WE OWE the notion of a centralized force dedicated to maintenance of the public order to no less a personage than the Sun King himself, Louis XIV of France, who in 1667 issued an edict creating a lieutenancy of police for Paris. Louis evidently intended to replace the medieval city’s crazyquilt of rudimentary law-enforcement measures with something a little more consistent and predictable.
The notion of a standing law-enforcement body eventually made its way across the English Channel to that other early modern megacity London, where in 1749 a judge and author named Henry Fielding convened an office of six men at his Covent Garden magistracy. Authorized to apprehend lawbreakers and deliver them to Fielding for trial, these were the Bow Street Runners, generally regarded as the prototype of the modern professional police force. But they were not yet a patrol force. That distinction belongs to the Metropolitan Police, organized by Sir Robert Peel under Parliamentary charter in 1829, and it is with this formation that law enforcement in the Anglosphere assumed much of its present shape, with all the familiar paraphernalia of precincts and uniforms, ranks and badges, beats and beat officers.
There matters more or less remained until 1950, when the newly appointed chief of police in Los Angeles, William Parker, undertook a reorganization of that city’s force that made use of lessons learned in World War II counterinsurgency operations, and somewhat more distantly from the American occupation of the Philippine Islands. In what Mike Davis has referred to as one in a series of “pathbreaking substitutions of technological capital for patrol manpower,” Chief Parker’s LAPD dispersed its officers across the octopic sprawl of the city in a supple, responsive grid of mobile units dispatched via radio. In these radio cars, the LAPD seemed to be everywhere at once, overmastering an urban topology determined by the internal-combustion engine like no other in history.
And here another long period of conceptual stagnation enveloped law enforcement, at least in the United States. Despite having been tasked over the past two decades with the front-line responsibility for homelessness, school safety, and the management of mental-health crises, and very much despite the occasional vogue for well-intentioned reforms like “community policing,” the archetypal contemporary American police force remains much as Parker envisioned his LAPD. Pleased to regard themselves as a “thin blue line” — the phrasing is another of Parker’s innovations — between polite society and the red-in-tooth-and-claw savagery an unpoliced community would otherwise surely descend to, the United States’s patrol officers now hunker down behind the tenets of a “warrior” mentality, explicitly embracing the virtues of paranoia and instantaneous, overwhelming response to situations (and people) perceived as threats.
Increasingly militarized, in large part through the so-called 1033 program, a federal initiative that transfers war-surplus materiel like automatic weapons and armored vehicles to public-safety departments at zero cost, they maintain a sullen, wary, often hostile distance from the communities they are formally chartered to protect and serve.
All too frequently, these circumstances intersect, with lethal effect. The deaths of Michael Brown, Freddie Gray, Philando Castile, Sandra Bland, and Eric Harris, as well as far too many other human beings whose names will never be more broadly known to the public, indict American policing as it is currently practiced. They stand as evidence that an armed formation comprised of people whose pumped-up bodies, fragile psyches, and febrile politics are all disturbingly reminiscent of Klaus Theweleit’s “armored male” is fundamentally unsuited to maintaining the peace of the 21st-century United States. Although the term feels both bloodless and pitiably inadequate, it is clear that policing as we know it has entered a profound crisis of legitimation.
Two new books propose, however, that this breakdown may also constitute an unusual opening, perhaps even a punctuation of the 75-year-long equilibrium that’s overtaken the praxis of policing. In The Rise of Big Data Policing, the legal scholar Andrew Guthrie Ferguson suggests that a new public-safety paradigm is upon us, enabled by a convergence of separate and distinct technologies that affect where, when, how, and to whom policing is applied. Meanwhile, in The End of Policing, Brooklyn College sociology professor Alex S. Vitale fundamentally questions the institution of policing itself, asking whether it’s remotely appropriate to the absurd variety of tasks we demand of it, and whether it might be supplanted by some more suitable set of arrangements. Should either one of their visions prevail, it’s clear that whatever public-safety formation remains in the aftermath will bear precious little resemblance to the one bequeathed to us by Peel and Parker.
