Idealizations and ideal policing

1. Ways of life and deep inequalities

The basic structure of society, determined by our major institutions, is significant because it has an enormous impact on the kinds of lives that can be lived within society. The distribution of rights and duties is central to the distribution of what is valuable within a society. As Rawls puts it,

[T]he major institutions define men’s rights and duties and influence their life prospects, what they can expect to be and how well they can hope to do. … In this way the institutions of society favor certain starting places over others. These are especially deep inequalities. Not only are they pervasive, but they affect men’s initial chances in life; yet they cannot possibly be justified by an appeal to the notions of merit or desert. (1999, 6–7)

Our major institutions, then, cannot be justified in the standard terms of interpersonal justice. There are three criteria for an institution to count as major: they must (i) determine the kinds of lives we can live and (ii) produce deep inequalities that (iii) are not justified by merit or desert. In determining principles of justice that select major institutions, Rawls engages in a full compliance idealization wherein the members of society are basically like us but are disposed to conditionally follow all of the rules. The domain of partial compliance theory drops that idealization. In characterizing that domain, Rawls lists theories of punishment and just war (1999, 8).

One thing that is easily missed is that these criteria apply to our institutions of law enforcement and not just the courts. To see this, let us consider the major institution criteria in turn. For now, I shall drop the full compliance idealization. Later, I argue that applying that idealization does not deprive police agencies of their status as a major institution or exclude them from the basic structure of society.

First, modes of policing largely determine the kinds of lives we can live. Being arrested and incarcerated drastically changes one’s life. The police play a major role in how public spaces can be used within in a community. Whether a park is used for recreation or off-book entrepreneurial activity is determined, in part, by policing (Venkatesh 2006, 200). Some departments are more deferential to those inhabiting the spaces than others, and the same is true for precincts or districts within a department. Whether people can gather for conversation on a sidewalk, let their get-together spill out of their house and onto public spaces, busk, sit on a stoop just to people watch, jog through a park at night, or count on quick responses to crimes or disturbances in progress all play a role in determining what their lives look like.

The automobile is a powerful illustration. Not only a symbol and means of freedom, the car is also a way of expressing oneself. In her history of traffic enforcement, legal historian Sarah Seo details a CBS documentary (Black on Black) that chronicles the importance of a car to express individuality in predominantly Black South Central Los Angeles. The “lowrider” wasn’t just a means of self-expression in LA; it was also a target for law enforcement (Seo 2019, 215). Similarly, lifestyle decisions such as having dreadlocks or wearing “lots of gold jewelry” make up the drug trafficker profile created by the Drug Enforcement Administration in Operation Pipeline (Seo 2019, 256).⁴ In one manual for effective criminal patrol, the author describes bumper stickers with Grateful Dead or religious imagery both as possible evidence of drug trafficking (Remsberg 1995, 55). Decisions about one’s car and one’s appearance are costly, and their costs are determined in part by policing decisions.

Second, law enforcement produces significant inequalities. For instance, interactions with police officers significantly affect individual well-being. Interactions with police can produce trauma and anxiety. More frequent, intrusive, or aggressive interactions, unsurprisingly, result in higher levels of trauma and anxiety (Geller et al. 2014). Because most Terry stops (“stops and frisks,” named for the Supreme Court case Terry v Ohio that legitimized them) often do not result in an arrest, these negative and unequal effects can accrue to the innocent (Apel 2016).

As “filters” on the input of the court and prison system, the nature of policing has a serious influence on the nature of punishment. Because the criminal legal system moves slowly, nearly half a million people are now held for pre-trial detention—that is, detention before they’ve been found guilty, and most people in jails are there pre-trial (Sawyer and Wagner 2019). Those who are unable to post bail risk losing their jobs, custody of their children, the ability to care for their pets, and so on. Incarceration, pre- or post-trial, is physically, emotionally, and financially painful (Henrichson et al. 2017). The leading cause of death in jails is suicide; in prisons, 6 percent of deaths are suicides—double the homicide rate (Bureau of Justice Statistics 2015). Whether people bear these costs is partly determined by police decisions.

The costs of punishment do not end with incarceration. Reentering society from prison is incredibly difficult. Individuals who undergo pretrial detention are probably less likely to find future employment (Dobbie, Goldin, and Yang 2018). Those with convictions are probably less likely to be hired (Agan and Starr 2018). They are also less likely to be admitted to higher education (Stewart and Uggen 2020).

