How Big Data is Cracking Down in Impoverish Neighborhoods

Victorina Joy Santos
2 min readMar 21, 2021

In this blog, we will be focusing on Chapter 5 of Cathy O’Neil’s work, “Weapons of Math Destruction” (WMD). In Chapter 1, O’Neil brought up the recidivism model used as the sentencing guidelines for each inmate. In Chapter 5, another WMD that is involved with the justice system is addressed. This WMD was a predictive program created by PredPol that measures the likelihood of crimes occurring in certain areas of a town/city. Low-income areas do not have enough resources to support a large police force. Therefore, predictive models such as PredPol’s are used to maximize a small police force’s utility.

The model is efficient; it lowered the rate of burglaries by 23 percent in Reading, Pennsylvania. Unfortunately, the model is unfair because it mainly targeted areas that were stricken with poverty. The police force focused on catching individuals that are committing petty crimes within these impoverished area. Meanwhile, they overlooked crimes committed by middle/high-class individuals because crime is less likely to happen in wealthy neighborhoods. O’Neil emphasized the need for equality within the justice system. Thus, it is unfair that an innocent person born in a poor neighborhood faces heavier surveillance while a criminal surrounded by law-abiding citizens does not.

While PredPol’s predictive model was efficient, it was also unfair. Efficiency is easy to achieve through data, but the concept of fairness is difficult to define using data; therefore, WMDs cannot measure fairness. To import fairness, everyone in society should face the same surveillance level from the police force, but will people agree with that idea? Those who disagree are more likely to be individuals living within good neighborhoods. When models are created, if they are efficient, their creators will believe that they are working correctly and overlook other troubling aspects like the lack of equality. To prevent inequality in WMDs, creators of such models should base the success of a model not only on its efficiency but also on equality in the justice system.

--

--

Victorina Joy Santos

Undergraduate majoring in Computer Science and Minoring in Mathematics