Not just war: Violence in climate protest

Abstract

This paper argues that reductivism, the dominant form of contemporary just war theory, cannot justify violent climate protest, despite Andreas Malm's suggestion that just war theory might help confront the challenge of the climate crisis. The argument has three stages. It begins by showing both the relevance of Malm's attacks on the climate movement's strategic pacifism to questions about the appropriateness of violence in climate activism in political philosophy and the dominance of reductivism in contemporary philosophical discussions of violence. It then shows that reductivism's individualism will make it difficult to establish liability for climate harms substantial enough to be met with defensive force. Finally, it argues that violent climate protest is unlikely to be necessary to protect individuals, since it will not effectively reduce individual vulnerability to climate change. The collective character of the threat of climate change is in tension with reductivism's commitment to justifying all violence in terms of individual defensive rights.

At COP21, 196 parties signed the Paris Agreement, pledging to limit climate change to 1.5 °C. The 1.5 °C target is partly a response to a concern about ‘tipping points’ in the Earth's climate, understood as thresholds beyond which abrupt and irreversible warming occurs (Lenton et al., 2019; Yangyang & Ramanathan, 2017). Tipping points include the collapse of Arctic winter sea ice and the mass bleaching of coral reefs. It is often claimed that the 1.5 °C target requires the global economy to be net-zero carbon by 2050 (see for example IPCC, 2018). However, this is misleading in at least two important respects. First, the emissions scenarios that underpin the claim that net-zero by 2050 is compatible with 1.5 °C or even 2 °C are qualified in response to uncertainty: they standardly hold that net-zero by 2050 leaves a one-third chance of breaching the temperature target (e.g., Doorman, 2022: 123; Moellendorf, 2022: 60). Second, the net-zero by 2050 scenarios compatible with 1.5 °C all assume the largescale deployment of Negative Emissions Technologies (NETS). Building the deployment of NETS into climate models allows global emissions to exceed the atmospheric concentrations of CO₂ equivalent compatible with 1.5 °C, since the technologies would, in principle, allow the excess to be removed. One problem with this assumption is that NETS are currently in development and unproven at scale; another is that exceeding the atmospheric concentrations of CO₂eq compatible with 1.5 °C risks crossing the tipping points that motivated that temperature target in the first place (Doorman, 2022: 123–130).

Even if we set these worries about risky and unrealistic assumptions aside, serious problems for climate politics remain. It is extremely unlikely that the global economy will transition to net-zero by 2050. According to some prominent forecasts, the carbon budget for 1.5 °C has already been exhausted. ¹ Indeed, on current policy trajectories, predictions suggest that between 2.6 and 2.9 °C warming will occur by the end of the century (Climate Action Tracker, 2023). Global emissions continue to rise, reaching record levels in 2022, over thirty years after the Intergovernmental Panel on Climate Change (IPCC)'s First Assessment Report. Major economies – those with the capacity to most substantially reduce emissions – continue to subsidise and licence fossil fuel exploration, despite this being incompatible with keeping warming to 1.5 °C (International Energy Agency, 2021: 21; Tong et al., 2019). Failing to keep to 1.5 °C risks catastrophe. At temperatures of 2 °C, an additional 65 million people are exposed to exceptional heatwaves, and plants, insects and amphibians lose a third of their climatic habitat (Lynas, 2020; IPCC, 2014). At 3 °C, a temperature level the Earth has not experienced for three million years, there could be a sea level rise of up to 2 m, a 500% increase in drought and significant yield declines in staple crops (Lynas, 2020).

The scale of the threat of climate change, and the failure to take anything like the steps required to mitigate it, has led some to consider how climate injustice might be resisted. Climate protest has become an increasingly prominent feature of the politics of many states. A striking feature of such protest is that it has remained almost exclusively nonviolent. Although activists have blockaded major transport links, disrupted major cultural and sporting events and engaged in school strikes, they have generally avoided confrontation with law enforcement or causing direct harm to persons or property. But if policy remains so disconnected from the escalating levels of climate harm, it is not clear we can or should expect climate protest to always remain so committed to nonviolence. Indeed, prominent voices in the climate movement have recently called for an escalation in tactics. In his influential How to Blow Up a Pipeline, Andreas Malm (2021) argues that the climate movement must abandon its commitment to pacifism and embrace limited forms of violence. He argues that climate activists should damage or destroy property linked to climate harm through targeted sabotage. It seems like his arguments have inspired action: a group known as the Tyre Extinguishers, for example, have engaged in the widespread sabotage of SUVs, a tactic outlined by Malm (BBC, 2024). His work has attracted considerable debate, including in prominent public venues (Butler, 2021; Kornick, 2021) and in the literature in political theory (Arridge, 2023; McLaughlin, 2026; Scheuerman, 2022), as others have considered whether climate protest might escalate along the lines he suggests.

This paper uses these developments as an invitation to consider the status of violence in climate protest. We begin by showing that the issue of protest has, until recently, remained rather peripheral in debates about climate justice in analytic political theory. We suggest that treatments of climate justice often make assumptions about the nature of our duties toward climate and the dynamics of political change that allow them to avoid difficult questions about the limits of acceptable protest. The intransigence of climate politics leaves these assumptions open to challenge. In turning to Malm's claim that protesters are permitted to engage in more directly confrontational forms of action, it makes sense to look to just war theory, whose central subject is the justification of political violence. In the next section, we introduce the currently dominant strand of just war theory, reductivism, which focuses on individual defensive rights. We argue, however, that reductivism is unable to ground permissions to use violence in climate protest, for two main reasons. First, reductivism needs to hold individuals accountable for harms to other individuals in order to justify using defensive violence against them or their property. But we argue at length that it is difficult to connect individual emitters, even as members of collectives, to climate harms in ways that would render them liable to defensive force. Second, we argue that violent climate protest will be unable to satisfy the reductivist necessity condition given the possibility of adaptation and ineffectiveness of violent protest in reducing individual risk. Thinking in terms of individual defensive rights, as the reductivist must, makes it difficult to grasp the threat of climate change, which is collective and mediated through social and political institutions and infrastructure. The paper arrives at something of a puzzle: although we find it plausible that violence will sometimes be permitted in climate protest, the most prominent strand in contemporary just war theory cannot provide justifications for such violence.

