Abstract
This article evaluates responses to air pollution that lie beyond the domain of state policy and nonviolent civil disobedience. I begin by presenting evidence that fine particulate matter pollution is more serious than commonly recognized. As a result, attempts to address it deserve more attention from those who are concerned about achieving a just and livable world. I argue that, in part because of its physical characteristics, air pollution pushes the limits of civil disobedience as moral communication. I evaluate one nonviolent and one unarmed violent strategy for collective self-defense against air pollution and make at least two concrete recommendations for engaging in collective self-defense against air pollution.
Introduction
Air pollution is likely to kill more people than climate change, HIV/AIDS, COVID-19, and all wars. This is true each year and it is likely to be true across the aggregate of human history. Air pollution represents the most lethal engineering-political problem in the world. However, there remains a dearth of reflection in the literature about how to think about air pollution as an urgent political problem.
Activities like running internal combustion engines, burning crops, and cooking food produce small pieces of matter that hang in the air. Particles smaller than 2.5 μm (PM2.5) may pass into the body, usually by inhalation. These particles are tiny: it would take thirty of them to span a human hair. Once inside the body, these fine particles can cause significant damage to a variety of organ systems, especially the lungs and heart. For an intuitive sense of the problem, imagine spritzing a fine poisonous grit inside a delicate, sophisticated engine.
This article begins by presenting evidence for the seriousness and lethality of air pollution today. I discuss various estimates of air pollution’s effects in the literature and suggest that air pollution ought to be taken seriously by those who aim to build a more just and livable world and especially by those who are concerned about human survival as a concrete political question (Mohorčich, 2021). While some state interventions have reduced air pollution, they have also been slow, unreliable, unequitable, and fragile. I consider non-state interventions intended to reduce vulnerability to air pollution, which means discussing civil disobedience. I discuss the limitations of a conceiving of civil disobedience as moral communication and emphasize instead using direct action as a means of harm reduction and collective self-defense.
As a way of concretizing this discussion, I evaluate two specific interventions against air pollution. One of these interventions is nonviolent and one involves property damage. The first uses nonviolent techniques like passive resistance to establish buffer zones between car traffic and places where people vulnerable to air pollution (the very old and very young, e.g.,) gather. A number of historical precedents for this sort of street closure exist, as do precedents for forms of passive resistance that have succeeded in repurposing public space. The second intervention involves an organized campaign of sabotage of privately-owned cars parked in cities. The potential harm-reduction benefits of such a campaign are high and there is some empirical evidence supporting the efficacy of sabotage in democratization movements. But several counterproductive effects are possible, including increased policing and compensatory emissions.
Air pollution is serious and urgent
Humans have evolved in environments with mostly clean outdoor air, no widespread industry, and few anthropogenic emissions. This background, combined with the fact that most of air pollution’s effects are long-term and not a source of strong selection pressure, mean that human bodies have evolved few good defenses against fine particulate matter. Increases in human industrial activity, however, have led to substantial increases in the amount of fine particles in the air over the past two centuries. Humans have increased many times over the presence of a pollutant against which we have few effective defenses.
As the tools for measuring air pollution have grown more sophisticated, scholarly estimates of its health effects have become more ominous. In Burnett et al. (2018) published a model for estimating deaths from PM2.5 exposure. Their model, the Global Exposure Mortality Model (GEMM), predicted that deaths in 2015 from PM2.5 exposure were approximately 8.9 million (95% CI: 7.5–10.3 million). This was the latest in a line of estimates that had placed human deaths from air pollution higher than previous work in the literature. For example, widely-used modeling by Cohen et al. (2017) a year earlier based on developing integrated exposure-response (IER) functions for several main causes of death (chief among them heart attacks, strokes, and lung cancer) had estimated global deaths from air pollution at 4.2 million. The 2010 Global Burden of Disease comparative risk assessment attributed 3.2 million deaths in 2010 to fine particulate matter (Lim et al., 2012). This same study ranked “ambient particulate matter pollution” as just the ninth-most severe risk factor for disease burden. (Using contemporary numbers, PM2.5 exposure would now be ranked at least second in global disease burden.) Earlier work from Cohen et al. (2005) put yearly premature deaths at 800,000. Earlier still, the literature is full of scattered warnings but few robust scientific objects like global mortality estimates. Consider accounts of the London smog of 1952, in which a temperature inversion in the Thames River Valley and high levels of low-quality coal combustion veiled London in thick yellowish-black smog from December 5th to the 9th. About 12,000 people died over the next 3 months, 4000 of them more or less immediately (Bell and Davis, 2001). To describe such an event as a canary in a coal mine is grotesque because it is too literal. And although hospitals and morgues were swamped, there is little evidence of citywide panic, certainly nothing commensurate to the smog’s lethality. Only later, in a careful public health accounting that has spanned decades, has the magnitude of the danger emerged. The same process that occurred in the decades following the London smog is now unfolding for global particulate matter. The tools for measuring fine particulate matter’s presence and health effects have grown more sophisticated and convincing. As the literature has become more detailed since 2010, credible estimates of death totals and disease burden from air pollution have roughly tripled in size and severity. All the world is London.