Furthermore, we cannot justify (all of) the deep inequalities that policing can generate by appealing to merit or desert (criterion iii). If full enforcement were a realistic option, we might be able to. But as a practical matter of fact, many lawbreakers will go unnoticed, not apprehended, not charged, or not convicted. Whether an individual is arrested can be determined by to whom, what, and where police direct their attention. How departments manage their resources, and how individual officers manage their discretion, means that a non-negligible amount of the inequalities that result from our law enforcement institutions are not justified by appeals to criminal behavior. Resource constraints entail that we will always have to make difficult decisions about allocating police resources. Those decisions inevitably bear on the legitimacy of police agencies and institutions of law enforcement.

The police do not act alone, but they do determine who enters the system and who does not. The nature of policing in a society partly determines the kinds of lives that can be lived (criterion i) and the kinds of inequalities that arise within that society (criterion ii) in a way that cannot be entirely justified by desert or merit (criterion iii). More on this last point later.

2. Interactions with other major institutions

Police agencies on their own meet plausible criteria for major institutions in the basic structure of society. But for those who are skeptical, it is worth pausing to reflect on how policing decisions interact with other paradigmatic major institutions. Again, these include political decision-making institutions, economic institutions, and families.

Interactions with the police reduce political engagement, and reducing political engagement is a way of reducing political power. This most clearly occurs via felony disenfranchisement, a practice that has probably been instrumental in the results of presidential and congressional elections (Uggen and Manza 2002). It also has the tendency to reduce substantially the likelihood of voting even without formal disenfranchisement (Weaver and Lerman 2010, 828). One of the most fundamental motivations of sensible political philosophy is to justify political power to those subjected to it. That a common exercise of political power has the tendency to reduce the justification of political power is especially troubling.

As we saw above, policing and law enforcement decisions have the effect of favoring some starting points over others, economically speaking. In addition to the effect that incarceration has on employment and admission to higher education, over-policing in schools transforms detention-able infractions into arrest-able infractions, depriving students of education and in turn the credentials needed for economic opportunities (Rocque and Paternoster 2011; Gottfredson et al. 2020). The likelihood of commercial investment in a neighborhood is partly determined by the security of foot traffic in an area. Ineffective or inadequate policing can, to the extent that policing can reduce crime but fails to do so, affect the kinds of economic opportunities available in a community (Rosenthal and Ross 2010).

Other well-known costs of policing accrue to families. Under-policing can make it difficult for families to spend their time together in public spaces, as can over-policing. Spouses and children bear tremendous costs associated with incarceration. Children with incarcerated parents have more mental health and behavioral problems (Geller et al. 2009). Certain kinds of law enforcement, such as drug prohibition, create rather than stymie violence (Resignato 2000). In these cases, innocent people can suffer serious consequences. This is the result, at least in part, of the kind of policing a community receives.

The distinctive features of major institutions apply to policing directly, as I argued in section 1, and apply indirectly, as argued in this section. In what follows, I argue that the third criterion is met even in highly idealized models of society.

3. Full compliance, traffic enforcement, and additional idealizations

The controversial claim required to show that policing should count as a major institution is that the inequalities (partly) caused by policing cannot be justified by merit or desert. The claim is not that merit and desert have no relevance here, for they clearly do. The claim is that a portion of the resulting inequality is independent of merit or desert. Undeserved inequalities could be eliminated only by perfect, full enforcement or by eliminating the need for police entirely (by perfect, full compliance). Yet, police are constrained by resources and a necessarily indeterminate criminal code. Additionally, full enforcement or compliance would be desirable only if the broader criminal justice system were infallible. In police scholarship, full enforcement is described as a persistent myth (Berg 1999, 262). The full compliance move, then, would successfully block an active enforcement agency from the basic structure only with methodologically and substantively objectionable idealizations. The cost of applying them is to render the resulting model much less philosophically interesting or useful.

Consider first that some laws we ask the police to enforce are unjust. In a society characterized by full compliance with all laws, we would have to eliminate these laws for the society to look normatively desirable. If we want to engage in that kind of idealization, then we must pause to reflect on what additional idealizations would be required. We would also need to idealize the democratic and legislative processes. This is because non-ideal versions of these processes yield laws that we lack reasons of justice for wanting people to obey and police to enforce. Such an idealization requires idealizing the electorate such that they lack these unjust political preferences. It would also require idealizing both the legislators and the legislative process. We would have to idealize away problems of regulatory capture and the political incentives that make decarceration so difficult to achieve.