Climate change and the politics of mobilisation

The literature on climate justice in analytic political theory is now voluminous, covering a range of different issues. These debates have often sought to identify ‘first-order duties’ (Caney, 2014), which tell us who ought to do what toward mitigation and adaptation, and on what grounds. Questions about the limits of acceptable climate protest do not arise in these discussions: if everyone complies with their first-order duties toward mitigation and adaptation, the threat of climate change will be averted, and there is no need for protest of any sort. Protest becomes relevant only in accounts that take the obstacles to climate policy seriously.

For example, Elizabeth Cripps has claimed that people have a duty to engage in collective action that might reduce the threat of climate change, including through political protest (Cripps, 2013; Garcia-Gibson, 2023). Although Cripps views protest as a potentially important way of driving political change, she does not discuss it in detail, for example, by specifying how to protest. Nonideal theories of climate justice that seek to address noncompliance also sometimes highlight the importance of protest (Green & Brandstedt, 2021). Simon Caney's (2014, p. 141ff) nonideal theory, for instance, claims that some people have a duty to engage in civil disobedience. He argues that participating in civil disobedience could allow people to discharge ‘second-order duties’ grounded in the urgency of avoiding the harm of climate change, by ensuring that those who are currently failing to meet their first-order duties towards climate change mitigation and adaptation do so in the future. But again, the issue is dealt with rather quickly: Caney seems to appeal to civil disobedience as a stand-in for the general way protest might press for political change, and he assumes that it will straightforwardly be permissible.

Whether or not it is plausible to assign climate protest a peripheral position in discussions of climate justice depends on how it contributes to political change. If civil disobedience and other peaceful forms of activism are the only effective ways for protest to press for more urgency in climate policy, then it is unsurprising that protest is rarely explored in any depth. The threat of climate change represents a grave injustice, which can surely be resisted through nonviolent civil disobedience. Assuming the effectiveness of such action avoids the need for difficult discussions about the trade-offs involved in adopting more confrontational tactics. It would not be surprising if this assumption had taken hold in debates about climate justice, since it enjoys wide currency in the climate movement itself. Drawing on work in social science, particularly Erica Chenoweth and Maria Stephan's Why Civil Resistance Works (2011), prominent strands in the climate movement have claimed that it is only through mass-participatory nonviolent civil disobedience that the ends of protest can effectively be promoted, and that climate activists must strictly avoid violence as a result (Extinction Rebellion, 2023; Hallam, 2019).

The problem with this position, clearly highlighted by Malm (2021), is that it relies on misunderstanding past political struggles and empirical studies of them. Chenoweth and Stephan's (2011: 12, 13) work does not establish that violence would necessarily undermine the ends of protesters, since they only aim to show that mass action which is ‘principally nonviolent’ will be more likely to achieve its goals than ‘revolution, plots (or coups d’etat), and insurgencies’ are. And indeed their analysis does suggest that violent revolutions and insurgencies, which depend on small and highly disciplined groups of dedicated ideologues, are less successful in overthrowing authoritarian regimes than broad-based movements which evaporate the state's coercive power rather than defeating it militarily. But climate protesters could in principle engage in forms of violence – such as targeted property sabotage – that do not involve taking up arms in the relevant way. ²

Indeed, Malm (2021: 57–61) argues that, far from being demobilising, violence can be necessary for the success of mass movements. As he points out, a number of the movements classified as ‘nonviolent’ for Chenoweth and Stephen's purposes in fact centrally depended on violence. The Egyptians whose massive gathering in Tahrir Square overthrew Hosni Mubarak in 2011 provide for Chenoweth and Stephan (2011: 229) ‘a particularly stunning example of why civil resistance works’. Yet they did not manage to gather in there in huge numbers ‘by offering flowers to the police’ but used ‘gas canisters, pieces of pavement and other projectiles’ to fight ‘their way through’ (Malm, 2021: 59). Contemporary newspaper reports make it clear that what Malm (2021: 60) calls ‘unarmed collective violence’ was central to the ousting of Mubarak, describing protesters breaking ‘down a burnt-out vehicle to salvage defence materials and put together catapults and slingshots’ as well as preparing to use stones and firebombs against regime supporters (Shenker, 2011). As a result, Malm (2021: 61) arrives at the opposite conclusion to that drawn by strategic pacifists influenced by Chenoweth and Stephan, challenging them to find a single case in which an entirely nonviolent movement has achieved transformative political change.

Malm's challenge may overstate the role unarmed violence plays in protest movements. However, it is clear that neither Chenoweth and Stephan's work nor empirical social science more generally show that unarmed violence is necessarily counterproductive, let alone fatal, to protest movements in the way that strategic pacifists in climate movement typically claim. As Chenoweth (2023: 57) wrote in 2023, after 12 years of exploration of their work with Stephan, empirical assessment of the role of unarmed violence specifically suggests it has ‘ambiguous effects’. Indeed, some studies of those effects ‘report several substantial immediate, medium-term, and long-term strategic gains’ from the use of unarmed violence in protest movements (2023, 65). That suggests there are ‘significant and unsettled questions about whether nonviolent discipline is as vital to the success of mass movements as was previously understood by scholars of civil resistance’ (2023, 71). At the very least, the blanket refusal to consider whether peaceful protest might be supplemented by some use of violence by climate activists by the dominant strands in the climate movement seems unwarranted.

The prevailing view in the climate justice literature also seems to be that mass-participatory peaceful protest is the only effective way to press for compliance with mitigation and adaptation duties. To the extent that this is true, its accounts of protest will be similarly vulnerable to Malm's criticisms. For Cripps (2013, 2022), for instance, individual duties toward climate change are duties to collectivise in order to become an effective agent. As we have seen, she understands protest as one among a number of ways actors might discharge this duty. Citing Chenoweth and Stephen, Cripps (2022: 165) claims that individuals can make a difference by adding numbers to a social movement (see also Táíwò 2022: 183); we cannot make a difference as uncoordinated agents acting on our own, argues Cripps, but we can contribute to building the peaceful mass movements that can make a difference.

