Algorithms in practice: Comparing web journalism and criminal justice

By Angèle Christin. 2017

307 citations recorded by the source corpus.

Big Data evangelists often argue that algorithms make decision-making more informed and objective—a promise hotly contested by critics of these technologies. Yet, to date, most of the debate has focused on the instruments themselves, rather than on how they are used. This article addresses this lack by examining the actual practices surrounding algorithmic technologies. Specifically, drawing on multi-sited ethnographic data, I compare how algorithms are used and interpreted in two institutional contexts with markedly different characteristics: web journalism and criminal justice. I find that there are surprising similarities in how web journalists and legal professionals use algorithms in their work. In both cases, I document a gap between the intended and actual effects of algorithms—a process I analyze as “decoupling.” Second, I identify a gamut of buffering strategies used by both web journalists and legal professionals to minimize the impact of algorithms in their daily work. Those include foot-dragging, gaming, and open critique. Of course, these similarities do not exhaust the differences between the two cases, which are explored in the discussion section. I conclude with a call for further ethnographic work on algorithms in practice as an important empirical check against the dominant rhetoric of algorithmic power.

Subfields: Communication, Safety Research, Sociology and Political Science

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Communication

In Safety Research

In Sociology and Political Science

Citations in the corpus

Fairness and Abstraction in Sociotechnical Systems (2019, background): “Second, studies of monitoring, evaluation, and measuring technologies demonstrate that they produce reactivity behaviors , thus altering the very social context that the original design was meant to support [15, 22, 25].”
Closing the AI accountability gap: defining an end-to-end framework for internal algorithmic auditing (2020, background): “Large gaps between the intended and actual uses of algorithms have been found in contexts such as criminal justice and web journalism [12].”
Human Trust in Artificial Intelligence: Review of Empirical Research (2020, background): “…facilitating positive initial trust among humans by encouraging safe and mutually acceptable conversational content (Isbister et al., 2000) and increasing mutual trust by facilitating more fair offers in negotiations and decision-making (de Melo, Marsella, & Gratch, 2016, 2017; Lee & Baykal, 2017).”
Algorithmic management in a work context (2021, background): “One key reason is the high-cost and uncertain returns of algorithmic systems, which usually have to be sourced from a third-party vendor (Christin, 2017).”
The Principles and Limits of Algorithm-in-the-Loop Decision Making (2019, background): “There are several indications that our results accord with real-world outcomes, however: judges su er from many of same cognitive illusions as other people [38], are skeptical about the bene ts of algorithms [10, 11], and exhibit disparate interactions when using risk assessments [1, 16].”
ORES (2019, background): “Many debates about COMPAS and related systems have placed less attention on the models as an element of a broader socio-technical software systemwithin organizations, although there are notable exceptions like Christin’s ethnographic work on algorithmic decision making in criminal justice, who raise”
Algorithms at Work: The New Contested Terrain of Control (2020, background): “Yet, although a labor process perspective draws attention to how algorithmic control can result in negative outcomes for workers, studies have also shown that there is variation in worker outcomes across organizations and individuals (Christin, 2017; Griesbach et al., 2019; Lehdonvirta, 2018).”
Simple rules to guide expert classifications (2020, background): “In the criminal justice setting, past work indeed suggests that judges may not apply the recommendations of risk assessments in a consistent manner (Christin, 2017; DeMichele et al., 2018).”
The Society of Algorithms (2021, background): “…through US networks but illegal for citizens by the Fourth Amendment of the US Constitution, which prohibits unreasonable search and seizure.9 With the help of data-hungry algorithms, citizenship was reframed from a binary, institutional category to a continuum of behavior (Cheney-Lippold 2016).”
This Thing Called Fairness (2019, methodology): “Ethnographic research by Angele Christin suggests that like other tools that are introduced into the workplace, the introduction of risk recidivism tools has met with resistance [21].”
Human Decision Making with Machine Assistance (2019, background): “Christin [15] studies the use of algorithms in the criminal justice and web journalism domains, finding that the intended and actual use of algorithms often differ, and that practitioners often develop strategies for minimizing the impact of algorithmic input in their daily work.”
A systematic review of algorithm aversion in augmented decision making (2019, background): “This means that the successful implementation of algorithmic decision making requires motivating, or incentivizing, human decision makers to utilize algorithmic aids in order to balance the costs of effort with the benefits of decision performance (Christin, 2017).”