Lelieveld et al. (2020), using the GEMM methodology from Burnett et al., estimate that if all “potentially controllable anthropogenic emissions” that contribute to air pollution were eliminated, global life expectancy would increase by 1.7 years (CI: 1.4–2.0). If all fossil fuel emissions (a subset of all potentially controllable human emissions) were eliminated, global life expectancy would rise 1.1 years (CI: 0.9–1.2). All air pollution (human-caused and not) depresses global life expectancy by 2.9 years. 1 For comparison, human-to-human violence lowers global life expectancy by about 0.3 years. One way to think about Lelieveld et al.’s findings is that you personally should expect to die 2.9 years earlier than you would in a world without widespread air pollution. (If you live in Europe or Asia, this number is probably a little higher.) One implication of Lelieveld et al.’s findings is that if you had two buttons in front of you, one labeled ‘World peace’ and one labeled ‘Reduce air pollution by 20%,’ you should consider pushing the latter.
Cohen et al. (2017) find that deaths from air pollution have increased from 3.5 million in 1990 to 4.2 million in 2015—and this estimate is based on an IER model now considered too conservative by 40–50%. It should be noted that air pollution’s effects are somewhat, but not completely, concentrated in south and east Asia. Cohen et al. (2017), using an IER model, find that 59% of the global health effects from exposure to fine particulate matter are spread from south to east Asia. Nagpure et al. (2014) estimate that Delhi alone saw 18,229 deaths from particulate matter exposure in 2010, up from an estimated 11,394 in 2000. Each residents of that city would receive on average more than 9 years of additional life if air pollution were reduced—not eliminated—to WHO-approved levels in Delhi (EPIC, 2021). In a metro area of 31 million, this represents nearly 280 million years of human life lost to air pollution. As harrowing as these numbers can be, it is crucial not to think of PM2.5 pollution as mostly affecting major cities in northern India and eastern China. Air pollution remains a global problem, including in rich European countries with putatively robust environmental protections. Europeans, in part because of their countries’ embrace of diesel vehicles, breathe some of the most consistently polluted air on Earth. PM2.5 levels exceed the WHO’s recommended annual concentration of 10 μg/m³ in a majority of the measuring stations in 83% of European countries. A review of WHO data from 2016–2018 reveals that only seven European countries of 41 studied (Spain, Estonia, Finland, Norway, Portugal, Iceland, and Sweden) avoided breaching the 10 μg/m³ limit in a majority of their cities (WHO, 2018).
Even the WHO’s 10 μg/m³ limit for fine particulate matter, a limit the majority of the world’s population spends its time well above, is somewhat arbitrary. No safe level of PM2.5 exposure has been established, and the literature implies that even modest concentrations have severe effects. For example, Apte et al. find measurable increases in stroke and ischemic heart disease beginning as low as 6 μg/m³ (Apte et al., 2015). Measurement limitations and available data make studies on particle concentrations much below that level difficult. It seems likely that, as with lead exposure, there is probably no ideal level of PM2.5 above zero.
It is important to note that the air pollution being modeled in these papers is almost always fine particulate matter and not greenhouse gas emissions associated with human activity, like carbon dioxide or methane. The climate effect of these gases is therefore separate from the death totals attributed to air pollution itself. Most, though not all, efforts to reduce PM2.5 emissions will reduce carbon emissions and vice versa.
Air pollution kills ∼five to thirty times more people per year right now than climate change is projected to kill during some of its most severe decades later this century (Weinberger et al., 2017; WHO, 2021). But air pollution rarely features as a major topic of discussion for political science and theory. In the last 5 years (2016–2021), Political Theory, APSR, the Journal of Politics, and Polity together have published one article, or less than 0.2% of their aggregate output, to do with air pollution. Perhaps this is because most air pollution discussion occurs in specialist journals that focus on interactions between environmental and political questions. These journals do publish more work on air pollution, but the quantity of articles is far from overwhelming. Environmental Politics has published six articles about air pollution in the last five years, or about 2.5% of its total. Environment and Planning C: Politics and Space has published 12, or about 3.5% of its output. None of this is to condemn any journal or subfield; it is simply to substantiate the claim that air pollution does not enter in the first rank of political or social theorists’ concerns, even among those who are explicitly concerned with questions of politics, environment, and space.
What would collective self-defense against air pollution look like?
If a government or other group of people were killing 8.9 million humans per year, nonviolent direct action to prevent this would likely be justified. Political philosophers have widely regarded legal protest and civil disobedience to be legitimate forms of resistance to unjust policies (Rawls, 1999: 319–43; Sabl, 2001). The harder line of questions that I wish to open has to do with action that rearranges the boundaries of civil disobedience. I mean ‘rearranges’ both in the sense of “reorders internally” and “moves closer in or further out.” As a way of investigating these questions, I will discuss two concrete examples of resistance that could reduce harm from air pollution: the nonviolent taking of space meant for cars and the sabotage of combustion vehicles that are stored and used in cities for private purposes. Sabotage is generally set beyond the pale of conscientious civil disobedience. It is my hope that examining specific tactics of violence and nonviolence—whatever the verdict—enables clearer thinking about air pollution and the tactics and choices available for reducing its harms.