So far, so good, we might think. This does not look all that objectionable to those sympathetic to the full compliance idealization. But it is easy to miss all of the things that go into these kinds of idealizations. For one, it is not obvious which policies best achieve their goals; well-intentioned policies often have unexpected outcomes. We can accidentally end up with bad laws with which it is not ideal to comply. To avoid this, we have to imagine legislators as omniscient, or much closer to it than they are now. Furthermore, even the most extravagant idealizations in political philosophy do not imagine legislators as already possessing the wealth of information they’ll need. Information is not free. Given resource constraints, acquiring that knowledge will invariably take resources away from other ends. The idealizations that go into making the laws worthy of full compliance are much more expansive than it might otherwise seem.

But it is not just the legislators (and the democratic process that selects them) and the legislation they produce that must be idealized. Citizens and the enforcement environment will need to be idealized as well. For one simple illustration, consider that traffic enforcement necessitates partial, discretionary policing even in highly idealized models of society. Given how important transportation is, this illustration is simple but not inconsequential. In the early 1900s, traffic problems pre-dated traffic divisions in police agencies. Traffic codes and traffic law enforcement developed alongside the adoption of the automobile. While early traffic law was based in tort law, that quickly transitioned to an expanding criminal law (Seo 2019). One of the unavoidable but unfortunate features of traffic law is its complexity. Because there are so many ways for driving to go wrong, the legal code—in an attempt to be fully determinate—is unwieldy and complex. If you drive, you violate traffic laws.

Crucially, traffic violations arise even when motorists intend to follow the law. This is because, for one, perfect driving is difficult. It is also because traffic infrastructure often encourages illegal driving even in courteous and conscientious motorists. If lanes are wide and there is a protected medium between traffic lanes, motorists will naturally drive more quickly; if the speed limit is set lower than people naturally travel on such roads, the infrastructure encourages illegal driving. If we want to imagine that a full compliance idealization is possible, we will have to make motorists nearly omnipotent (regarding driving). If we want to imagine that a full compliance idealization is desirable, we will also have to idealize all traffic infrastructure. Again, any money spent on better traffic infrastructure comes at a cost to funding for education, healthcare, and so on. The full compliance idealization, then, requires idealizing away various faults in legislator knowledge, motorist ability, and the resource scarcity that normally prevents full compliance from being possible and just.

The automobile played a large part in the rethinking of how to apply the Fourth Amendment. It highlights, then, the difficulties of a fully determinate legal code but also the ineliminable vagueness of much of the law. In his 1974 Oliver Wendell Holmes Lectures, legal theorist Anthony G. Amsterdam makes the point that there are no bright lines that can regulate police conduct. Here is one version of this point:

As applied to law enforcement activities, the terms “searches,” “seizures,” “persons,” “houses,” “papers,” and “effects” could not be more capacious or less enlightening. The plain meaning of the English language would surely not be affronted if every police activity that involves seeking out crime or evidence of crime were held to be a search. When the policeman shines his flashlight in the parked car or listens at the tenement door, what else is he doing than searching? When he climbs up a telephone pole and peers beneath a second-story window shade, what on earth is he doing up that pole but searching? … Unless history restricts the amplitude of language, no police investigative activity can escape the fourth amendment’s grasp. (1974, 395)

Vagueness, then, requires discretion in the enforcement of law, rendering it non-uniform and partial. We’re left with a dilemma between vague and determinate law. No matter how we idealize society, the law, or law enforcement, full compliance and full enforcement seem impossible, undesirable, or both. No set of rules can be sufficiently determinate, and the push to make them more determinate can create new problems. The solution, from the perspective of justice, is not to fully enforce traffic laws, even if that were possible. Given unavoidable facts about human drivers and the complexity of traffic laws, full compliance from motorists or full enforcement from police officers is not ideal. In virtue of the unavoidable complexity in a legal code, partial, discretionary enforcement is required to handle the problems that are not idealized away.

These are problems that extend beyond policing and the criminal code. We see them arise in other bureaucratic contexts as well. If the idealized world has state benefit programs, the imperfect system of rules for determining who gets what benefits will produce similar problems (Zacka 2017). Ideal legislation requires idealizing away problems (including problems of reasoning and access to information) that are not usually moral failings. There’s nothing humans can do about the fact that our social world is highly complex and hard to predict and that it is practically impossible to create a set of rules that fully specifies acceptable behavior without backfiring. But whether full compliance looks ideal depends entirely on if we’ve gotten the rules right. To put the point generally, then, the complexity of the social world, and vagueness in all systems of rules, are incompatible with the standard full compliance idealization. Since legislative imperfections result from normal human circumstances, full compliance can be a normative ideal only if we imagine away those normal human problems.

