Big data and the New Transparency: Measuring and representing police killings

Abstract

Controversies about recent killings by police officers in the United States have prompted widespread questioning about the scale and changes in police use of force. A perceived lack of transparency about the frequency of police killings amplifies concerns that many such killings are unjustified. This commentary considers efforts by journalists and activists to comprise databases that document and measure police violence, particularly in terms of how these endeavors exemplify the New Transparency.

In “The New Transparency: Police violence in the context of ubiquitous surveillance,” I offered a historical view of transparency, a ubiquitous term and a broadly held normative value in contemporary democracies (Brucato, 2015b). The article explains how video cameras—components of surveillance systems or civilian smart phones—are producing new mediated visibility of policing, the result being exemplary of new developments in transparency (Brucato, 2015a; Brucato, 2015b). In this commentary, I apply my concept of the New Transparency to various efforts to collect, analyze, and report on police use of force. As in the prior article, I elucidate the New Transparency by associating John B. Thompson’s (2005) concept of the new visibility with the political concept of transparency, the latter of which is ideologically and practically rooted in the Enlightenment. Further, I explain how efforts to create databases measuring police violence do similar work to that of civilian video recording of police: the work of adapting an enduring tradition of transparency to the contemporary neoliberal context and to new technological affordances.

Developing an early modern concept

Transparency is an Enlightenment-era concept. The term was first used by Jeremy Bentham, who believed that

the persons and objects of that [social] world must be weighed and counted, marked out and identified, subjected to the brightness of the public light, the better to be seen by the public eye. Only then could they be controlled and security made possible….” (Hood, 2006: 8)

Bentham’s maxim was that “the more strictly we are watched, the better we behave” (in Hood, 2006: 9). In this early formulation, an enduring association between visibility and knowledge was wedded to conceptions of security.

Transparency was hoped to reconcile the “vertical inequality” between lawmakers and citizens (Machin, 2012). The normative values for transparency encourage governments and their agents to make themselves visible to their publics via self-disclosure (Birchall, 2011). The powerful establish the legitimacy to hold and wield their power through visibility, exemplifying the Enlightenment project “to make power visible—to illuminate it” (Andersson, 2008: 1). When transparency is lacking, legitimacy declines; in its absence, governments rely on coercion over consent. Without legitimacy, either democracy or governments fail (Brucato, 2015b).

The New Transparency retains early modern qualities but has adapted to the proliferation of information and communication technologies (ICTs). Formal governance by state organizations and informal governance by communities increasingly depend upon ongoing, dynamic collection and analysis of data (Couldry and Powell, 2014). ICTs not only enable, but also are fundamental to how transparency is produced. Paradoxically, transparency is most often mediated. New surveillance technologies extend the capability for visibility to be unidirectional, to transcend temporal or spatial distances, to be preserved on some medium, to be cataloged and sorted, and for its access to be controlled through institutional or technical protocol. Transparency is produced with documents that permit truthful inferences about reality. In doing so, it borrows the Enlightenment distrust in human subjectivity, while adding new technical affordances to produce seemingly neutral, third-party documentations that are treated as self-evident (Harris, 2011).

Transparency now extends beyond voluntary self-disclosure by state actors aspiring to the Enlightenment ideal. It is now produced through “sousveillance,” or “watching from below,” exemplified by participatory media and data production. Birchall (2011) explains that transparency is being reasserted because trust in governments is failing. Civilian voluntarism steps up especially when government surveillance is considered deficient, negligent, or underfunded. Publics are now active as amateur watchdog journalists and as monitors for environmental and health indices (Kimura and Kinchy, 2016). In “The New Transparency” (Brucato, 2015b), I outlined two varieties of such civilian voluntarism:

Civilians volunteer as a kind of redundancy, when they believe there are gaps in extant data or when they want to quantitatively improve or enhance data quality through repetition. This variety may indicate a lack of trust, for instance if official monitoring is considered remiss or if some perceive information is being withheld.

Typical of a neoliberal age of austerity, civilians supplement or replace activity historically under the purview of governments when agencies scale back.

Government self-disclosure of police violence

Police are not immune from the expectation that, as an agency of government, they should publicly disclose their activities. For decades, various attempts have been made to document and report police outcomes (Klinger et al., 2016). Controversial use-of-force incidents are often cause for various publics to declare expectations that investigations be handled transparently and for agencies to improve reporting.