Political philosophers have, at least since Rawls, generally argued for four central features of legitimate civil disobedience. In decreasing order of consensus and increasing order of controversiality, these features are conscientiousness (the act must be thought through with respect to those it affects), communication (the act must seek to transmit a perceptible meaning to others), publicity (the act must be committed, even planned, in the open), and nonviolence. Conscientiousness has to do with a serious, sincere, and sedulous accounting of an act’s likely effects. It remains crucial for justifying political action of all types. Communication and publicity are helpful values in many contexts. But they are not quite essential to, say, taking streets from cars in a way they may be essential for other acts of civil disobedience whose aim is to increase awareness of a given injustice (see, for a related argument, Celikates, 2016). It is possible that setting up sawhorses to keep non-essential vehicles away from schools would ‘raise awareness’ of air pollution and its effects. Awareness has its uses. But centering communication in this case is a little like saying the primary point of putting on an oxygen mask in a plane is to raise awareness that the cabin is depressurized. This is potentially a nice side-benefit, but it is not the central purpose of wearing an oxygen mask. The point of putting on an oxygen mask is to avoid dying of hypoxia and to enable yourself to help others put on their masks. In much the same way, the central purpose of the actions considered here is to protect yourself and others from as much harm caused by air pollution as possible. Because the values of communication and publicity are auxiliary to this purpose rather than central, and because the protection of life and wellbeing is central, the actions considered here are better thought of as forms of collective self-defense and harm reduction first and as communicative civil disobedience second. (This is not to say that moral communication is worthless. On the contrary, it is important. But it is not always the first or only concern.)
The idea of civil disobedience as communication structures debates about what counts as legitimate civil disobedience, the purpose and uses of nonviolence, and the permissibility of violence for political ends (see e.g. Celikates, 2016; Çıdam et al., 2020). Conceptualizing civil disobedience first as plea obscures questions of survival, self-defense, and harm reduction, and makes it harder to formulate strategies around these considerations. Values like publicity and the transmission of meaning to observers are important for civil disobedience, the story goes, because the function of civil disobedience is largely to demonstrate or display to others why they ought to regard a certain state of affairs as unacceptable. The wellbeing of the person committing civil disobedience is of importance only insofar as the suffering of that person can be used to communicate to third parties the injustice of what is taking place. This conceptualization often remains intact even in works that purport to examine the status and formulations of civil disobedience within law and political philosophy (Cohan, 2007). This conceptualization, however, tends to foreclose possibilities of harm reduction and self-defense not just in analyses of civil disobedience but also within the nonviolent tradition more broadly, a tradition that is understood mostly as a means for creating forums for different sorts of moral appeals. Making your actions perceptible to third parties, the idea goes, differentiates what you are doing from vandalism, terror plotting, theft, plain crime—that is, from tactics that depart from what is public and can be publicly communicated. But, of course, there are forms of action (some of which appear in this paper) that are neither nonviolent communicative action aimed at convincing third parties (‘civil disobedience’) nor the other thing for which we reserve words like ‘crime’ and ‘violence.’ This gap, between civil disobedience and plain violence, is why it is important to partially extract the question of violence and nonviolence from the question of communication. Nonviolence is for more than moral suasion, exposing injustice, or engaging in certain forms of public communication. It can be used to save lives and reduce harm directly.
But so can violence. Or so the story goes
Discussions of violence and its role in political action have traditionally centered on certain ethical dimensions of violence and nonviolence: is violence justified? If so, under which material and historical conditions? Who can do perpetrate it and who suffer it? These are excellent questions and rightly command scholarly attention. Jennifer Welchman, William Scheuerman, Andreas Malm, Steve Vanderheiden, Michael Martin, and Thomas Young are among the writers who have offered qualified defenses of ecological sabotage (Malm, 2020; Martin, 1990; Scheuerman, 2021, 2022; Vanderheiden, 2005; Welchman, 2001; Young, 2001). Vanderheiden, for example, uses elements of just war theory to conditionally defend sabotage like that committed by EF! and the ELF. However, he cautions against using the language of self-defense to describe ecological sabotage. He does this in part because these acts of vandalism are only indirectly connected to the bodily wellbeing of the saboteur and other humans in general. Ecological sabotage should be “reserved for extraordinary circumstances and then narrowly targeted to the exigencies of specific cases” (Vanderheiden, 2005: 444). This is a sensible and indeed frequent recommendation in the literature on ecological sabotage. But what if, as the foregoing sections demonstrate, there exists a direct, urgent danger to one’s own life and health—and the life and health of those in one’s community—emanating from a specific, physically proximate source susceptible to sabotage? This tight coupling of emission and harm is not true of greenhouse gases. Carbon emitted down the street or across the world has the same long-term effect on, say, the residents of low-lying islands in the South Pacific. PM2.5 emissions are different. It is enormously consequential to the children on a playground whether PM2.5 is being emitted two or two thousand meters away. Air pollution, because it geographically binds sites of emission with sites of harm, necessitates a more direct discussion of self-defense.
In part because I am concerned with self-defense and harm reduction rather than a civil disobedience whose communicative moral force is central to fulfilling its ends, this article is primarily interested in the question of whether violence works to achieve the aims of those carrying it out. Can violence, as Butler (2021) asks, be reimagined and repurposed to protect us from violence? Recall that famous nonviolent advocates like Te Whiti o Rongomai, Mohandas K. Gandhi, Martin Luther King, Jr, and Cesar Chavez make arguments not only about the moral force of nonviolence but also about its effectiveness. Similarly, advocates of tactical or emancipatory violence tend to claim not only that it is morally justified, but that it works (Churchill and Ryan, 2017; Fanon, 2002; Gelderloos, 2007). Meanwhile, Maria Stephan and Erica Chenoweth famously argue that ‘civil resistance’—a form of resistance Stephan and Chenoweth intend to encompass broader tactics than civil disobedience, but that is still chiefly nonviolent—remains more effective than violent resistance in most cases. Specifically, they find instances of nonviolent resistance from 1900 to 2006 were about twice as likely (53% against 26%) to succeed as were violent resistance movements (Stephan and Chenoweth, 2008). The difference in success rates, they argue, is partially explained by the fact that nonviolent resistance is likely to win a broader base of support than violent resistance. Violent state reactions to nonviolent protest, for example, are likely to convince other members of the political community that the state is reacting disproportionately and the protest movement is justified. Violent movements, by contrast, are less likely to be seen as legitimate and therefore less likely to win a broad base of support, to win elite support, or to incite loyalty-switching by police or military, all of which help movements achieve their aims.