To briefly recap, the need for partial, discretionary enforcement entails that the inequalities that result from the criminal justice system are not justified entirely by merit or desert. This satisfies the final criterion for major institution status. The reply that full compliance undercuts this claim is unsuccessful in light of what it would require to make full compliance a normative ideal. The additional idealizations required to make the full compliance idealization attractive, or ideal in the utopian sense, are many and wide in scope. How do we determine, in a principled manner, which idealizations are methodologically unattractive?

The kinds of idealizations common in political philosophy are justified, if they are, on the grounds that we are trying to imagine what kinds of institutions and rules people would have good reason to obey, so it might be valuable to imagine what the political and social world would look like when we idealize away moral failings such as opportunistic rule violation. Idealizations should eliminate “won’ts,” as in, Don should respect Vic’s intellectual property, but he won’t. Idealizations should not eliminate “can’ts,” as in, Don should eliminate all poverty, but he can’t.⁵ The problem is that the idealizations that go into avoiding the issue of unequal enforcement, or making full enforcement desirable, are not justified on the same grounds. Rather, they take us away from the “circumstances of justice” by ignoring unavoidable problems of rulemaking and following. We’ll see this problem arise for the moral angel idealization in the next section.

4. Moral angels, urbanization, and idealization as abstraction

There’s a common mistake in ideal theorizing about justice: assume that individuals in their private roles are non-ideal while assuming that in their public roles they’re ideal and derive conclusions about just institutional structures from this assumption (Freiman 2017). I’ve argued that the full compliance idealization requires a host of other problematic idealizations to obviate the need for a police role in society that is sometimes partial and discretionary. But I should emphasize that I’m not assuming that in our public, police role, we’ll be perfect and therefore that we can ignore problems of enforcement. Rather, the fact that we won’t be perfect in that role shows that the inequalities that result from policing cannot be justified entirely by desert. Traffic enforcement is a simple illustration of this point. Now I shall turn my attention to another need for an active enforcement role even within a model society characterized by extreme idealization. I’ll argue that this need persists even with a stronger idealization: what we might call the “moral angel” idealization. If we were moral angels, there would still be a need for a police role in managing the use of public space. The reason is that, in contrast to traffic enforcement, rules managing the use of public space are inherently vague. To idealize away this need would require yet more methodologically suspect idealizations.

The discussion so far has focused on a particular kind of policing: law enforcement. Most people associate the police with their law enforcement function. But most police officers are not detectives, and most of what they do is not law enforcement (Wilson 1978, 16; Bayley 1994, 22, 32). Rather, quite a bit of police work falls under the heading of “order maintenance” conducted by patrol officers.

Even if we find the above-described idealizations to go along with full compliance methodologically desirable, they do not eliminate the need for order maintenance policing that responds to various kinds of disagreement or “social friction.” We would have to idealize the citizenry to eliminate perspectival diversity and to eliminate problems that are not moral failings. We would have to idealize away the “burdens of judgment” that lead reasonable people to persistent disagreement (Rawls 1996, 56–58; Levy 2016, 324). This amounts to an abstraction more than an idealization. In other words, we would have to remove these features from the model, but this is not clearly an improvement. Reflecting on the order maintenance policing role makes this clear.

Shopkeepers don’t want people loitering or sleeping in front of their stores, residents don’t want people loudly arguing or fighting near their window late at night, people generally dislike public drunkenness and loud music, homeowners typically frown on neighbors working on or storing cars in their driveways, and so on. These things prevent people from using public or even private spaces, and historically, the police have spent most of their time attending to these issues. This work is described by George Kelling and James Wilson as making streets safer even without reducing crime (Kelling and Wilson 1982; Wilson 1978). Order maintenance policing sometimes relies on enforcing laws, although the aim is not law enforcement. When patrol officers can diffuse a fight rather than make arrests on disorderly conduct charges, they often opt to do so.

An active enforcement or administrative role survives full compliance or moral angel idealizations because such idealizations do not eliminate the burdens of judgment, perspectival diversity, or mental health problems that give rise to much of the order maintenance role. One common police activity is the “wellness call.” Police will respond to reports that someone has gone missing or is acting in a strange or potentially dangerous manner. Often, this is a result of mental health problems. These calls are not occasioned by moral failings. To avoid this, we would need to idealize away a non-culpable, very human problem. The justification for idealizing away common human problems such as greed or selfishness to imagine what justice looks like on a camping trip is that these problems are still moral failings. Even if everyone were moral angels, there would still be the occasional wellness call. To be sure, in many cases, it is preferable to dispatch a medical health professional to these calls. But in at least some of them, the individual in question may pose a threat to themselves or others, and they may need to be forcefully restrained until proper medical care can be administered. This is as true in emergency rooms as it is in the field.

