Algorithms As Prosecutors: Lowering Rearrest Rates Without Disparate Impacts and Identifying Defendant Characteristics ‘noisy’ to Human Decision-makers
Daniel Chen (TSE / IAST / Université Toulouse )
Elliott Ash (U. of Warwick)
Daniel Amaranto (NYU)
Lisa Ren (NYU)
Roper Roper (NYU)

Abstract : This paper investigates how machine learning might bring clarity to human decisions made during the criminal justice process. Our data comes from all cases at the New Orleans District Attorney’s office for the years 1988-1999. We exploit random assignment of prosecutors, prosecutorial discretion, and heterogeneity across prosecutors in charge rates to compare prediction models to judicial decision makers. Our model of defendant rearrest, trained using de- fendant and offense characteristics, selects higher-risk individuals to prosecute than its human counterparts did. In particular: given a set charge rate, our model would reduce rearrest rates by five to nine percentage points. This model could have several important policy implications: it might identify defendant characteristics that are particularly ‘noisy’ to prosecutors; it could suggest ways of alleviating criminal caseloads without increasing crime rates; and it might provide important insights into how a prosecutor’s background relates to the quality and nature of their charging decisions.

Missing Rich Offenders: Traffic Accidents and the Impartiality of Justice
Madina Kurmangaliyeva (European University Institute)

Abstract : This paper estimates the effect that wealth and power have on criminal justice outcomes by exploiting the random matching of drivers to pedestrians in traffic accidents. If justice is impartial, we should observe the same share of rich offenders both for poor and rich victims, conditional on location and time. Rich victims act as a control group to estimate the proportion of missing rich offenders whose victims are less powerful. I use this estimation approach on data from Russia, and find that its justice system is not impartial. The same approach can be applied not only to other countries but also to other characteristics that should be irrelevant to judicial outcomes in an impartial legal system, such as race and gender.

Predictors of Organized Crime and Subversion: a Machine Learning Approach
Patricia Prüfer (CentERdata, Tilburg University)
Pradeep Kumar (CentERdata, Tilburg University)

Abstract : The share of offences that can be related to organized crime and subversive activities is increasing. Organized crime engages in systematic violation of the law with serious effects on society and is able to cover these violations in a very effective way. The same is true for subversion, which can be seen as “the paralysis of the (regional) society”. Notwithstanding the serious negative effects of these activities, it is difficult to combat them as they are performed illegally and ‘under the shadow of the law’. In this paper, we disentangle patterns of organized crime and subversion on Dutch industrial areas. Based on patterns and best predictors, we develop a method with which policy makers can judge the level of criminal and subversive activities taking place in a certain area. This method provides an instrument, think of a ‘sliding scale’ or an ‘organized crime hot spot’ providing insights in the state of a certain industrial area. Besides data we scrape from the Internet (and eventually the Darknet), we use all types of available data and information related to different aspects of organized crime and subversion. For example information from a suspicious person or firm; all sorts of files, for example on ‘clients’; annual reports and balance sheets; tax information; and registries from the Chamber of Commerce. In theory, also social media data and the browse/search behavior of a suspicious person or firm can be analyzed with respect to interesting trends and patterns. To mine, cluster, and analyze these structured as well as unstructured data sources, we mainly use techniques related to natural language processing and machine learning techniques. Based on all the patterns found, we can derive the best predictors for organized crime and subversion on Dutch industrial areas. This provides a method and an instrument with which policy makers can judge the level of criminal and subversive activities that take place in a certain area.