Stephan and Chenoweth’s account has been both supported by further work, like that by Teorell (2010) on the determinants of regime change, and criticized on methodological grounds, especially for the manner in which certain cases that seem to involve some measure of violence are coded as nonviolent and for the ways that apparent nonviolent failures are excluded from the data set (Chenoweth, 2016; Gelderloos, 2007, 2015; Lehoucq, 2016; Malm, 2020; Onken et al., 2021). Anisin (2020) both expands and criticizes Stephan and Chenoweth’s work. He expands the dataset of oppositional movements to 1800 and uses four distinctions instead of Stephan and Chenoweth’s two. Anisin finds that nonviolent campaigns succeed 48% of the time, movements that use “unarmed violence” succeed 61% of the time, those using “reactive unarmed violence” succeed 60% of the time, and “fully violent” campaigns succeed 30% of the time. Anisin’s work demonstrates something familiar to many students of social movements but sometimes backgrounded in attempts at quantification: the quality of violence and the place it assumes in political and narrative structures matter more than its mere presence or absence. The firmest takeaway from the literature around violence and nonviolence is probably not, as Stephan and Chenoweth’s findings have sometimes been used to argue, ‘avoid every form of violence, including self-defense or property destruction, in activist movements’ but instead something closer to ‘avoid, if at all possible, armed conflict with the state.’ (Kadivar and Ketchley (2018) reinforce this conclusion when they suggest that “remaining unarmed may be more consequential for a democracy campaign than adhering to nonviolence.”) What the work of Stephan and Chenoweth, Anisin, and other defenders and detractors of violence underscores is how much depends upon the particulars of each case, how much can be accomplished with nonviolent means of self-defense and harm reduction, and that substantial prospects and dangers worth studying lie beyond nonviolence.
In any case: if we do want to limit the harm caused by air pollution, but our interventions will likely harm others, how do we maximize harm prevented and minimize harm caused within a given set of practical constraints? Let us evaluate two possible interventions: the nonviolent establishment of car-free zones and the sabotage of cars in cities.
Car-free zones and harm reduction
Most PM2.5 from on-street sources comes from the tailpipes of internal combustion vehicles. Consider a project of establishing of car-free or car-light zones around the places and people most vulnerable by air pollution. This would include schools, especially those with street-level entrances, exits, and playgrounds, retirement communities, and neighborhoods where existing disease burdens and PM2.5 levels are unusually high. In New York City, for example, it is not uncommon for children’s playgrounds to be separated from active roadways, many of which are truck routes, by a chain-link fence and two meters of sidewalk. Chain link fences do not filter out anything smaller than a grapefruit (PM100,000), so children are exposed to considerable PM2.5 from the tailpipes of passing vehicles. Spira-Cohen et al. (2011) demonstrated how much particulate matter students in the South Bronx are exposed to by giving them air quality monitoring devices to carry on the way to school. These monitors established that the children were exposed to fine particulate matter levels regularly exceeding the EPA’s already-high 35 μg/m³ limit. (Recall that increases in ischemic heart disease and stroke are linked with levels as low as 6 μg/m³; see Apte et al., 2015). About one in five elementary to middle-school aged students in the South Bronx attend school within 150 meters of a major road, and all attend schools sited directly on smaller roads open to car traffic (Fernandez, 2006; Spira-Cohen et al., 2011).
Because of atmospheric diffusion, standing just a few meters from a pollution source instead of right next to it sharply reduces a person’s exposure to that source. Polidori et al. (2010) and Zhu et al. (2002), for example, have found that airborne pollutant concentrations drop rapidly as measurements are taken further from the source of pollution and that these reductions are initially quite rapid before leveling out. For example, Polidori et al. find that if PM2.5 is at 22 μg/m³ next to an urban roadway, it can be expected to drop to ∼16 μg/m³ just 25 m away, before leveling off to 14 μg/m³ at >50 m of distance. The outcome differences between 16 and 22 μg/m³ are substantial—following Apte et al. (2015), probably in the tens of thousands of deaths across a population of several million. In other cases, such as a truck loading and unloading zone studied by Polidori et al., the drop-off in PM2.5 is spread over greater distances. In this case, about 70% of PM2.5 reduction occurred in the first 250 m. The remaining reductions took place gradually from 250 to >3000 m (Polidori et al., 2010). Even modest differences in proximity to the pollution source matter a great deal: imagine the difference between a puff of perfume in your face and the same spritz across the street. This means that keeping cars and trucks even the width of one city street (3–20 m) away from school loading zones, playgrounds, hospitals, assisted living facilities, and so forth would provide substantial protection to the people most vulnerable. Keeping combustion vehicles even further away, say by establishing one square city block of space (70–270 m) where cars are not admitted, would prevent considerable suffering and death both by keeping emission sources further from vulnerable groups and by disincentivizing driving, which reduces total emissions.