Another police activity that survives appropriate idealizations is the response to disputes about the acceptable use of public space. In highly segregated cities this is easy to overlook because people who spend their free time indoors live near people who do likewise, and people who spend their free time on their porches, in their front lawns, or playing in the streets live near people who do likewise. For another example, the use of public space in New York City’s Times Square or on Bourbon Street in New Orleans’ French Quarter is vastly different from the use of public space in other parts of those cities just a short distance away. Playing live music on the street is expected, and not disorderly, in these areas. If one were to set up an electric keyboard and amplifier on the streets of DC’s Georgetown, however, one should expect a prompt visit from the police. Whatever we want to say about objective, universal principles of justice, surely there’s no way of settling questions about public disorder ahead of time and from the armchair. Where individuals disagree about this, we will need a way of settling disputes and maintaining (hopefully equitable) equilibria. This need arises primarily in urban settings where diversity and dynamism guarantee disagreements at geographic and temporal boundaries.

One of the major problems of American cities in the twentieth century was the “skid row” or “vice district.” Skid rows, defined by sociologist Samuel Wallace as “a distinct ecological area for homeless men in the city,” have high rates of inhabitants with mental illness and especially substance abuse disorders (Wallace 1965, 143–45). The people who lived there, if they worked, were typically itinerant workers without family connections and spent much of their time off work doing very little. They also regard social workers attempting to help them with deep skepticism (1965, 156). Skid rows were an innovation of sorts: “the emergence of skid row meant that the vagrant at last had a place he could call home when he cared to live there. The term homeless … became specifically applied to the single unattached worker who lives on skid row between jobs” (1965, 18). They were the result of complex factors that cannot be explained entirely in terms of moral failure. One factor was the temporary nature and changing location of much manual labor at the beginning of the twentieth century.

Skid rows mostly emptied towards the end of the century, but some remain. Los Angeles’s is perhaps the most famous remaining skid row. Open-air drug markets, sex workers, and substance abuse disorder, public urination, and other “social incivilities” are common (Berk and MacDonald 2010, 814). Many U.S. cities now have homeless encampments that share important similarities even if they are not called skid rows. They are not, however, entirely a result of underfunded homeless shelters (which we might chalk up to a moral failing of those unwilling to fund them). New York City is a “right to shelter” jurisdiction, meaning people can’t be turned away. Yet for various reasons, people regularly opt not to take advantage of resources from the Department of Homeless Services. Especially in warmer climates, existing homeless shelters often go unfilled; “When asked why they ‘chose’ to camp as opposed to other alternatives, the camp residents referred to the shelter in nearly every case, but rarely ever to its inaccessibility. Instead, they referred to the material and moral benefits of the camps over the shelters” (Herring 2014, 306). Those residents described them as autonomous spaces of self-governance providing a set of moral resources absent in the state-provided shelters (Herring and Lutz 2015). At least some encampment residents are there because they reject the “prevailing socio-economic norms” and “contours of America” (Smith 2014, 42).

The vice district is another related phenomenon. While the Bowery neighborhood was New York City’s skid row into the 1970s, Times Square was at that time more straightforwardly a vice district characterized by pornography stores and theaters, sex workers, and drug dealers (Traub 2004). Vice districts and homeless encampments sometimes, but don’t always, overlap. They both rely on uses of public space that, due to the burdens of judgment, many people find objectionable and would like to do without (at least near their home).

In light of the perspectival diversity that characterizes actual societies, this kind of disagreement is inevitable.⁶ But we don’t even need to endorse something as strong as perspectival diversity to get this result. Suppose we’ve reached consensus on which laws are completely just and also that everyone obeys them and respects one another’s rights. Suppose further that everyone is a morally perfect angel with a strong desire to do the right thing at all times. Are all of the things that produce social friction really matters of morality? Will people, under these assumptions, drink on their stoops? Will they move their television onto their driveway to take advantage of a cool breeze? Will they decide to consume recreational drugs and fall asleep in public? Will they play music on the sidewalks or carry on loudly as they walk from place to place? Will they busk on the street or in the subway station in the hopes that you tip them? Will they reject the contours of American life, or suffer amoral problems in life, and opt for the homeless encampment? Will people with mental health problems struggle to live “typical” lives and make use of public space in controversial ways?