Let’s get the bad news about The Rise of Big Data Policing out of the way right off the bat: despite being crammed with useful detail, and summarizing many of the issues with data-driven policing better than any single volume I know of, in places it’s perilously close to unreadable. Someone hell-bent on making these abstruse issues accessible to a broader readership has clearly advised Ferguson that he’d be best served by rendering crucial parts of his argument in a choppy, would-be Ellrovian vernacular, and these ham-fistedly noirish passages become a tiresome slog. (A sampling: “The night patrol. At night, you stay in the squad car. At night, you wait for the radio run. At night, good guys look like bad guys. In the dark, you wait.”)
Such lapses aside, the case Ferguson makes is inherently interesting, and increasingly urgent. He introduces us to a panoply of technologies that arise at the intersection of pervasive sensing, large-scale data collection, and persistent storage with sophisticated computational analysis, and argues that they amount to nothing less than a revolution in the way the ends of public safety might be pursued.
A police department equipped with these technologies — threat prediction, risk terrain modeling, biometric recognition, and so on — can predict with a high degree of granularity where and when certain kinds of crimes are likely to happen, and even who is likely to commit them. For example, by geotagging all reports of burglary, a department equipped with the necessary analytic software can build up a highly detailed model of the particular blocks likely to be burgled in the future, even specific houses. And by weighing the factors associated with handgun violence — prior arrests, a pattern of association with people who have also been the victims or perpetrators of violence, physiotemporal proximity to previous murders — those individuals in the city with the highest propensity to commit future violent acts of this sort can be identified. These individuals are then notified by the police, or arrested and taken out of circulation if there is any constitutionally viable justification for doing so, and with any luck the crime itself is in this way prevented from ever happening. In both cases, the power to anticipate drives a fundamental change of stance in the way we police, from reactive to preemptive.
At times, the book is fascinating. We learn about “automated suspicion algorithms” capable of recognizing certain stereotyped gestures and interpreting them as various forms of criminal behavior — flagging up the repeated passing of small objects from one person to another, for example, as a reliable indicator that street-corner drug dealing is taking place. We learn that for all their putative objectivity, facial-recognition algorithms authored in Korea, China, and Japan do far better at recognizing East Asian faces than those developed in the West (the converse is also true). We learn about Persistent Surveillance Systems, aircraft-mounted sensors that allow their operators to orbit aloft in invisible, mile-high gyres and siphon up terabytes of imagery from the city below. We learn that in Oakland, California, the police department’s own data revealed that African-American men were four times more likely than white men to be handcuffed in the course of a routine traffic stop — and that 20 percent of officers were responsible for 67 percent of handcuffing incidents. Above all, what we learn is that the process of supplanting traditional methods of policing with data-driven approaches to threat determination, patrol allocation, and preemptive interdiction, and even self-examination, is well advanced.
Ferguson’s overarching conceit is to divide the various possibilities of data collection and analysis into buckets he calls “black,” “blue,” and “bright data,” and he spends far and away the most time considering the first of these: the use of the new technologies in a relatively conventional mode, allowing a police department equipped with them to anticipate and preempt both violent crime and offenses against property.
As Ferguson tells it, this technology-intensive, data-driven mode of policing, with its sheen of neutrality and scientific objectivity, has irresistible appeal to municipalities in the post–Black Lives Matter moment. Local governments chastened by the revelations of discriminatory conduct on the part of police departments nationwide, large and small, may be looking for a way to shroud their decisions with an aura of cool, detached scientism, on the theory that algorithms can’t plausibly be accused of racism.
And there is a genuine promise to all this, one that should be welcomed by anyone who’s ever been beaten or mugged, come home to find their front door pried ajar, or, for that matter, been harassed on the street by armed officers of the state. That promise consists in the idea that the state’s crime-control resources might be concentrated where they properly belong: on the very small proportion of individuals who constitute genuine threats to the commonweal, rather than entire communities of working people. For better or worse, in 2017 the Overton window has moved so far to the right that any notion that the police might address the criminal mayhem suffered by the poor and vulnerable, without at the same time levying an additional burden of oppressive official attention upon their neighborhoods and lives, counts as genuinely progressive.
Why, then, hang the grim label “black” on this set of practices? As Ferguson recognizes, the Achilles’ heel of all such technical deployments is that they take place inside history, and have future trajectories that are to varying degrees determined by that history. By dubbing the ordinary use of predictive technologies in American law enforcement “black data,” he hopes to simultaneously acknowledge two critical facets of the way in which history plays out.