In some cases, local governments in cities like Mexico City, Oslo, Bolzano, and Bogotá have themselves endeavored to establish some network of car-free zones. However, such protections are often partial (e.g., dissuading older model cars from driving into the city center on certain days), unevenly implemented (rich neighborhoods are more likely to receive car-free programs than others), slow, unagile, susceptible to challenges from those invested in the status quo (in New York City, for example, police often refuse to enforce street closures and drivers frequently remove or destroy barriers to cars—see Cruz, 2020), and do not aim to protect the most vulnerable (see in particular Montoya-Robledo et al., 2022 for a discussion of domestic workers’ exposure in Bogotá).
These gaps and failures, coupled with the urgency and lethality of exposure to tailpipe emissions, invite the question of whether extrastate efforts to develop physical protection against cars are justified and how such efforts might be carried out.
A variety of precedents exist for reclaiming streets from cars exist. In the United Kingdom in 1991, EarthFirst! began occupying streets and roads, using nonviolent tactics to take space from cars and repurpose it for children’s play areas, political action, and noncar recreation and transportation. These efforts have continued as a part of the Reclaim the Streets movement, which spread to Finland in 1997 and to much of the world after. In Los Angeles, Rojas (2010) has documented how immigrants from Latin American have “retrofit” the city’s “auto-oriented built form to make it pedestrian friendly” through new patterns of behavior and the retaking of space meant for vehicles. During the COVID-19 pandemic, Berlin, Amman, New York City, Athens, and dozens of other world cities closed streets to cars to allow for more cycling and walking. Often, these programs were as simple as setting up community-maintained sawhorses at each end of a block and directing cars to go around. How effective would a similar, community-led program of seizing streets for self-defense against air pollution be?
Let’s start by estimating the ameliorative effects at one specific site. Consider Public School 184m Shuang Wen in Manhattan. PS 184 serves about 650 pre-K through 8th-grade pupils with around 100 staff and is bounded by Cherry Street (one car travel lane and two parking lanes) to north, Montgomery Street (two travel lanes, two parking lanes, two bike lanes) to the east, and South Street (two travel lanes and one parking lane) to the south. FDR Drive, a six-lane elevated highway, stands 16 m south of Shuang Wen’s playground. The 3-year average for the nearest PM2.5 monitoring station on the first days of September and October is 26 μg/m³. Because PS 184 is so close to FDR Drive, its PM2.5 exposure likely higher than this station indicates, increasing the potential impact of ameliorative efforts. In any case, we know from Apte et al. that reducing PM2.5 from 25 to 10 μg/m³ avoids substantial amounts of sickness and death, mostly by reducing heart attack and stroke mortality, although COPD and lung infections also factor in. Reducing PM2.5 exposure to 10 μg/m³ avoids ∼90% of air pollution mortality in the 11–25 μg/m³ range. So, if Shuang Wen’s average estimated PM2.5 exposure could be lowered by about 62% (to ∼10 μg/m³), students and staff could avoid about 90% of the expected future mortality at that school. The effect is nonlinear: lowering PM2.5 by 42% (to ∼15 μg/m³) would still avert something like 70% of the expected mortality at Shuang Wen.
We know from Polidori et al. and Zhu et al. that atmospheric diffusion means we can expect reductions of around 27% from 25 m of distance from a roadway and around 36% from 50 or more meters of distance from a roadway. Closing Cherry Street between Clinton and Montgomery, Montgomery Street between Cherry and South, and South Street between Montgomery and Clinton to vehicular traffic would increase the average distance between the school building and playground perimeter to the roadway from <5 m to about 50 m. This increased buffer would lead to an expected reduction in PM2.5 of about 35%, or to about 17 μg/m³ (only considering car-derived PM2.5 for the moment—other PM2.5 sources, like industrial production or fugitive dust, would not be directly reduced). This reduction would lead to a reduction in expected future mortality of around 65% of at PS 184. If the average student at Shuang Wen is expected to die ∼2.9 years earlier due to air contamination (Lelieveld et al., 2020), if students are in school for about 9 months, and if young bodies are especially vulnerable to air pollution (as they are—see Gauderman et al., 2015; Gilliland et al., 2017; Kim, 2004; WHO, 2005), one school year of street closures around PS 184 should save something like one to 3 weeks of expected life per student. Recall than Shuang Wen serves about 650 students. The central test for this intervention, then, is if closing one block on three residential streets (Cherry, Montgomery, and South) to car traffic (involving the loss of 35–60 parking spaces) for one school year is worth ∼20 years of the student body’s life expectancy per year. (A reminder, here, that using life expectancy produces a conservative estimate of air pollution’s negative effects, because it only indirectly measures health problems, lowered quality of life, etc., all of which attend particulate matter exposure.)