If our idealizations eliminate these behaviors, then political philosophy risks moralizing in the objectionable sense of the term and risks being objectionably intolerant of various kinds of diversity. And if we wave our magic idealizing wand to eliminate these problems—including mental health problems, difficulties in convincing people to take advantage of unused shelter capacity, and unusual preferences to live unattached, itinerant lifestyles—we can no longer appeal to the common justification for idealizations. So whereas Rawls imagines the circumstances of justice involving resource scarcity and conflicts of interest between people roughly similar “in physical and mental powers,” even this more moderate form of idealization looks methodologically unattractive (Rawls 1999, 110). Again, one good reason to idealize away opportunistic rule violations is to figure out what sorts of institutions we ought to accept. Idealizing away problems, such as perspectival diversity and typical human problems, including dissimilarities in physical and mental powers that are not moral failings, cannot be justified on the same grounds.

Idealizing perspectival diversity away would require idealizing away both (some) reasonable conceptions of the permissible use of public space and the predictable disputes that result from dynamic land use. These are not problems to be idealized away such as the unreasonable anarchist or the intolerant ideologue who insists on optimizing for their own idiosyncratic perspective on justice. This is not an idealization, but a methodologically and substantively undesirable abstraction. Similarly, idealizing away the sorts of problems that generate wellness checks, while an improvement, is just an abstraction from the kinds of amoral human problems that generate the need for human cooperation and institutions.

The strategy that we’ve largely settled on is to restrict certain kinds of land use to certain areas through formal zoning or informal vice district policies (Garnett 2010). But this is still not entirely sufficient. It appears impossible to fully specify the acceptable use of public space even if we imagine everyone as perfectly motivated to follow the rules. Vagueness will be embedded in the rules. Furthermore, vagueness will inevitably interact with conflicts on the margins of formal or informal districts given that land use is dynamic and without sharp boundaries in practice. We cannot, in other words, idealize away the dynamic nature of social life that can upset local equilibria any more than we can idealize away pluralism regarding the acceptable uses of public space. It is one thing to assume that everyone is perfectly motivated. It is another to assume that people will never upset the local sensibilities. We know from experience that these problems and diverse perspectives will create social friction and disagreements. But there is no reason to think they can be solved by legislatures alone. Rather, something like the police officer as “streetcorner politician” is needed to adjudicate disputes (Muir 1977). How the police render their services will clearly influence the kinds of lives people can live in a way that is unlikely to be justified by desert. Given this, discretionary enforcement is a practical reality and necessity even under highly idealized full compliance or moral angel conditions. There is, therefore, always the likelihood that the inequalities resulting from the criminal justice system will be determined by policing decisions rather than merit or desert alone.

5. Is this a police force?

To briefly recap, the last two sections are aimed at demonstrating two things. First, moral idealizations are insufficient to eliminate the need for some amount of active enforcement or administration of justice. This need remains even if we’re fully compliant moral angels. Second, because we’re caught between the problems of determinacy (over-criminalization) and vagueness (indeterminate criminalization) in the legal code, partial enforcement is unavoidable; idealizations can reduce, but not eliminate, inequalities and impacts on ways of life that are not justified by merit or desert.⁷ So, law enforcement institutions are entitled to major institution status.

Before concluding, one major objection naturally occurs. Some readers will object that the kind of enforcement discussed so far doesn’t require police. Some will insist that there will be no need for coercion to settle disputes over public space if we’re moral angels. Perhaps the remaining enforcement functions do not actually amount to a police force, at least if we understand the police to be essentially coercive. Perhaps what’s left looks more like social work than police work, and perhaps it can be supplied entirely by private cooperation. Tight-knit communities might regularly meet to fine-tune their policies on disorder, and members might take turns on community watch groups to smooth the remaining rough edges. Medical professionals might respond to wellness calls. Traffic enforcement might be conducted by unarmed security guards. Maybe there is no need for the police in ideal theory after all. There are several things to say here.

First, the point I’m making is not that there would be a need for government even if people were morally perfect (Kavka 1995). Rather, the point is just that there will always be some need for the role that police sometimes fill now, a role characterized by the active administration of justice involving the adjudication of street-level disputes and the enforcement of some rules to serve that end. I’m taking no stand on the anarchism or privatization questions; the police function could be supplied by public police forces, private protection agencies, community organizations, or some combination thereof.

