Abstract
Brayne’s Predict and Surveil opens with a scene illustrating how tools like Palantir can deliver on the promise to save police precious time and resources, in this case facilitating the arrest of a suspect involved in the aftermath of a gang-related murder. The manuscript, however, goes on to complicate this notion and propose that we should be critical of the actual functions and impacts of these technologies, as they make police decision making more opaque and exacerbate inequalities. Big data technologies displace, but do not eliminate, bias and discretion in policing. Predict and Surveil offers a critical account of the rising popularity of tools like Compstat, PredPol, HunchLab, or Palantir that draw from public and private data sources to answer queries, flag patterns, and predict which people, networks, places, or ‘times’ might be of interest for policing.
Privileged, powerful groups like police, lawyers, and judges remain relatively understudied, partly because access can be tough for researchers to negotiate. Difficult as it may be, independent research that ‘studies up’ and interrogates privileged actors is vital. Brayne’s tour de force ethnography contributes to a rich tradition focusing on organizational and institutional sites to understand how professionals adopt or resist policies, practices, or technologies. I was impressed and dismayed by the dystopic extent of state surveillance facilitated via traffic cameras, consumer tracking, and social media. I appreciate the book for what it teaches us about how private markets exploit the digital traces we leave behind, and the need police agencies have to invest in a veneer of ‘objective’, fair, and reliable decision making.
Brayne’s project is impressive in part because of the access she negotiated and the five-and-a-half years she spent in the field. She offers us multiple vantage points from which to appreciate police culture; taking part in helicopter surveillance missions, spending time with officers at crime scenes, in squad cars, and at the bar, attending surveillance industry conferences, and visiting tech teams and research and intelligence agencies. This relatively short manuscript is engaging, clear, illustrated thoughtfully, and rich with field notes and interview excerpts that bring key arguments to life. Introducing police jargon and describing the frigid environment of the squad cars or the frenzy of vendor conferences for example, the vignettes help us appreciate the context in which this data was collected.
The first part of the book reviews how public police have long borrowed from military logics and tools, and how they now also collaborate with private companies to build data-driven prediction tools that quantify and professionalize police work. There are some similarities with public/private partnerships that have also led to the proliferation of risk assessment tools in corrections or social work. Actors in many fields are facing similar managerial pressures to track, analyse, and report performance to gain legitimacy, while simultaneously being asked to do more with less. With Brayne’s work, we learn that pressures to address systemic racism and inequality in policing are used to justify the use of data-based decision making. Terrorist attacks were also used to justify the creation of surveillance hubs designed for counterterrorism, which now also get used for municipal policing—one of many striking examples of function creep. This provides us the foundation we need to understand how public police have come to depend on big data and technologies developed by the military and private sector. This section helps us appreciate the appeal of new tools police use in surveillance and prediction, and how that relates to ever-growing institutional and social pressures to be efficient, accurate, and fair.
One part of the book that stood out to me shifts the focus away from officers watching the public, to tools used by police to surveil other police. Brayne shows that while there is potential for these tools to improve policing of the police, many factors make this unrealistic—including union-supported resistance and limited adoption of existing oversight mechanisms. Ironically, police officers do not appreciate surveillance and predictions tools ‘turned on them’ by their own organization. When the watchers become watched, they complain about function creep, injustice, privacy concerns, and threats to occupational identity. This response is not uniform, with sworn officers, civilian analysts, and management perceiving these tensions differently. This part of the book makes the point that surveillance is ‘relational’ and experienced negatively by those on the receiving end, even if they are relatively privileged and powerful.
Brayne’s engagement with sociological questions is ever present, weaving through chapters and carrying the conclusion. Impressive in its breadth, the manuscript engages with notions of social change, social capital, social movements, social control, socially impactful events, the discretion of (biased) social actors, and the social functions of institutions. It also presents relevant social-historical context, to situate and explain social injustice, systemic racism, perpetuation of inequality, and the social impacts related to policing. To some extent, the conclusion left me wanting more, as I remained intrigued by the notion that ‘big data is fundamentally social’ and that social patterning informs how it is used (p. 137). Reviewing the claims advanced throughout the book, we can appreciate why contemporary policing practices should be considered from a sociological perspective, as they are shaped by while also shaping the social. Brayne highlights that:
Society has become increasingly digitized, automated, and organized according to algorithmic analysis and prediction. We now live in a ‘big data environment’ (p. 3), where consumer and citizen tracking has been normalized, and private companies play a large role in how institutions and professionals operate. Socially impactful events like 9/11 have helped justify the creation and funding of data hubs, which facilitate unprecedented information sharing. Social movements focused on police violence and racial profiling, cries for justice, and demands for police reform are also included in narratives that are said to legitimize the development of big data tools. Big data is a form of social capital, which helps professional organizations respond to calls to improve services. It is a resource that helps police leaders manage positions in their field, by bolstering claims of objectivity and performance, by relying on the appeal of quantification and computational methods. Big data is ‘fundamentally social’ (biased and subjective), since it is embedded in power structures, shaped by organization priorities and reimagined according to actors’ discretion at every point and in every way (p. 4). Indeed, people decide when and how data should be collected, how it should be stored, how consent should (or should not be) obtained, who should access or share the data, for how long, and for what purpose is should be used, and as evidence of what. Practices informed by big data tools worsen social inequality and continue to drive discriminatory and racist practices and outcomes. Data-driven predictions can drive up police contact in ‘hot’ areas, and with suspected individuals triggering feedback loops that perpetuate cycles of suspicion and controls. Suspected individuals are dragged into the system in ways that are often invisible, yet can be life changing. This goes well beyond the reach of a traditional criminal record, punishing (presumably innocent) people. The law is not sufficiently sociological, and is therefore detached from empirical reality. This leads to judicial mismanagement of important questions like prediction and ‘police suspicion’ (p. 121), privacy and data storage issues, or the false assumption that data tools are race neutral (p. 132). Globally, these contemporary policing practices serve a social function, empowering the institution with professional knowledges that support the ongoing surveillance and control of people deemed to be (potentially) problematic. Socially constructed platforms like Palantir are inherently malleable and locally adapted, driven by human decisions. Brayne argues convincingly that the source of problems (see point 5) is also the source of promise with these technologies. Private market ‘trade secrets’ obscure much of the decision making that informs these tools, yet she proposes we can and should find ways to avoid ‘tech washing’, to curtail net-widening and secondary surveillance, and to otherwise temper consequences big-data-driven police tools have on people and on society more broadly (e.g. better remediation of errors).
I recommend this book to anyone interested in contemporary public police work, big data tools, punishment, surveillance, social control, private–public partnerships, organizations, or risk management. All will appreciate, as I did, Brayne’s engaging storytelling, compelling evidence, and convincing arguments that push us to recognize how predictive logics have transformed the landscape of public policing.