The first is straightforward: due to the fundamental complexity and obscurity of information technology, these tools are “black” in the sense of philosophy’s proverbial black box, i.e., the way in which they work is opaque and largely inexplicable to nonspecialists, and therefore not well suited to existing procedures of governance and democratic accountability. As communities and polities, we just don’t understand these tools well enough to regulate them effectively.
The second sense in which Ferguson means “black” is far more pressing and explosive, albeit at the cost of introducing an unpleasant degree of semantic torque. Put simply, his argument is that the new data-collection and -analysis tools are invariably racist in practice, if not in intent.
This is the central claim of The Rise of Big Data Policing, and Ferguson marshals a great deal of evidence to support it. To start with, predictive systems of the sort described tend to fall afoul of the Heisenbergian certainty that to measure something is to intervene in its fortunes. Police officers, like any of us, can be primed to interpret their perceptions of the world by information they are presented beforehand; knowing that a given location lies within the bounds of a computationally-derived “red box” predisposes the patrol officer to understand the things they see there as evidence of crime, or at least its precursors — and every face that of a potential criminal. (As one skeptical criminal-justice scholar characterized the insinuation, “I go in this box, and everybody’s Michael Brown.”)
Whatever arrests are generated in this way, of course, then feed back into the perception that the location in question was in fact a hotspot of criminal activity, and the cycle begins anew. Meanwhile, identical crimes may be committed elsewhere, in districts not so heavily attended to by the police; they’ll go undetected, never show up in the database, and never cause those locations to be rendered on a map of predicted future hotspots.
A fundamental problem with building predictive models this way is that arrests are not, and historically never have been, fully co-extensive with criminal acts. In fact, just the opposite is true: arrests are discretionary, and — unsurprisingly to anyone who’s paid the slightest attention to the history of American law enforcement — the discretion of the arresting officer is applied in patterns which map directly to racial repression and exclusion. We know that white, black, and Latino/a Americans smoke marijuana at similar rates, for example, but are charged with possession at wildly different ones; as the Oakland data suggests, much the same goes for traffic offenses. So any predictive engine that uses a history of prior involvement with the police as part of its threat model will disproportionately tag nonwhite citizens.
The vendors of predictive technology swear up and down that their systems “aren’t looking for race,” and indeed they may well not be. But as Ferguson reminds us, race correlates strongly with other factors we might choose to measure; train your algorithm to look for those indicators, and you are indeed folding a historical pattern of racial discrimination into your results, whether or not that was ever your intention.
And that pattern is damning. It is invariably poor communities of color that suffer most from crime, and in the United States it has become a commonplace to observe that such communities are simultaneously over- and underpoliced. Black and brown Americans are subjected to heavy manners no majority-white community would tolerate. The litany of humiliations they experience at the hands of the police has perhaps become overfamiliar, but is nonetheless worth enunciating again: they endure endless hassling over what are at worst quality-of-life offenses, specious charges like “failure to comply” or “walking in roadway,” and oppressive stop-and-frisk tactics sharply increasing the chance that a blameless individual will come into contact with the law. Overwhelmingly, it is they who suffer the kinetics of police aggression.
It simply isn’t possible to meaningfully discuss, or understand the implications of, technological change in policing without addressing this state of affairs. As Ferguson makes painfully clear, deployments of technology invariably take place within a social context — and in the United States at this moment, as in any given moment of the past three centuries, that context is one of murderous state violence directed at black and brown bodies.
This is the reason why so many of us are uncomfortable with the new technologies of policing, even where it does seem likely that a proposed innovation might potentially shed some real benefit upon the broader public. In the hands of an organization fully accountable to the citizenry, perhaps, these capabilities might not trigger quite the same premonitory twitching of the peripheral nerves that they do in the here and now. But the police as we know them do not in fact appear to be accountable — and when we’re told they’re moving from a reactive to a proactive posture, too many of us have excellent reason to shudder.
As a useful organizing principle, “black data” is rhetorically unfortunate. Forcing the word “black” to connote both the opacity of data-driven policing and its obscured structural racism is asking too much — too much of language, too much of readers. The term strains beneath the load. The tags Ferguson uses to describe his other buckets — “blue” and “bright” — aren’t so loaded, and as a result wind up faring slightly better.
