the surface of the potential for mass surveillance. Ferguson (2017) predicts that in the near future, real-time facial-recognition software will connect surveillance camera footage and biometric databases in order to identify individuals with open warrants.
What are some concerns of big data policing?
Although the concept of big data policing is innovative, efficient, and logical, this policing technique has several issues. Specifically, two major concerns have been raised; issues of privacy and objectivity.
A) Privacy concerns: Before the implementation of big data policing, any action committed in a public setting could be subject to police surveillance; however, considering the current and future extent of mass surveillance, will any action remain private from law enforcement? In light of the advanced nature of police surveillance technology, the privacy of all individuals may soon be diminished, regardless of whether or not they are a suspect for a crime. Mass surveillance is becoming a possible and affordable reality, as the cost of storing all audio data for the average individual in one year is an estimated two cents (Joh, 2014). While some argue that big data policing through mass surveillance is a positive policing initiative, the controversy regarding issues of citizen privacy is ongoing.
B) Objectivity concerns: The software and algorithms that are used by police agencies to analyze big data have sparked concerns regarding the objectivity of big data policing. The algorithms behind
predictive policing cannot be entirely objective; they are created by human beings, and are therefore subject to human bias. This point is emphasized by Ferguson (2017), who claims that “data is not blind. Data is us, just reduced to binary code.” For example, if certain neighbourhoods are segregated by race and/or socioeconomic status, then predictive algorithms will likely “forecast crime” in a discriminatory manner. Furthermore, if a big data policing system is analyzing past crime reports, this will result in the over-policing of communities in which predominately poor individuals of colour reside. As Ferguson (2017) points out, this generates a self-fulfilling prophecy, where police only find crime in these areas because this is solely where they patrol for criminal activity.
What is the verdict on big data policing?
Now that we have analyzed the various strengths and concerns related to big data policing, the question remains: should law enforcement continue to practice big data policing methods? As one can assume, the answer is much more complex than a simple yes or no. Every big data policing initiative undertaken by police agencies should undergo rigorous evaluation and revisions before it is integrated into policing practices. This will ensure that these techniques are efficient, do not produce biased results, and align with the principles of evidenced-based policing. Another reason for recommending individual evaluations is because some forms of big data policing are significantly more flawed than others. One particularly concerning example is Chicago Police Department’s “heat list”. The “heat list” is based on an undisclosed algorithm that assigns every individual arrested with a “threat score” ranging from 1-500.The issue lies in the fact that men of colour are disproportionately included on this heat list (Ferguson, 2017). Examples such as this highlight that although big data policing can positively contribute to policing work, there are several issues that must be addressed before these techniques are implemented into police best practices. This blog post was inspired by Andrew Ferguson’s book “The Rise of Big Data Policing”, which explores the strengths and concerns of big data policing in further detail.