Darrel Moellendorf (2022) provides a particularly clear statement of the view that climate protest ought to take the form of peaceful mass-participatory collective action. Drawing on Martin Luther King and his strategy of nonviolent disobedience, Moellendorf (2022: 129) aims to present a positive vision for the future around which a peaceful mass mobilisation might occur. He claims that ‘[i]t is an important practical principle of progressive political realism that the power of money in politics is best met by the power of mass mobilisation’. For a view like Moellendorf's, questions about the limits of appropriate protest are straightforward to answer. Protest is understood as a general way of increasing compliance with duties toward climate change mitigation and adaptation, and nonviolent mass mobilisation is assumed to be the most effective way to exert this pressure. Mass mobilisation does not appear difficult to justify given the stakes, and so no other protest or resistance tactics are worth considering, since they, anyway, would not effectively press for more ambitious climate policy. In this context, it is reasonable to expect participants in debates about climate justice to reject calls for confrontational and even violent forms of climate action.

Indeed, Moellendorf's emphasis on the importance of a hopeful climate politics of mass appeal leads him to a scepticism about the appropriateness of confrontational tactics. If ambitious climate action requires a mass mobilisation, it follows that climate activists should avoid tactics that make such a mobilisation more difficult. For Moellendorf, this likely rules out the type of sabotage Malm advocates, which he worries will be ‘merely symbolic’ and ‘demobilising’ for the broader movement (van Diemen, 2021). Scheuerman (2022; 2023; 2026: 98ff), a stronger critic of Malm, echoes this point. He claims that the sabotage of high-emitting property risks being futile: only mass pressure can hope to unsettle the influence of fossil fuel interests, and this requires attention to movement building.

But just as the strategic virtues of pacifism have been exaggerated in debates about climate protest, these concerns about confrontational action are overstated. Neither Moellendorf nor Scheuerman provide strong empirical grounds for thinking that mass mobilisation would be undermined by acts of limited violence by some protesters, nor do they consider the different justifications protesters might use when resorting to such action. Lai and Lim (2023), for example, have developed a fairness-based argument for uncivil climate protest. They claim that potentially violent tactics like ecotage – a form of sabotage that disrupts property that causes environmental harm – enables protesters to be more discriminating, targeting those with particular responsibility for climate injustice. ³ Relatedly, Alexander Arridge (2023) has extended Malm's argument, suggesting that activists have rights to use defensive violence against those who threaten unjust climate harm. He believes that ecotage, in particular, would in many cases represent an appropriate form of self- or other-defence, at least when it targets certain property associated with luxury emissions. If people have fairness-based reasons to engage in ecotage or defensive rights against particular people for their harmful emissions, then it would not necessarily be decisive that violence set back a mass mobilisation in the climate movement, even if we had empirical grounds for thinking that were true. Of course, these arguments will need to be scrutinised.

Here we will focus on the argument from individual rights to defensive violence, which draws on the most prominent strand in contemporary just war theory, known as reductivism. Our conclusion is that seeking to justify violence in climate protest on the basis of individual defensive rights will be extremely difficult and most probably impossible. The collective character of climate change seems to mean it cannot pose a threat in the way that reductivism requires to justify violence. Perhaps other arguments, like Lai and Lim's, or others from the burgeoning literature on uncivil disobedience (Delmas, 2020), will be able to justify ecotage or other forms of violence in climate protest. However, given the influence of just war theory and its central focus on the justification of violence, it is significant that its dominant strand struggles to comprehend the threat climate change poses. Yet this is what our arguments suggest. The most prominent version of just war theory is, we claim, systematically unable to get to grips with the threat posed by climate change.

Reductivism and climate protest

Those who have thought about whether violence can be justified in political resistance have often appealed to ideas associated with just war theory. In our context, both Malm and Arridge believe that just war theory can shed light on the distinctive normative challenges posed by climate politics. Arridge (2023: 422) claims that viewing climate justice in terms of ‘aggressors, victims, defense and attack’, as a ‘struggle between perpetrators and victims’, might enable theoretical accounts to properly engage with the persistent obstacles to ambitious climate policy. Malm (2021: 32) refers to just war theory to bring out what he sees as the moral bankruptcy of strategic pacifism dominant in the climate movement. For him, the refusal to consider when the use of violence might be justified in this context involves surrendering ‘to suffering and atrocity’. Activists should instead accept that violence can be justified in response to grave threats, as just war theory suggests, and start thinking about when and how it might be appropriate as a way of averting the extremely grave threat of disastrous climate change. ⁴ Otherwise, their claim that their pacifism is not an implausible form of moral absolutism looks insincere. As Malm (2021: 32) puts it, referencing Fabre (2015), ‘a pacifist who makes exceptions is a just war theorist’.

Malm's own discussion is primarily concerned to show that property damage can be an effective response to climate injustice, and so is not structured by a detailed engagement with just war theory and its criteria for the permissibility of violence. Apart from using it to attack strategic pacifism, Malm (2021: 107) only otherwise refers to just war theory to defend sabotage against the allegation that it is a form of terrorism, pointing out that it does not aim at direct physical harm to anyone, let alone breach the principle of non-discrimination. Arridge makes much more direct and detailed use of just war theory though, framing his arguments in support of Malm around its criteria. Nor, as we have already mentioned, is this unusual in assessing protest. For example, Pasternak (2019: 391, 389, 398ff; 2025) has influentially used this tactic in her assessment of political riots, which she defends as at least sometimes being necessary, successful and proportionate means of ameliorating the ‘pervasive, gross violations of fair social cooperation’ that continue to ‘mar at least some contemporary democracies’.