But why think there would be any need for police agencies of any kind? What about peaceful anarchist means of resolving these disputes? As argued in the previous section, moral angels will still have disagreements, and there’s no reason to assume that a desire to resolve them peacefully will always result in their peaceful resolution without any need for institutions of some kind to facilitate the resolution. There are likely to be second-order disputes about whether the system for adjudicating our first-order disputes have gotten things right (Kavka 1995). Assuming that these disputes can be handled in some deliberative fashion without the need for order maintenance requires a philosophically uninteresting level of idealization.

Could the police in this model society simply remind people of the local ordinances that they’ve violated without relying on coercion? In some cases, surely the answer is yes. In others, however, a dispute occasions the relevant violation. Often, the dispute will involve the violation of a vague statute. Consider a loitering complaint in which the alleged loiterer is in front of a store. We can imagine that situation arises as a result of blameless factors. Still, the proprietor feels annoyed by this and worries that people are opting to shop elsewhere to avoid any potential hassle. The police are called and expected to resolve the situation. In such a case, both think they’re within their rights, even if they recognize that the other is unhappy. Morality, they know, does not demand that one’s actions please everyone involved. And it will often not be clear, from the language of the statute, whether the law has technically been broken. But the conflict needs resolution, and facilitating a dialogue between the two parties is not guaranteed to work. The officer must make a decision that has to be binding. It is true that in a world with compliant angels, the decision will probably not require the actual use of force. Nevertheless, the outcome of the interaction is that someone must change their behavior, and even if they accept the result, they were still forced into it.

We might imagine that moral angels will recognize the outcome of the decision-making procedure as just and authoritative and therefore voluntary endorse it. If so, this would not be coercive. One problem with this response is that the procedure would have to be infallible, or close to it, for disputants to simply substitute the outcome for their own judgment. Another problem with this response is that vagueness in the law will enable the second-order disagreement to remain. In these cases, there probably isn’t a single, determinate, correct outcome. To idealize away this (admittedly extremely mild) form of coercion requires, again, idealizing away cognitive imperfections in addition to vagueness and complexity in the law.

The second thing to say about the objection is that our popular understanding of the police might be misleading us. The reason this objection—that what remains is not a police force—is intuitive is simply because the popular conception of policing is skewed so heavily towards its law enforcement function. The image of a SWAT officer tossing a flash grenade through a window before violently serving a drug warrant correctly pushes us towards thinking that there would be no need for the police if only our society were more ideal than it actually is. Much of patrol work, however, is a kind of social work, and calls for “community policing” are calls for a different kind of policing. To say that there is a need for a police function in reasonable kinds of ideal theory is not to say that there is a need for all of the current police function. The streetcorner politician who occupies the order maintenance role in an idealized society takes up a drastically reduced police role.

Third, and most importantly, disputes about how to provide these remaining functions are ultimately disputes about how to structure police agencies and provide police services. Perhaps the functions will be so diffused that nothing like a familiar, singular police agency remains. The de-policing initiatives in some cities today shed some light on this. Instead of treating homeless encampments—a problem that could persist in methodologically attractive versions of ideal theory—as a “police problem,” some cities have turned to other agencies to manage encampments. Portland, for example, has begun using city contractors. They notify residents that they have 48 hours to leave the encampment. Whatever possessions remain after that time are confiscated and discarded by the contractors (Saslow 2021). This strategy is basically the same as those employed at various times by the NYPD and LAPD who confiscate and discard the possessions of the homeless or unlicensed vendors to reclaim the public space (Duneier 1999, 185; Stuart 2016, 240). What we see, then, is that moving away from what we now think of as a police force does not entail moving away from active, at times coercive, policing of the use of public space.

The claim is not that this use of city contractors is ideal. It is just that preferences for how to provide the functions that I’ve argued will survive appropriate idealizations are preferences for how we institutionalize policing. Extensive idealizations drastically reduce the need for policing but do not eliminate it. We are then left with questions about how ideally to respond to such problems. Certainly, the ideal looks drastically different from what we have now. But this is not to disqualify policing from “major” status. It is just to take a stand on the issue of the structure and nature of police institutions.

6. Ideal policing

Once again, police agencies serve as input filters to the criminal justice system. Police power and the criminal legal system post-arrest generate a wide variety of morally significant costs and inequalities. Because of the unavoidable fact of partial, discretionary enforcement, police agencies create inequalities that are not fully justified by merit or desert. Policing shares enough features with other major institutions when we drop the full compliance (and enforcement) idealization to earn the status of major institution. But those remain to a lesser degree even if we apply idealizations to bring our model society as close to full compliance or moral perfection as possible.