Barring these streets to cars is not costless, of course. Individual drivers are likely to be inconvenienced (especially staff who park near the school—about half of the parking spaces on Cherry and Montgomery Streets are reserved for Department of Education employees during school hours—although these same staff are likely to be net winners from street closures, as they will benefit disproportionately from reduced pollution). Motorists may have to take longer routes to get where they are going or to park further from their destinations. The initial stress and conflict associated with even a nonviolent campaign may be substantial: to close these streets may provoke violence (e.g., barrier destruction) from drivers, often with the tacit or explicit approval of the police, as was reported during New York City’s open streets pilot program (Cruz, 2020). However, drivers comprise a small minority of the neighborhood around Shuang Wen (87.4% of residents in the neighborhood surrounding PS 184 do not drive to work), and this harm will be concentrated among those who are also disproportionately helped by this closure (e.g., Shuang Wen staff and parents). More broadly, the inconvenience to drivers or potential economic losses from making, say, box truck delivery routes slightly less efficient remains mild compared to the life expectancy losses at Shuang Wen—which remain an underestimate along a number of dimensions, e.g., they do not include positive spillover effects from cleaner air in the densely-populated blocks nearby (hundreds of apartments face Shuang Wen from the north, east, and southwest).
Extrapolating nonviolent street closures globally, to schools, hospitals, assisted living facilities, retirements homes, and so on would lead to hundreds of thousands fewer deaths per year and to millions of life-years saved. Clearly, the benefits are immense. But how feasible are nonviolent street closures? There are some precedents for communities blocking off streets and for this recapturing of public space to become permanent. In some cases, the state does eventually come to sanction the intervention. This usually occurs in conjunction with strong levels of local support. In New York City, where motorists moved or destroyed barriers designed to open streets to pedestrian use during the COVID-19 pandemic, carrying out the intervention in opposition to local groups like car owners and parent motorists may risk large levels of destructive backlash. However, carrying out a program in partnership with drivers and parents (or, at least, sufficiently attentive to their concerns to weaken their opposition) by, for example, setting up marked drop-off zones a half-block from the school rather than simply blocking off streets with little explanation may provoke less resistance and physical harm to the barriers and community members. It is important to remember that in the neighborhood Shuang Wen serves, 87.4% of the residents do not use a car to commute. PS 184 is surrounded by dense mid- and high-rise apartments, much of which is public housing. Car ownership is rare. But while motorists are a minority, cars are physically large, so their presence on the street is forbidding. Those who wish to close streets to cars will be at risk in street confrontations because cars are large, heavy, and can easily harm the people outside of them.
There are precedents for the seizure of public or semi-public spaces with and without state approval. I have already mentioned action by EarthFirst! beginning in the 1990s and continuing today, that nonviolently seized streets from cars and repurposed them for play, education, and nonmotorized transport. Consider too the histories of Can Masdeu and Can Batlló in Barcelona. In 2002, guerilla urbanists successfully took control of Can Masdeu, an erstwhile leper hospital. They seized Can Batlló, a former factory, in 2011. In Can Masdeu, activists used nonviolent passive resistance techniques like locking themselves to objects to resist the police that had come to evict them. For example, two squatters on one of the upper floors of the former hospital sat on either end of a long beam balanced through two windows; neither could be removed without causing the other to fall. The Spanish police were unable to remove the protestors from Can Masdeu. Three days into the siege, a judge ordered the police to withdraw. In Can Batlló, the community demanded that a former factory that had been rezoned for public use as far back as the 1970s be so used. (Years of private and public foot-dragging and the effects of the global financial crisis meant that by 2009 little had happened at the site.) Residents near Can Batlló formed the Plataforma Can Batlló, a community group that called for the conversion of the space into “a school[,] a park, affordable housing and an urgent care center[,] among other services” (Davidson, 2020). In response to further intransigence, the Plataforma began to lay concrete plans for occupying the factory, even holding conferences that described how the new space was to be laid out once it had been seized. A few days before the Plataforma’s deadline for occupation, the Barcelona city government acquiesced, yielding a block of the Can Batlló factory to the Plataforma. Both Can Masdeu and Can Batlló remain communal spaces as of 2022.
These instances of occupation, resistance, and seizure offer a few lessons. First, government support—or at least noninterference—is generally a requirement for success. While both Can Masdeu and Can Batlló began with nonstate action by community groups, both were secured for common use only when Barcelona’s city government agreed. Second, community support for or lack of opposition to were similarly essential for success. Both movements sprung from or drew support from the community and neither encountered substantial grassroots resistance. Third, the transition from non-state action to state acquiescence is where permanence is often achieved but where many movements fall apart. The interface between community action and state action often determines whether or not an intervention can transform itself into a durable feature of the community and built environment or whether it will be removed or curtailed. These points echo empirical work by Hess and Satcher which finds that community support, coalition building with local government, and the relative absence of counter-mobilization are associated with greater remediation on environmental justice issues (Hess and Satcher, 2019).
In sum: the negative effects of non-state creation of buffer zones to protect vulnerable populations from PM2.5 emissions from cars are minimal and the positive effects are probably very large (for one medium-sized school, decades of life-years saved per year of intervention). Successfully seizing such spaces is not hard in the short term, but winning state acquiescence to make them permanent remains a crucial point of transformation at which many efforts fail.
This invites the following question: If the negative effects of airborne pollution from cars are so great that certain means of nonviolent resistance are justified, are more violent forms of resistance defensible?