Any attempt to understand the appropriateness of violence in protest through the lens of just war theory will have to take a view on a central disagreement in that literature. As Seth Lazar notes, contemporary just war theory ‘is dominated by’ a division between traditionalism and revisionism. Traditionalists ‘aim to provide … morally defensible foundations’ for international law as it applies to war, and ‘especially the law of armed conflict’. Revisionists, in contrast, deny that many of the central features of international law governing military force, such as the entitlement of all combatants to use lethal force against each other, are morally defensible (Lazar, 2020). While traditionalism remains influential ‘outside philosophy departments’, within them it has been subjected to ‘exacting scrutiny’ and is now widely rejected. In this sense, ‘we can attribute much of the recent success of revisionist just war theory to the preponderance of moral philosophers in the field’ (Lazar, 2017: 38, 41).

Revisionist philosophers continue to disagree in various ways about when and why military violence is acceptable, but they typically agree that traditionalism's attempt to justify the law of armed conflict is inadequate at least in part because it requires treating war as in some sense morally sui generis. Instead, revisionist philosophers are usually also reductivists, who hold that war must be justified in the same way as any other kind of violence – ‘that war is regulated by the same moral principles that apply to interpersonal morality’ (Renzo, 2018: 708). ⁵ Showing that killing in war is justifiable requires showing that it would be justifiable in response to an otherwise identical threat outside of war. Justified killing in war then reduces to ‘justified acts of individual self- and other-defense’. ⁶ Our ‘fundamental rights to life and liberty’ do not disappear simply because a war has been declared, and so overcoming the barriers they present to military violence must be the fundamental question about its permissibility (Lazar, 2017: 40). For example, Massimo Renzo takes it that there is a difficulty in justifying wars to defend political self-determination because this focus on interpersonal morality demands an explanation of its value that is both sufficient to justify lethal violence and depends only on individual rights, and so cannot appeal straightforwardly to the collective goods political self-determination realises. ⁷ The ‘predominant approach to just war’ requires showing how the defence of individual rights can by itself warrant killing, including to protect collective self-determination (Renzo, 2018: 708). ⁸

Both of the two uses of contemporary just war theory to consider the justification of violence in protest mentioned earlier draw on reductivist arguments. Pasternak accepts reductivism by treating questions about defensive harm in general and just war theory in particular as continuous. Political rioting is ‘a form of defensive harm’ and so should be assessed using ‘the various constraints offered by just war theorists' (Pasternak, 2019: 386). And Arridge's account of permissible harm draws heavily on reductivist just war theorists like Jeff McMahan and David Rodin (Arridge, 2023: 415, 416; Lazar, 2017: 40, 2014: 12). Damaging oil pipelines is governed by the same principles as defending yourself against a murderous attacker. That both Arridge and Pasternak effectively treat reductivism as identical to contemporary just war theory illustrates and so substantiates Lazar's and Renzo's claims about its dominance. As such, if we are to assess Malm's call for more forceful climate protest by, as he suggests, drawing on just war theory, then reductivism is the obvious choice as the ‘predominant approach to just war’.

However, we argue that it is in fact hard to see how reductivism permits any violence in climate protest. ⁹ Reductivists generally accept ‘that justified self-defense’ – and so any permissible violence – ‘must meet four conditions’: that ‘the defender … face[s] an unjustified threat’; that there is some reason to protect ‘the defender's interests’ rather than those of whoever threatens them; that ‘the force used must be proportionate to the threat averted’; and that that force ‘must be necessary to avert the threat’ (Lazar, 2012: 3–4). Yet while climate change undoubtedly poses an unjustified and serious threat to billions of people, the second and fourth of these conditions are difficult to meet in the case of violent climate protest. The second condition, that it is appropriate to protect a defender's interests rather than an attacker's, usually means showing anyone targeted for violence is responsible for the threat it hopes to avert. But it will be difficult to connect individuals to particular climate harms to the extent that would be required to render them liable to defensive force. The fourth condition, the requirement that defensive harm is necessary, demands avoiding violence where there are alternative means of averting a threat. Whatever the merits of violent climate protest as part of a broader political strategy, it is not an effective way to avoid individual harms associated with climate change. A person will generally have other options available to them that will be more reliable and effective in reducing their vulnerability to climate impacts than any piece of violent climate activism, which is very unlikely to have any effect on their vulnerability at all.

The upshot of our argument is that reductivism will almost inevitably judge violent climate protest as impermissible. One conclusion to draw from that finding is that violent climate protest is indeed impermissible, and we do not argue here that it would be wrong. However, we think that rather than rejecting violent climate protest, it may be more plausible to reject reductivism. In our view, its commitment to justifying all appropriate violence by reducing it to violence that would be permissible in interpersonal contexts is ill-suited to a threat like climate change, which has no relevant individual analogue. While we do not have the space to develop that position here, it is nonetheless important to note that there are two responses to our argument, and that we are inclined to reject the claim that all permissions to use violence can be reduced to permissions individuals have to defend themselves rather than foreclose the possibility of legitimate violence in climate protest.

Responsibility for individual climate Harms

The first challenge for a reductivist account of violent climate protest is to provide an account of why the targets of such protest are liable to defensive harm. Reductivism sees the justifiability of all violence as depending on the permissions available to particular individuals to defend themselves or others. Yet we are individually typically only entitled to defensively harm someone if they are responsible for the threat the defensive harm means to avert. Notably, for McMahan (2009: 158), ‘moral responsibility for an unjustified threat is the basis of liability to attack’. A runaway trolley hurtling towards someone could kill them if it hits them. The risk it poses passes the threshold to license defensive harm. However, whether and to what extent anyone else is liable to that defensive harm depends on their relation to that risk. It would be wrong to push an innocent bystander in front of the trolley to divert it, whereas doing the same to the attacker who deliberately set it loose would not be.