Partly in response to public scrutiny in the wake of the Rodney King beating by officers from the Los Angeles Police Department in 1991, the Violent Crime Control and Law Enforcement Act of 1994 established a federal mandate for the collection and reporting on use of force by police in the United States. In 1995, the National Institute of Justice and the Bureau of Justice Statistics (BJS) established a research agenda to fulfill this mandate. Although the 1994 law requires federal compilation of data on police shootings, there are no requirements that local police departments provide requisite data. Recently, the BJS found that more than a quarter of killings by police are not recorded in the two federal databases (Schweppe, 2015).

Because local police agencies are not required to collect nationally standardized data, these federal statistical programs are hampered by incomplete and inconsistent data (Klinger, 2012). Localized discrepancies have plagued other national use-of-force monitoring. The Centers for Disease Control and Prevention collects data about injuries inflicted by police. As an example of their partiality, only 17 states participated in their National Violence Death Reporting System. The Federal Bureau of Investigation (FBI) maintains records of civilians killed by police, but both their Arrest-Related Deaths database and their Supplementary Homicide Reports exclude about half of known deaths (Krieger et al., 2015; Schweppe, 2015). In 2011, FBI spokesperson William Carr claimed the bureau has no mandate to collect information on police use of force, that it would take an act of congress to change that, and that budgetary and practical factors would prohibit police officers and agencies from collecting the data (Las Vegas Review-Journal, 2011).

Nongovernmental monitoring of police killings

Perceived failures in government self-disclosure have prompted civilian participation in the production of policing’s transparency (Brucato, 2015b). In the wake of controversial deadly force incidents, citizens have demanded increased self-disclosure by government agencies about investigations and discipline, and for more thorough tracking of use of force. Because government monitoring of killings has proved difficult, tracking the far more frequent use of nonlethal force is treated as infeasible.

Despite past resistance to reporting policing killings, in 2015 FBI Director James Comey said the lack of information provided by his agency is “embarrassing and ridiculous.” Comey says: “It is unacceptable that The Washington Post and The Guardian […] are becoming the lead source of information” (Davis and Lowery, 2015). The Washington Post publishes an online database called Fatal Force in downloadable and interactive online formats. Its data comes from news reports, public records, internet databases, and original reporting. The site launched in 2015 and reported 990 killings by police that year in the United States. In 2016 it reported 963 killings.

The Guardian’s site, The Counted, includes an online and downloadable database alongside ongoing investigative in-depth reporting on police violence in the United States. Citing an “unacceptable” lack of self-disclosure by government and police agencies, The Guardian declares “accounting is a prerequisite for an informed public discussion” (The Counted, n.d.). The Counted relies on original reporting based on police reports and witness statements, monitoring of regional news outlets, and drawing from open-source reporting like Fatal Encounters and Killed by Police, discussed below. Additionally, this project is developing a unique process of verifying and fact-checking crowdsourced reports. The Counted puts the number of police killings at 1092 in 2016 and 1146 in 2015. In accounting for per capita racial disparities, The Counted shows 7.13 Blacks per million population were killed by police in 2015, compared with 3.48 per million Latinos, 3.4 per million Native American, and 2.91 per million whites.

Killed By Police started as a Facebook page on 1 May 2013, and posted news articles reporting on any person killed in the United States by a police officer, alongside a running count. Now, Killed By Police also runs a barebones website with a static chronological spreadsheet, each row listing the name of a person killed with a link to a related news article. Where available, the spreadsheet includes the location of the killing and the victim’s age, gender, and race. In 2014, FiveThirtyEight found that 85% of entries in the spreadsheet “were clear-cut police shootings in which the victims were fatally shot by officers acting in the line of duty” and that 8% involved other arrest-related deaths (Fischer-Baum and Johri, 2014). The remaining 7% were accidental deaths caused by officers or intentional killings by officers while off duty.