By “blue data,” Ferguson means the same, highly granular collection and analysis technologies involved in the production of “black data” being turned inward, to track the police themselves and assess them in the performance of their duties. In a proposal sure to be resented by police officers everywhere, if not actively sabotaged by their unions, he calls for the use of “worker surveillance technologies” to track beat cops, rank them by response time, and screen them for early signs of a propensity toward the inappropriate application of force. He extends the hope that a patrol organization that knows the public can see them — and can in fact see the public seeing them — is one bound to exercise a greater degree of respectful care in its dealings with that public.
This curiously Foucauldian notion is, in context, laudable. But if Ferguson believes that transparency underwritten by data analytics must surely entrain better behavior on the part of the police, his optimism is probably misplaced: a large, rigorous recent study suggests that body cameras have had virtually no detectable impact on patrol-officer conduct. (It is in this chapter, as well, that Ferguson inadvertently puts his finger on the deepest problem of all with data-driven decision systems: that whatever inconvenient truths any such system might turn up will simply be disregarded. When an older risk-assessment system, for example, flagged over half of the Charlotte-Mecklenburg, North Carolina, patrol force as having the potential to use excessive force, it was retired rather than being attended to and acted upon. Thus always to the tellers of truth, evidently.)
It is with “bright data,” the final and most speculative of Ferguson’s three modes, that he comes closest to naming the deeper opportunity created by emerging data technologies. If there is one thing that machine-learning-based data-analysis techniques excel at, it is pattern detection and recognition, and Ferguson proposes that we use these capabilities to “reveal hidden problems and remedies” across the wide belt of circumstances we now call upon the police to manage.
From decent housing to the provision of adequate childcare or mental-health resources, these remedies, happily, don’t require the interventions of an armed patrol force. Indeed, once having recognized that the “data-driven ability to identify risk does not necessarily require a policing remedy,” it’s just a short jump to understanding that many, perhaps most of the situations we currently ask them to handle “do not have to be addressed by the police,” and in fact that “a predictive policing approach may be just as effective without direct police involvement.” This is achingly tentative language for what is actually a profoundly radical proposal. Using computational analysis to unpick the latent and unsuspected correlations between seemingly discrete events, in the hope that we can intervene in time to prevent the worst outcomes from arising, undoubtedly substitutes a faith in technology for a commitment to politics. Nonetheless, here we have a first harbinger of what is sure to become a central point of contention in progressive thought in the decades to come: can algorithmic technology achieve the liberation that centuries of organizing, activism, agitation, and even revolution could not?
If Ferguson sees a new policing paradigm aborning, though, he never quite questions the institution of the police itself — and this despite the harms he so accurately diagnoses. What was probably intended to strike an eminently justifiable tone of ambivalence all too often reads instead as an aversion to the conclusions warranted by the facts he’s spent so much effort assembling. Perhaps this is why the measures he calls for in his conclusion are at best tepid and anticlimactic: quality-assurance protocols for third-party data, a set of purchasing guidelines to police administrators contemplating the acquisition of new technologies, perhaps a community-run Police Awareness System capable of identifying abusive cops.
The task of imagining something fundamentally different falls to Alex S. Vitale, in his new book The End of Policing. In successive chapters, he details how the basic post-Parker model of a mobile patrol force has been applied to realms in which it is wildly inapposite, from managing homelessness, emergent mental-health crises, or undocumented immigration, to constraining the practice of democracy as it emerges in the form of protests, strikes, and demonstrations.
Throughout, one of Vitale’s basic points, backed by a solid accumulation of empirical evidence, is that no reform in staffing, training, equipment, or procedure is likely to alter the fundamental hostility of the police force to the policed public, especially when that public is black, brown, and/or poor: “At root, [reformers] fail to appreciate that the basic nature of the law and the police, since its earliest origins, is to be a tool for managing inequality and maintaining the status quo. Police reforms which fail to directly address this reality are doomed to reproduce it.”
This finding is unlikely to faze, say, critical legal studies scholars, let alone anyone with personal or familial exposure to oppressive policing, but it will abrade those inclined toward technocratic centrism — it’s an insight that Ferguson never seems to quite get his head around, for example, despite brushing right up against it in his section on “bright data.” Where others might shy away from the implications of this line of thought, though, Vitale pursues them to their logical conclusion: in context after context, public safety is enhanced, and the public interest is best served, when the police are eliminated from the equation.