The idea that responsibility is necessary for liability to defensive force is reflected in defences of violent climate protest and activism, which tend to focus on defending the destruction of property owned by agents seen as responsible for climate harms. Arridge (2023: 409ff), for instance, argues that ecotage is appropriate when, along with being effective and proportionate, it is targeted at ‘[c]limate aggressors’ who ‘are culpably and wrongly harming many present and future people’ (see also Malm 2021: 84–97). As such, we will begin by considering whether we can establish liability to defensive force on the basis of individual responsibility for discrete climate harms. Various reductivists have challenged McMahan's identification of liability with responsibility, and we will engage with their arguments in the ‘Alternative grounds for liability' section. Still, the centrality of responsibility both in debates about climate activism and in reductivist just war theory makes it an appropriate place to start.

Connecting particular individuals to discrete climate harms is very hard, at least to enough harm to warrant a violent response. Holding even high-emitting individuals responsible for specific climate harms seems extremely difficult. The emissions caused by the most destructive lifestyles are, at an individual level, dwarfed by the size of the problem. For instance, in 2015, the top 0.1% of the global income distribution produced 4% of annual global carbon emissions (Kartha et al., 2020: 9). That implies that on average, those roughly seven and a half million people each emitted about half a millionth of a percent of annual global carbon emissions. Even if the 0.1% of the 0.1% stood in the same relation to overall emissions in that group as the 0.1% does to global emissions, those roughly seven and a half thousand individuals would have, on average each emitted one fifty-thousandth of a percent of annual global carbon emissions. Only a proportion of global emissions orders of magnitude higher could be connected to particular climate harms.

The climate system is incredibly complex and chaotic. Greenhouse gas emissions are absorbed by different parts of the biosphere for varying lengths of time, and the gradual accumulation of many of these emissions in the atmosphere, as the capacity of the global emissions sink is exceeded, drives changes in weather patterns which cause harm as they interact with social sources of vulnerability. In that sense, ascribing responsibility for particular climate harms to particular emissions is vexed. Even after the fact, once particular harms may be probabilistically linked to climate change in general, the complexity of the system makes attributing particular harms to individual-level emissions problematic. When those emissions are such tiny fractions of the global total, we cannot be confident they made any difference at all. ¹⁰ And if we cannot be confident that tiny fractions of total emissions make any difference, we cannot hold individual emitters liable to be attacked to prevent climate harm. The six seconds worth of annual global emissions that the 7500 wealthiest people in the world each individually produced in 2015 simply are not significant in the relevant sense. However, additional luxury emissions may be impermissible, they do not seem a plausible basis for liability to defensive violence as a means of preventing particular harms to particular individuals. Liability for defensive harm cannot be based on infinitesimal contributions to unpredictable outcomes generated by the complex and chaotic interactions of billions of people's uncoordinated actions all across the world.

One way to try to get around this problem of individual insignificance is by giving up on linking specific harms to specific emissions. Instead of holding people responsible for the consequences of their emissions in particular, they could instead be held responsible for the proportion of the expected harm caused by climate change that matches their contribution to it. All we need to know then is the expected harm for a given level of aggregate emissions and the fraction of total emissions an individual contributed. They are then responsible for that fraction of the expected harm. The huge amount of harm climate change will do means that this suggests that the harm done by individual emissions is non-trivial. For example, Broome (2019: 111) estimates that the lifetime emissions of a well-travelled academic will typically do around $40,000 worth of discounted damage while they remain in the atmosphere. Similarly, Nolt (2011: 9) argues that the lifetime emissions of the average resident of the United States can be expected to kill or seriously injure two people over the next millennium.

However, calculations like Broome's and Nolt's do not show that any individual has any defensive rights against climate emissions, even if we grant that it makes sense to sum risks spread across billions of people and hundreds of years into one figure. The risks all those billions of people face across hundreds of years are individually miniscule, far smaller than would justify violent self-defence. Nolt's (2011: 8) calculation implies a 1 in 50 billion lifetime chance of death or serious injury for each of a population of 100 billion. To put it another way, this is a risk from which we would be very unlikely to see any deaths or injuries if it were imposed on every single human alive today. Even the 2015 emissions of the 0.1% only imply a 1 in 5 billion risk (Nolt, 2011: 5; Kartha et al., 2020: 9). There cannot plausibly be a right to violent self-defence against such a risk. Broome does not give details about the basis for his calculation, but given that billions of people face risks from climate change, the additional risks caused by even a lifetime's worth of emissions must be tiny. Just confining the risk to the roughly eight billion people currently alive, the expected harm to any one of them of $40,000 of climate damage is 0.0005 cents. No one has a right to use violence to avoid an expected loss of one two-thousandth of a cent. As McMahan (2021: 237) says when discussing summing the expected harm of individual emissions, individual losses of that size do not give rise to any ‘morally significant complaint’. Broome's (2019: 111ff) repeated invocation of the death of a child caused by a decision to go for a drive on a Sunday whose total expected cost is only $1 is then somewhat hyperbolic.

Individual responsibility for collective climate harm

The only way to connect individuals to climate threats large enough to warrant defensive harm on reductivist grounds is through the threats posed by groups of emitters. But linking individuals to unjustified threats posed by collectives can be difficult. It is easiest in the case of highly structured or tightly connected groups, whose members cooperate to achieve particular goals. All the members of a gang of bandits that kill and rob travellers are responsible for all those murders, and could be attacked to prevent any of them. That is because whatever they do, they do together: their individual acts are guided by and make sense in terms of decisions about what to do as a group (Kutz, 2000: 113ff). Passing responsibility from the group to its individual members becomes much harder without that cooperation (Kahn, 2024: 36–39). Young's (2011: 99ff) case of Sandy, a single parent who becomes homeless, is illustrative of the difficulties involved in connecting individuals to harms caused by uncoordinated collectives. Sandy's landlord sells her apartment to developers. She needs to find a new apartment that is convenient both for her work and for her children's schools and childcare, but can no longer afford to live in an area that has easy access to each. As Young points out, there does not need to have been a shared plan, let alone one to treat her wrongly, for Sandy to experience an injustice. Landlords respond to market incentives and their own financial pressures rather than to a plan to marginalise and exclude her personally or even single parents more generally. Yet without such a plan, it is not straightforward to see how any individuals could be personally responsible for Sandy's plight, which none of them alone could have caused and which may not have been easy to foresee; after all, many people in Sandy's situation do not become homeless. And if none of them are personally responsible for Sandy's homelessness, a reductivist account of permissible violence cannot easily hold any of them liable for defensive harm in order to avert the threat it poses to her interests.