The sabotage of privately-owned vehicles
Private cars line the streets of cities, mostly unsurveilled and unprotected. These vehicles sit still for ∼95% of their service life (Barter, 2013). When active, they, visit substantial harm on the people who share those streets with them. In many neighborhoods in large, dense cities, the vast majority of the residents do not own cars themselves. Moreover, cars are complicated and somewhat fragile: even modest damage can cause expensive repairs and long downtimes. It is a bit puzzling, therefore, that the sabotage of privately-owned cars stored on public streets has been so rare. John Lanchester, reviewing a series of books about climate change in 2007, spelled out the logic of vehicular sabotage this way: “in a city the size of London, a few dozen people could in a short space of time make the ownership of [SUVs] effectively impossible, just by running keys down the side of them, at a cost… of several thousand pounds a time. Say fifty people vandalising four cars… six thousand trashed SUVs in a month” (Lanchester, 2007). Lanchester never revisits this glib proposal, but others have taken the idea more seriously. Two arsons against SUV dealers in California in 2003 have been linked to the Earth Liberation Front, as well as some sporadic spray-painting of vehicles from spring 2003 to spring 2005 in California, Oregon, Washington, and Arkansas (Tamaki et al., 2003). As this article was in review, a group of young climate activists began deflating SUV tires in New York City, leaving leaflets on cars explaining the vehicles’ environmental harm (Milman, 2022). The group had borrowed their leaflet and tactics from earlier campaigns by “Tyre Extinguishers” in the UK (Tyre Extinguishers, n.d.).While “ecotage” by EarthFirst! the ELF, and other groups has been well-documented in the literature (Plows et al., 2004), and the work of the tire extinguishers in the US and UK is suggestive of a broader strategy of resistance, there have not yet been any sustained, widespread campaigns to disable cars as a strategy for reducing vehicular traffic and PM2.5 exposure in cities.
This section evaluates the following intervention: concerted sabotage of private cars intended to raise the perceived and real costs of owning and using a car. Expensive cars (with an estimated Kelley Blue Book value of not less than, say, the average purchase price of a new car the year before, which was around $47,000 in 2021) could be targeted to reduce potential regressive effects of sabotage. No humans or other animals should be harmed and no safety-critical systems in the car should be targeted. The preferred forms of damage would be obvious, cosmetic, and time-consuming to fix.
Why damage cars? Sabotage is expensive and annoying. There are direct effects (e.g., a car that is a sabotaged cannot be driven that day and will emit no PM2.5), but greater effects come from that fact that as things become more expensive and annoying, people do them less. Consider the relationship between fuel price and total fuel consumption, which is an excellent proxy for total emissions. Goodwin et al. (2004) estimate the demand elasticity of fuel consumption for fuel cost as about −0.25 in the short term and as much as −0.64 in the long term (≥5 years). This elasticity estimate implies, making a few simplifying assumptions, that for every 10% fuel prices rise, total fuel consumption (and therefore emissions) falls by 2.5% in the short run and 6.4% in the long run. Much like rising fuel prices, sabotage or the credible threat of sabotage represent additional costs to car ownership and use. If a sustained campaign of sabotage against private cars could raise the (real or perceived) cost of driving by, say, 20%, prevailing elasticity estimates imply that vehicle usage and therefore emissions could fall by something like 4–6% in the short run and 11–14% in the long run. The resulting reductions in PM2.5 pollution would, as we have seen from Apte et al.’s work above, save on the order of hundreds of thousands of lives per year. The sabotage of private cars, then, seems justified on Singerian grounds if on no other: if you were walking by someone slowly drowning a child, and you had to ruin their suit (or their tires) to save the child, costing them a few hundred dollars, surely you’d be justified in doing so (Singer, 1972).
But does sabotage work? Kadivar and Ketchley (2018, henceforth K&K) attempt to tease out the effect of “protest defacement,” which includes sabotage, on the likelihood of democratization movements’ success. They helpfully distinguish sabotage from other forms of unarmed collective violence (like riots) and from nonviolent resistance. K&K find that protest defacement is positively correlated with political liberalization, at a slightly higher degree of confidence than is found for riots (see Kadivar and Ketchley, 2018: 7–8). But K&K are studying democratization movements. These movements’ goal is regime change. This paper is concerned with a more specific objective (air pollution harm reduction). This fact introduces a few important differences, one of which is that democratization movements often undertake “protest defacement” as a form of moral communication (although sometimes, to be fair, sabotage is undertaken as ad-hoc self-defense). As a result, we can mostly interpret these findings as reassurance that sabotage probably doesn’t risk an obvious and crushing backlash, but not necessarily that sabotage for self-defense or harm reduction is similarly associated with increased levels of success. Finally, the tactics studied by K&K are generally directed at government (or, failing that, corporate) symbols and infrastructure rather than individual citizens’ personal property. This difference introduces considerable complication.
Partially encouraging findings from Kadivar and Ketchley notwithstanding, large-scale car sabotage is likely to cause several undesirable rebound effects. Chief among these: increased consumption for new cars and parts to replace those that have been damaged, increased policing and securitization of public streets, the risk of collapsing community support, and violence’s tendency to make more of itself. The first effect is straightforward: more people, especially in the short run, will choose to have their cars repaired or to purchase new ones. This will cause increased emissions in the form of increased production and transportation of new cars and parts. These rebound effects are unlikely to swamp the eventual benefits of decreased car use, but they are not negligible. These effects also reinforce the importance that the campaign of sabotage be sustained: if the campaign is one of sporadic, ELF-style arson and vandalism, this may produce a worst-of-both-worlds situation in which extra emissions from repairing the sabotage are absorbed along with scant reductions in car use.