Hirschfield (2015: 2) discusses what he calls “the most ambitious of [amateur journalist] projects,” Fatal Encounters, which combines data from internet searches, web-based news alerts, public record requests, and crowdsourced information along with “earnest cross-validation.” The site provides a downloadable database of police killings from 2013 to present and includes demographic and geographic information for most incidents. It is maintained by D. Brian Burghart, who says he believes that “in a democracy, citizens should be able to figure out how many people are killed by law enforcement, why they were killed, and whether training and policies can be modified to decrease the number of officer-involved deaths” (Fatal Encounters, n.d.).

Transparency in trouble

The discursive process of defining “missing data” about police violence as a social problem calls upon an enduring value for transparency. Often the references are plain, as when police use-of-force scholar Geoffrey Alpert (2015: 238) explains that “police agencies vary in their transparency concerning OIS [officer-involved shootings].” Transparency consolidates a demand for mechanical objectivity, that objectivity can unproblematically produce knowledge, and that this knowledge enables accountability. Butterfield (2001) implies an enduring distrust of subjectivity and advocacy when claiming the “lack of accurate statistics makes it virtually impossible […] to draw meaningful, big-picture conclusions about deadly encounters between the police and the civilian population.” FBI Director Comey believes “what we need are ideological agnostics who use information to try to solve problems” (Comey, 2015). Alpert (2015: 239) claims that without data “we cannot understand our world and make it better.” Alpert (2015: 239) asks:

How can we address concerns about “use of force,” and how can we address concerns about officer-involved shootings if we do not have a reliable grasp on the demographics and circumstances of those incidents? We simply must improve the way we collect and analyze data to see the true nature of what's happening in all of our communities.

Likewise, Krieger et al. (2015) claim that data is needed because it provides understanding of “the problem” of “law-enforcement-related deaths,” and that this knowledge is a prerequisite to “prevent the problem.” Here, an absence of transparency is evidence that some essential precursor for social action is missing.

Civilian participation in lieu of sufficient government self-disclosure exemplifies contemporary development of the centuries-old value for transparency. These projects share a sense that American police are too frequently using force without sufficient cause. They also share a sense that better methods of collecting, analyzing, and reporting on use-of-force incidents is a necessary early step to fulfill prior to intervention. They too treat the problem of police violence as a knowledge problem. Data is treated as offering some unique access to certain knowledge, without which neither governments nor publics could legitimately act to intervene. Among those in government, academics, journalists, and many activists, police use of force is a social problem to be resolved through better data collection, analysis, and reporting. This discursive maneuver articulates a view of transparency in which databases enable and legitimate social and political action. By implication, this work also functions to communicate that action may be illegitimate without recourse to data.

In “The New Transparency” (Brucato, 2015b), I looked specifically at expectations that video cameras can serve as neutral, objective recorders and reporters of truth. Here, we see that databases are expected to share these qualities. Camera footage turns viewers’ attention to individual occurrences, inviting analysis of “what really happened.” Databases, on the other hand, turn attention toward finding patterns and contrasts. With video, “what really happened” during an individual police killing is an outcome of a certified interpretation of a given video (Brucato, 2015b). The expectation that video will accommodate a lacking condition of transparency encourages viewers to treat cameras as neutral, mechanically objective recorders of truth. In the case of databases, we know that users of data consult them in order to effect change and that this involves not only the presentation of data but also the interpretation of presentations (Taylor et al., 2014). This is a basic problem of the use of information that video and big data share in common. Couldry (2010) explains that regardless of what data are accessed and how they are analyzed, the social processes that steer who can give these data voice, where they are articulated, who can hear them, and what sense can be made are perennial and unsolvable through recourse to information. The necessary step of data manipulation disappears behind the rhetoric of objectivity. With big data, these interpretations will be even more hidden and fixed in automated protocol. As Couldry and Powell (2014) show, the automated aggregations of analytic mechanisms that big data rely upon are closed to human interpretation and review. The calculative logic of analytics forecloses heterogeneous interpretations. Such difference is fundamental to political contention and any partisan advocacy that could influence significant changes in policing. The turn to big data for transparency is intended to avoid any need for interpretation, and thus sets databases up to obscure the interpretive work done by data scientists, data visualizers, reporters, and the recipients of their presentations. Privileging a mythical neutral perspective implies a lesser value of claims made by advocates for people victimized by police. This delegitimation exemplifies the underlying devaluation of the lives of those most likely to be subjected to the violence of police.