Reductivism avoids the difficulties of connecting individuals to wrongs by collectives in its central case of war in two ways. First, many of those liable to be attacked in war either use or order violence, and so are responsible for it directly. At least when that violence is unjustified, they are straightforwardly liable to defensive harm on grounds of responsibility for an unjustified threat. An officer who orders a subordinate to summarily execute a prisoner is as, if not more, responsible for the execution than the soldier under their command. Neither need to be connected to a larger collective outcome or project to be held responsible for an unjustified threat. Their own acts are unjustified threats to individuals, independently of connections to any other acts. As we have already seen, climate change does not involve individuals directly posing significant risks to other individuals, and so this way reductivism makes combatants responsible for unjustified threats is therefore not available to advocates of potentially violent climate activism.

Second, wars are fought by structured groups that orient their members’ acts around their objectives and plans to achieve them (Barzagan, 2013). All the members of an invading army are participating in an invasion: they are where and are doing what they are because their presence is and actions are part of a plan to capture and hold another state's territory. If that plan aims at or is organised around systematic rights-violation, then the invaders pose an unjustified threat, whatever they may be doing individually. Digging a supply trench or managing the flow of messages between different units is as much a part of the invasion as capturing a defensive position. Climate change is obviously not caused by an organised military directing the actions of individuals in line with clear objectives, and so work would need to be done to extend such an argument to the case at hand. Still, this is a more promising route for the reductivist, and indeed a number of accounts have sought to extend the logic of liability through participation to members of groups other than militaries. For example, Pasternak (2025: 121) argues that police officers can be appropriate targets for rioters as members of police forces in societies marred by ‘serious domestic injustices’ that almost inevitably involve systematic ‘racism, discrimination and disrespect’ from law enforcement. She claims that officers can legitimately be targeted, whatever their individual responsibility for that racism, discrimination and disrespect. What matters is their participation in a collective which cooperates to achieve goals which involve inflicting serious wrongs. ¹¹

Whether we can apply this analysis to climate change depends on how we understand the social processes from which that threat emerges (see e.g., Kahn 2024). Climate change could be understood in the terms of the model of structural injustice offered by Young and exemplified by the processes which leave Sandy homeless. That would make it impossible to hold individuals personally responsible for the threat posed by climate change, and so block arguments for liability based on either individual responsibility or participation. That model of climate change as a pure tragedy of the commons is not very plausible though (Aklin & Mildenberger, 2020). For instance, it does not take proper account of the way in which powerful interests have used all the means at their disposal to illegitimately block and stymie attempts to slow and stop the build-up of greenhouse gases, both domestically and internationally. Relatedly, powerful interests have only intervened to protect their ability to continue to profit from emissions because there are organised groups and institutions capable of limiting, reducing and, in the end, eliminating it.

Both the groups which have intervened to prevent effective action on climate change and the institutions they have prevented from acting are structured agents that organise their members’ activities to achieve their goals. Their members can be held responsible for what they do. Insofar as, for example, large fossil fuel companies like Shell, BP, Exxon Mobil or Aramco have exacerbated and accelerated the climate crisis, we may be able to hold their employees and shareholders accountable for the harm they have done. That accountability may serve as the basis for defensive harm. Equally, members of the wealthiest, most polluting societies might be responsible for the failures of those societies to reduce their emissions as quickly as they could and to cooperate to help others do so too. Of course, more would need to be said about this possibility, and it is noteworthy that Arridge (2023: 410ff), for example, does not develop or appeal to an account of individual responsibility for collective or corporate action in his justification for violent self-defence in climate protest. Instead, he argues only that corporate agents are responsible and so bypasses the issue of whether any of their members are liable to defensive harm as a result of their acts. However, assuming that we can move from the collective to the individual in that way, there are still two serious challenges a reductivist defence of violent climate activism and protest must overcome.

First, and building on the previous discussion, the harm done by the groups to particular individuals may not be sufficient to justify defensive harm. Even collectives associated with high emissions – wealthy states, for example, or international corporations – emit fractions of total emissions, and reductivist justifications of violence would need to show that these fractions posed substantial risks to particular individuals. In fact, that threat to particular individuals may be sufficiently small that even significant damage to their assets or disruption to their activities would be disproportionate. Nolt's calculations imply that producing 10% of total greenhouse gas emissions imposes a risk of around 1 in 300 of serious death or injury on every individual for the next millennium. This is slightly higher than the risk of dying as the result of a road traffic accident implied by the death rates in England and Wales in 2022 (Office for National Statistics, 2023). That risk, influenced by political considerations just as the harms of climate change are, does not seem to warrant significant harm even to property. Vandalising cars to reduce it would not be legitimate, let alone blowing up pipelines or other pieces of fossil fuel infrastructure, which would involve more serious economic losses and potential harm to individuals. Direct harms to individuals foreseeably risked by unarmed collective violence of the sort Malm appears prepared to countenance as part of efforts to block, stall or disable fossil fuel infrastructure are certainly ruled out.

Second, even setting aside this worry, responsibility for wrongs committed by a group does not necessarily mean full responsibility for those wrongs. For instance, the extensive literature on citizens’ responsibility for state wrongdoing tends to focus on explaining and justifying compensatory liabilities it assumes will be shared across the population as a whole (Jubb, 2014; Pasternak, 2021; Stilz, 2011). It does not seek to show that citizens can be treated as individually responsible for the full weight of wrongdoing, seeing the lesser task of justifying compensatory liabilities as already challenging enough. For example, Pasternak (2021; 2025: 157) explicitly excludes ordinary citizens from liability to direct attack in riots on grounds of participation in serious domestic injustice, despite elsewhere defending the claim that they typically have compensatory duties for state wrongdoing. And citizens seem much more directly involved in the policies of their own state towards their fellow citizens than they do in climate change, which is not and cannot be the responsibility of any one state. Indeed, many reductivists remain sceptical that ordinary civilians are liable to defensive harm in war despite the large and direct threat to very many individuals that wars pose (Fabre 2012: 76; McMahan 2009: 225ff).