Sustaining a campaign of sabotage, however, brings up the second rebound effect: increased securitization and policing of public streets by both private and public actors. It is not especially likely that motorists and police departments will stand idly by as dozens of cars are systematically disabled night after night, month after month. Whatever set of responses unfolds—police violence, mass surveillance, curfews, firearm purchases and use, the erection of physical barricades on streets—is likely to cause significant harm to the people living in those neighborhoods, and these effects are likely to be first and worst for poor people, people of color, and those who are housing insecure. A campaign of sabotage against private vehicles is also likely to increase support for ‘law-and-order’ policies, which brings up the third potential rebound effect: a collapse of community support.
It is not the case that increased levels of violence as a rule push people toward emancipatory or revolutionary ends. There is suggestive work—including by Kadivar and Ketchley (2018) and Ansin (2020), both discussed above—that certain forms of unarmed violence can be as effective or more effective than purely nonviolent means and markedly more effective than armed violence. However, the idea that violence can be controlled once started is one of the sirens’ greatest hits. Violence most reliably produces at least two things. The first is, as García-Ponce and Benjamin Pasquale write, higher levels of trust in police and state actors. Exposure to violence, they find, raises “reported levels of fear,” which leads to higher levels of “support for enhancing local political order,” especially in the form of policing (García-Ponce and Pasquale, 2015). The second thing violence produces is more of itself. If unarmed sabotage, like damaging SUVs, were guaranteed not to start an escalating cycle of violence, it would be easier to recommend. But no such guarantee exists. As history never tires of demonstrating, violence can quickly tip toward reactionary or repressive ends rather than liberatory or harm-reductive ends.
Finally, there is the risk of collapsing community support. Though I have chosen to center harm reduction and self-defense rather than moral communication as the central features of campaigns of resistance to air pollution, acts of self-defense do communicate moral concerns. They do so necessarily and by virtue of their public nature. If acts of self-defense are received as inimical by the surrounding community, they risk backlash. Kadivar and Ketchley give empirical reasons to think of certain forms of sabotage as effective political acts. But their findings are mostly about defacing symbols, government sites, and corporate property. A sustained campaign of sabotage against motorists’ personal property risks a collapse in community support for the campaign of self-defense. Such a collapse risks, of course, the campaign itself, for logistical reasons if no other—it is much harder to defend yourself in a setting where you are despised than one in which you are regarded neutrally or favorably. And it is much harder for a reviled movement to translate its successes into more permanent forms of state protection, should it find such translation necessary. Elements of moral communication, therefore, cannot simply be discarded even in campaigns for harm reduction: even when you are defending yourself, it matters what people think of your defense.
We are left with three stubborn facts in mutual tension: the practice of sabotaging cars, if sustained and scaled, could have significant positive effects. Moreover, there are some findings in the social movement literature that unarmed violence, like “protest defacement,” which includes sabotage, is positively correlated with democratization. However, widespread sabotage of personal property has few precedents and may provoke backlash effects that would limit or reverse its potential successes.
Conclusion
Air pollution kills people in astounding quantities. However, air pollution’s effects are not limited to human mortality. Widespread air pollution harms forests, bees, crop yields, solar panel efficiency, reduces birth weights and lifetime earnings, increases retinal damage, asthma, and rates of central nervous system diseases like Parkinson’s and Alzheimer’s, to say nothing of its lethal effects on the heart and lungs (Genc et al., 2012; Isen et al., 2014; Kaldellis and Kokala, 2010; McLaughlin, 1985; Thimmegowda et al., 2020). It even increases brain inflammation in dogs (Manisalidis et al., 2020). In reality, air pollution’s death toll only caps the iceberg of its effects. This article has used mortality as its primary metric because deaths are important in themselves and because they serve as a proxy for air pollution’s myriad harms. The positive effects of mitigating air pollution estimated by this paper are therefore conservative.
I have argued that air pollution is the major under discussed political-engineering problem on Earth at present. As an engineering challenge, large-scale reductions in air pollution are surmountable: move pollution sources further from people, reduce the number and severity of pollution sources, and filter the air where feasible. But air pollution is also a political object, created by a matrix of policies and choices, commissions and omissions. The US has achieved modest but important air pollution mitigation through state-sanctioned channels (Isen et al., 2014). Because air pollution is causing ongoing, irreversible, and severe damage, this paper has considered two extrastate responses that fall on different sides of the nonviolent/violent divide.
I argue that the nonviolent establishment of buffer zones around the physical sites where vulnerable people gather (e.g., schools, hospitals, retirement communities) has enormous protective effects. We can learn about its prospects for success or failure from past community and activist action: establishing car-free zones is by no means an assured success, but certain passive resistance tactics, when they have substantial community support, can win state acquiescence and therefore some measure of durability. I argue that a campaign of sustained sabotage against combustion-engine vehicles in cities may have significant protective effects, although not as large as the establishment of buffer zones. There is empirical evidence that sabotage can be an effective movement tactic. However, this evidence comes mostly from democratization movements. The effects of widespread sabotage against individual personal property are unknown. Movements that sabotage personal property may struggle to win community support or state acquiescence relative to nonviolent resistance, or they could be more effective. More work is needed on this urgent question.
The efforts described in this paper, even if carried out in every major city on the planet, would constitute only a modest start against a deadly opponent. Substantially more work, especially across disciplines like engineering, political science, and urban planning, is needed to understand what sort of interventions protect against air pollution, how these interventions can be carried out within and without the state, and how communities can act directly to protect their most vulnerable members.
Footnotes
Acknowledgements
The author’s thanks are due to C. McNeil, W. X. Tchou, T. Vasko, and two anonymous readers.
Declaration of conflicting interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