His calls for removal of the police from the schools, the decriminalization of most drugs, the legalization of sex work, a greater scope for restorative justice, and an enhanced degree of community involvement in public safety are all welcome. (Presumably even the police themselves wish that someone would relieve them of the responsibility of dealing with endemic homelessness, mental illness, and low-level quality-of-life offenses.) Vitale understands that living up to our own, oft-expressed ideals of justice means rethinking the ways in which we conceptualize and address the various forms of violence endemic in our communities.
He concludes by articulating “a larger vision that questions the basic role of police in society,” though one wishes he hadn’t waited until the last seven pages of a 228-page book to develop it; indeed, there’s something bleakly humorous in learning, on page 223, that following his advice would require “transform[ing] some of the basic economic and political arrangements in our society,” especially since he doesn’t specify any way how we might actually go about doing so.
“Access to decent housing and employment,” for example, certainly seems like something that might ameliorate the conditions that underlie crime, but in a society almost exclusively reliant on the unfettered market to allocate housing (and, what is more, facing the prospect of massive and permanent structural disemployment due to the spread of automation technology), that access feels like it might be a difficult thing to achieve. At the very least, anyone making meaningful progress toward resolving those issues will have achieved a great deal more than simply fixing the police.
The End of Policing’s great strength lies in demonstrating that if the shape of American policing is historical, it is also contingent. We could have made different choices regarding how we set about securing the public against the array of threats that confront it, and — refreshingly, at this moment of general despair — Vitale believes we still can. For all their logical consistency of his arguments, though, their critical weakness is how little they have to do with how power works in the world we actually find ourselves living in. (It reminds me of the charge an online commentator once lodged against the legal idealist Lawrence Lessig: “He knows everything about the way the world should work, and not a damn thing about how it actually does.”) While perhaps unfair to Vitale, and certainly unkind, this nevertheless drills right down to the bitter truth of things: it is, to say the very least, exceedingly difficult at present to imagine an America capable of taking heed of the excellent advice he offers.
The vision enunciated in those final pages is, to be sure, a deeply appealing one, but it would seem to require nothing short of a full-spectrum, Great Society approach to the distribution of goods and resources — and even so, any gains realized by such a program would likely prove to be just as fragile as those won by the original Great Society turned out to be, if not continually defended against those who would undermine them on ideological or pecuniary grounds. For better or worse, we’re compelled to pursue the ends of justice with the United States we have, not the one we wish we had.
Ultimately, however, though Vitale comes closer to getting it than Ferguson, both authors miss something central: the reason why a clearly injurious policing paradigm continues as it does isn’t because we don’t understand that it is broken. It’s because policing as an institution is working as fully intended, i.e., to suppress black, brown, and poor lives, contain their political emergence, channel public funds toward an array of favored private contractors, and perhaps to absorb surplus male aggression. Per the cybernetician Stafford Beer’s dictum that “the purpose of a system is what it does,” we evidently want our uniformed public servants to punish, to kick ass, and to look a certain way while doing it. If we didn’t, we wouldn’t tolerate any of this.
Which is not to say that there is no worth in supposing that things might be different. If at present the task of defending the public and keeping it free from harm falls to the constabulary force we call “the police,” a formation that is by heritage and conception poorly equipped to answer the challenges it now faces, Ferguson’s and Vitale’s books dare to at least sketch the contours of a different way of doing things.
Where The Rise of Big Data Policing and The End of Policing agree is that we face a rupture in our approach to the maintenance of public order. For Ferguson, that rupture consists in the full-spectrum, persistent awareness afforded by the emergent technics of data collection and analysis, as well as the fact that those tools have been placed in the hands of an institution increasingly inclined to conceive itself as apart from, if not above, the people we suppose it to serve; for Vitale, in the utter bankruptcy of traditional policing methods when applied to social problems, the obscene violence which all too often results, and the all but total lack of accountability for it.
If these scholars appear to be in rough consensus that some such discontinuity exists, though, it remains an open question as to what kind of public-safety arrangements we wish to see prevail on the other side of it. The great contribution of these books is in spurring us to realize that policing as it has come down to us from the days of Louis XIV is at or at least very near the end of its useful service life, and that it will be necessary to devise some new institution consecrated to the public defense if we wish for it to retain any popular legitimacy at all. Now comes the difficult, vexed, contested work of imagining what that institution might look like, and how we can design it to operate in a manner commensurate with the rights, obligations, and aspirations enjoyed by citizens of a free, democratic society.