Alternative grounds for liability

So far, we have argued that the reductivist frame generally adopted in contemporary just war theory is ill-equipped to make sense of the production of climate harm. Although climate change will lead to enormous harm in the aggregate, it is not an injustice done by one individual to other individuals. The obvious way then to hold individuals liable to defensive harm to prevent climate change is through their participation in collectives. There may be collectives large enough to have made contributions to climate change large enough to avoid being dwarfed into insignificance by the size of the problem, and it is plausible that participants in them may bear responsibility for the harms they inflict. It is not at all clear, though, that responsibility should be understood as liability to defensive harm to prevent a direct risk of harm to particular individuals. A reductivist account of violent climate activism based on participation in collectives would have both shown that individuals participate in collectives large enough to make a difference to individual climate harms and that participation was an appropriate ground for not just compensatory liability, but liability to defensive harm. Meeting that challenge seems not just difficult, but one that no reductivist justification of violent climate activism has yet seriously attempted.

Reductivists may instead turn to other grounds for liability to demonstrate that there are appropriate targets for defensive harm against climate change. One such alternative justification is that attacks on innocents would be a lesser evil (see, for example, Fabre 2012: 253 or Tadros 2012: 268). Indeed, despite McMahan's (2009: 114) emphasis on the significance of responsibility for liability, he does allow that deliberate defensive harms to innocents can sometimes be justified, if they avert a ‘substantially greater’ harm than they cause and so override the ‘presumption against the intentional infliction of harm’. Once we raise the stakes sufficiently high, it seems plausible to think that responsibility for an unjust threat is not necessary for liability to defensive harm. If one of us must be sacrificed to save thousands, then perhaps that is justifiable.

We are sceptical about lesser evil justifications for violent climate protest, for two main reasons. First, accepting such justifications puts pressure on the principle of discrimination, which requires that military violence be directed at properly military targets, and we doubt that many who are prepared to justify violent climate action would be prepared to accept direct attacks on civilians in war. Indeed, as we have noted, Malm (2021: 107) appeals to exactly that principle to defend ecotage against the allegation that it is terrorist. Second, even if justifications of violent climate activism were prepared to undermine the principle of discrimination, they would run into our arguments about necessity. Those arguments loom even larger for lesser evil justifications for violent climate activism because of the need for defensive action to avert ‘substantially greater’ harms.

Victor Tadros has separately questioned whether liability to defensive force requires individual responsibility for an unjust threat in the way McMahan suggests. On Tadros’ view, a variety of relations to the threatened harm, the aggressor intending to commit it or its potential victim may be enough to generate liability to defensive harm. Part of his motivation is to provide a more substantial role for considerations of fairness in our thinking about liability. For instance, Tadros (2014) argues at length that it can be unfair for those who benefit from wrongful threats – such as the citizens who have enjoyed the protection provided by a military that is now committing injustice – to be shielded from liability for them. Tadros (2016: 137ff) also claims that the unfairness of acting to benefit oneself in ways which impose risks on others can generate liability to defensive harm against those risks, even when seeking those benefits is permissible. However, Tadros’ arguments from fairness do not avoid the arguments we developed above. This is because they still require a link between an individual and a threat they have caused. The first of Tadros’ arguments redistributes liability from those responsible for a wrongful threat to those who benefit from it, while the second imposes liability on those who benefit innocently from imposing risks on others. But our arguments, if successful, show that there is no link between individuals and climate harm from which to derive liability in the first place.

There are yet further arguments which do not ground liability for defensive harm in either individual responsibility for or participation in the collective imposition of a wrongful threat. For example, Pasternak (2025: 161ff) has argued that Samaritan duties of assistance can be coercively enforced by defensively harming those who have them in ways which assist their beneficiaries. An individual could force another individual to save a child from drowning in a pond if they could not rescue the child themselves. Again, we doubt this argument applies to possible targets of violent climate activism. The enforceability of Sarmatian duties plausibly depends on it being possible for a bystander to easily intervene to save someone from an imminent serious harm. But this is not the situation anyone faces with respect to climate change, which poses a wide range of different risks to very many people all across the globe. We suspect that this discontinuity between climate change applies to other arguments which separate liability from individual or participatory responsibility. For instance, Tadros (2016: 204ff) presents a case where an individual knowingly buys goods from someone who will use the proceeds to hire an assassin. As it happens, the would-be murderer has already raised enough money. In this case, the buyer plausibly incurs liability without either participating in a collective project or contributing to the eventual assassination. The absence of clear causal responsibility for significant wrongdoing is a matter of sheer luck there, though, and no one has been individually lucky not to contribute to significant climate harm.

Still, our goal here is not to show that there could never be a reductivist justification for liability to defensive harm to prevent climate change. Doing that is likely to be impossible, given the range of possible justifications for liability to defensive harm open to reductivists. It is instead to show that the nature of climate change and its production pose systematic challenges to reductivist justifications for liability because of their focus on individuals and their defensive rights. Climate change is not an attack by individuals on other individuals, and starting by wondering about what rights individuals might have against other individuals who attack them does not seem to us a helpful way to get to grips with how it might be appropriate to try to prevent its worst effects.

Climate change and individual risk

Our discussion so far has primarily focused on how climate change is caused. The way that individuals contribute to and participate in the processes that cause harmful climate change makes it difficult to connect them with those harms sufficiently strongly to make them liable to defensive harm in order to prevent it. In this section, we turn to questions about the effectiveness of violent climate activism. Justifying such activism as a form of defensive harm requires showing that it is necessary to prevent unjust harm: that it would, in the first place, have a chance of averting the threat, and additionally that it would be appropriate in light of the alternative options. We argue that it is very unlikely that violent climate activism can meet the reductivist necessity condition, since it will struggle to overcome either of these challenges. The mechanisms which cause climate harm to individuals mean that violent climate activism will rarely be the only intervention that might avert the threat, and in fact, when considered from the perspective of an individual, it will not be an effective way of reducing the threat at all.