One objection to the full compliance and moral angel idealizations is that they are analogous to idealizing away resource scarcity in economics or friction in aerodynamics (Freiman 2017, 6; Levy 2016, 323). I have granted that idealizing away serious forms of injustice may be methodologically appropriate. Instead, I’ve relied on the weaker claim that idealizing away resource scarcity, complexity, error in the application of vague rules or principles, and non-culpable human diversity is inappropriate. The objection I’ve developed here is that the full compliance idealization creates a model that a society of humans cannot possibly achieve even if we were perfect.

But the goal is not simply to critique certain approaches to ideal theory. The point is that ideal theory has artificially restricted the scope of fundamental questions in political philosophy. The narrowing it induces undermines our ability to effectively theorize about our social world. This is true even if we reject utopophobia and accept that a theory of justice could be practically unlikely. As the discussion of policing suggests, there are deeply important issues of justice in the manifestation of our social and political systems that we can’t apply our theory of justice to if we idealize away the facts that determine how our institutions are realized.

One upshot is that theories of justice are incomplete if they ignore problems of administering justice. If I am right, police agencies and institutions are as central to political philosophy as the structure of constitutions, democratic decision-making bodies, markets, and families. This is true even under rather strong idealizations. Another is that, given the complexity of the social world, attempting to idealize individual features of it is fraught. Idealizations that appear desirable might interact with other features in surprising and non-ideal ways. And we should probably abandon the view that full compliance is genuinely ideal (unless we’re willing to take on the host of other idealizations to make it ideal, and we shouldn’t be).

So, what are theorists to do? For those who are attracted to ideal theory, theorists can identify certain important institutional features of policing against a background of (weaker) idealizations. If policing the use of public space is an ineliminable part of social life even in ideal circumstances, then a theory of justice had better have something to say about what that should look like. The considerations on order and the use of public space discussed throughout suggest at least one important fact about the institutional structure of policing: order maintenance policing should be provided as locally as feasible. Just policing requires sensitivity to local contexts that can only be provided by institutions that represent local interests. Majoritarian decision-making beyond the neighborhood level is likely to be unjust even if local interests are represented. This point stands in contrast to proponents of centralizing police power (cf. Ostrom, Parks, and Whitaker 1973).

The case I’ve made shows that we can’t stop there because idealizations that are less extensive than full compliance will introduce new, complex, and hard-to-predict problems. Hyperlocal decision-making, for example, could create problems of its own (as we now see in, e.g., housing and education policy). This is a difficult problem to solve. For example, residents who live near a park might have idiosyncratic views about its acceptable use that ignore the interests of outside stakeholders. It is unclear how pronounced this problem would be in highly idealized models, but it is one that must be accounted for.

The early history of policing suggests that polycentric governance could abate this problem. When Frederick Law Olmsted had completed Manhattan’s Central Park, order maintenance issues became especially pressing. People simply didn’t realize that certain uses of the park land that at the time were common (felling trees, allowing animals to graze, picking flowers) would render it unusable by others. Olmsted fought a political battle to keep the park police as a separate agency, one that focused on the park rather than on crime in the city more generally and one that protected stakeholder (not just resident) interests in the park. He ultimately lost that battle, and policing in the city underwent a steady stream of consolidation. But while he was in charge of the park police—he insisted on calling them “keepers” instead of “police”—the agency took up a lenient and educational, rather than punitive, approach (Thacher 2015). The point of this illustration is that thinking about ideally just policing requires thinking “defensively” about the problems we are likely to encounter, ones that persist even in the face of various idealizations.

Additionally, the ineliminable discretion that characterizes even idealized policing (due to problems introduced by highly determinate law, vagueness, and complexity) entails that a theory of justice will require principles for guiding the administration of justice. The full compliance project has obscured this by overlooking the impossibility of full compliance and enforcement. In light of the pluralism that survives methodologically attractive idealization, there is good reason to prefer the police to be as lenient as possible and to endeavor to police in a way that enables diverse ways of living.

It is true that many of the most significant questions about just policing will arise outside highly idealized contexts. Of course, that’s true of most questions about justice. Yet, if our theories of justice and legitimacy are fine-grained enough to require particular kinds of economic and democratic institutions, surely they will have particular institutional demands of police agencies. These demands should be the subject of robust philosophical scrutiny. This paper aims primarily to establish that claim. Defending in any detail the institutional demands is a task for future work.