The necessity requirement for the justification of defensive harm is usually understood as demanding that any defensive force is the least bad option, once the likely costs to those affected are weighted by both their likelihood and responsibility for the unjust threat that makes at least a significant risk of harm unavoidable (McMahan, 2018: 435ff). Although it would be proportionate to break an attacker's arm in order to stop a murder, it would be unnecessary and so illegitimate to do so if the attacker could instead be stopped by being tripped and then restrained. It is then crucial for reductivist justifications of violence that defensive agents do not have options available to them that are nonviolent. Violence is usually itself harmful, and so successfully averting a threat non-violently will usually mean causing less harm overall than averting it violently. For example, usually someone should only defend themselves against an attacker if they cannot successfully flee (Frowe, 2014, p. 91). And that may be true even when the nonviolent option would lead the defender to incur some costs: fleeing might still be required if, say, the attacker was threatening to steal property the defender would have to leave behind.

For our purposes, we need to know whether individuals have options available to them that would reduce their risk of suffering climate harm, which do not involve imposing the costs associated with violence on others. Let us look first at the alternative options available to individuals seeking to reduce their risk of climate harm. It is true that individuals cannot by themselves protect local infrastructure and social practices, and so some forms of adaptation, if they are possible at all, must necessarily occur at the collective level. But given the social mediation of climate harm, there might nonetheless be many ways for an individual to reduce the risk they face without displacing the same risk onto anyone else. Individuals might adapt by redesigning their homes or relocating, for example, or they might be able to reduce their vulnerability by taking up opportunities created by large-scale adaptation projects managed by governments. Indeed, in some cases, individuals might be able to avoid costs that collectives cannot. Whereas individuals might be able to change jobs at little cost to themselves, large-scale redeployment programmes at the state level can be enormously costly, demanding resources that might otherwise have been used elsewhere. Noting the possibility of individual adaptation is important, since it shows that existing accounts of violent climate protest construe the necessity requirement too narrowly. For example, Arridge (2023: 420) compares averting climate harms through violent protest with averting them through continuing to protest peacefully, and so ignores the fact that they might be protected against more directly through adaptation.

Of course, the ability of individuals to adapt to climate change is often limited. Individual adaptation measures like moving house or changing jobs are, in fact, often very costly for those involved. Nor can individuals easily reduce their risk by making use of collective adaptation programmes, as these initiatives are in practice often absent altogether or otherwise inadequate. More, it will be impossible to adapt to mitigate or prevent some of the harms of climate change, either because of their sheer size or the uncertainty surrounding them – or both. Such is the threat posed to large-scale social organisation by warming of 3 °C, for example, that there is significant uncertainty about the social and economic effects of this level of climate change. We might therefore wonder whether individuals really do have a range of adaptation options available to them that would effectively reduce their risk of climate harm.

We do not dispute that individual climate adaptation is often difficult and costly and sometimes impossible. But we are sceptical that a lack of effective options implies individuals could claim a defensive force necessary on reductivist grounds. This is because of the other condition introduced above: in order to claim that violent climate activism was necessary to reduce an individual's risk of climate harm, it would have to be the case that violent climate activism reduced that individual's risk of climate harm. However, violent climate activism is simply not an effective way for an individual to reduce their risk of climate harm. This is not to say that violent climate action would be politically ineffective in general. Ecotage, for example, might lead to change through the pressure it exerts on political actors, or through the way it dissuades investors from funding new fossil fuel projects. But important as this might be as part of a collective political struggle, it would not be a very direct way to defend individuals against climate change. An awful lot of pipelines would need to be blown up to make any discernible difference to any one person's risk of climate harms.

Indeed, Edmund Flanigan and Ten-Herng Lai (2026) recently published a defence of violent climate activism which takes as its starting point its futility as a form of defensive harm. ¹² They argue that violent climate activism is instead justified as a fitting expression of the unacceptability and injustice of climate change and the harm it does. We do not comment on the plausibility of that expressive justification for violence, which deliberately turns away from the reductivist just war theory and the ethics of defensive harm. Our point is just that if necessity requires an individual to respond to a threat by taking the action that leads to the least morally weighted harm to individuals, it is hard to see how they would be permitted to pursue an option that would not directly reduce the threat to individuals but would cause harm to others.

Conclusion

To meet a reductivist criterion of necessity, violent climate protest would need to be better at protecting individuals against climate harm than adaptation or other changes in policy and behaviour, which provide direct protection against climate harms rather than aiming to reduce emissions over time through a combination of political pressure and direct action. To identify targets reductivism would accept are liable, violent climate protest would need to show that those targets are responsible for grave enough harms to particular individuals through their contributions to climate change to lose rights against being attacked or having their property destroyed or damaged. This is challenging not only because individual emissions do not cause severe harm to other individuals, but also because an individual's liability to defensive force as part of a collective will be discounted according to the nature of the contribution they make to a shared project. Establishing either liability or necessity in this case seems difficult. Establishing both seems extremely unlikely.

Unless liability and necessity can be established, then the predominant position in contemporary just war theory, reductivism, seems to prohibit all forms of violence in climate protest. That leaves open at least two possibilities. One is that violent climate protest is impermissible. But as we stated at the outset, this is not the conclusion for which we argue. In fact, we believe that the literature on climate justice must take the possibility of legitimate violence much more seriously than it has to date. Malm's critique of the strategic pacifism it shares with the broader climate movement seems to us both compelling and to raise a series of important normative questions about the appropriateness of forceful responses to the threat of climate disaster. The other possibility is that reductivism's central claim, that defensive violence can only be justified if it can be reduced to violence that would be appropriate between individuals, is false. We have not argued for this conclusion, but given the barriers reductivism's individualism poses to grasping the collective structure of climate change and its politics, we believe it is the more promising way to respond to our argument.