What does it mean when we say that one factor is more or less important than another in identifying the causes of social conflict? Thomas Homer-Dixon writes here on causality in complex systems, in response to Alex de Waal’s earlier post Is Climate Change the Culprit for Darfur? and to Declan Butler’s June 28th Nature article Darfur’s climate roots challenged. Thomas Homer-Dixon holds the George Ignatieff Chair of Peace and Conflict Studies at the Trudeau Centre for Peace and Conflict Studies at University College, University of Toronto.
Here at the University of Toronto, we have been trying to understand links between climate change, environment and conflict for almost two decades. We’ve learned from this research that additive models of causation are rarely valid in complex ecological-human systems. Causation is almost always multiplicative (or, in the jargon of social scientists, interactive). Rather than an A+B+C relationship holding among causes, we find an AxBxC relationship.Interactivity has a number of critical implications, the most important of which is that arguments about the relative importance of one cause over another are usually a waste of time. While in an A+B+C relationship, it might be appropriate to claim that cause A is more or less “important” (or “powerful” or “significant”) than cause B or C, in an AxBxC relationship, all causes are equally important, so such claims are invalid.Much of the discussion about the role of climate change in the Darfur tragedy runs aground on this issue. Commentators and researchers adopt an implicit additivity assumption. They then argue over THE cause or set of causes of the conflict or over which factors are REALLY important and which are not.Counter-factual thought experiments (of the form, “if B had happened instead of A, then Z would have happened instead of Y”) also assume additivity of causes: the implication is that one can simply change or subtract a causal factor (in the Darfur case, climate change) and see what happens (in the thought experiment) to the dependent variable in question (in this case, conflict).Research on causation in complex systems over the last two decades suggests that such mental manipulations almost always produce erroneous and even meaningless results. The causal variables and links within complex systems are so numerous, so many of these variables and links are unknown, and so many causal relations are reciprocal and/or nonlinear that we can’t possibly know a priori the consequences of subtracting or altering only one factor. Changing one thing will have ramifying and unpredictable consequences through the entire causal network. In this case, we have absolutely no way of knowing the consequences of holding the climate constant in Darfur (whatever “constant” means).Jettisoning the additivity assumption isn’t easy for researchers, commentators, or policy makers. Researchers often stake professional reputations on the claim that one cause is key relative to all others. Commentators have a bias towards simplistic arguments in public discussions where info-glut reduces everything to sound bites. And policy makers want to know which factor is most important so they can determine where to invest their resources to produce a “solution.” But just because the additivity assumption makes people’s lives easier doesn’t mean it’s correct.In the case of Darfur, it’s pointless to ask about, or to argue over, the relative importance of climate change as a cause of the violence. But based on the evidence available, we can say with considerable confidence that any adequate description or explanation of the crisis must include climate change as a causal factor.I find Alex’s analysis in his essay on the SSRC website persuasive, and it fits well, I think, with the general framework I laid out in my Princeton book Environment, Scarcity, and Violence. But, I have a few comments on Alex’s entry along the lines of the points above.
1. In Alex’s first paragraph, he states that,
“depleted natural resources and livelihood transformations cannot on their own account for conflict, let alone armed conflict. The most important culprit for violence in Darfur is government, which not only failed to utilize local and central institutions to address the problems of environmental stress in Darfur, but actually worsened the situation through its militarized, crisis management interventions whenever political disputes have arisen. In turn, violent conflict has worsened Darfur’s ecological crisis. For many reasons, Darfur cannot now be reconstituted the way it was.”
I entirely agree with the first sentence. I’m troubled, though, by the “most important” phrase in the second sentence, for two reasons. First, it implies it’s possible in such a complex system to discriminate among causes by their relative weight or power. But the fundamental lesson of studies of complex systems over the last two decades is that such systems exhibit a disproportionality between cause and effect: small events can produce huge effects, and big events can sometimes produce virtually no effects whatsoever. Many complexity theorists believe that this disproportionality is the essence of nonlinear behavior, and the human-ecological system in Darfur (like all human-ecological systems) is clearly complex and nonlinear. I don’t think any of the evidence Alex adduces in the rest of his essay’s penetrating analysis supports the “most important” weighting claim he makes in this sentence. In fact, most of his analysis seems to implicitly acknowledge — not only multicausality — but also the impossibility of discriminating among the relative power of causes.
Second, in Alex’s later analysis in the essay he argues that the government’s causal role was often one of neglect — that it DIDN’T DO things that could have prevented a further deterioration of the situation. I’m always troubled by claims that suggest a necessary cause of a particular phenomenon was the ABSENCE of a particular factor. I’m not sure what “cause” means in such cases, because there is no thing or event that is doing the “causing.” Of course, this concern of mine doesn’t hold in cases where the Sudanese government’s “militarized, crisis management interventions” worsened the situation (as Alex notes later in the second sentence), because here we had real events/policies that were causes.
2. I agree that traditional land-management and stewardship practices are sustainable — but only if population densities aren’t too high. Swidden, for instance, is sustainable if population densities are low enough to allow overused land to lie fallow and to regenerate bush cover. Surely the quadrupling of Darfur’s population had an effect on the sustainability of the region’s traditional land-use practices. I’m not sure that the implications of population growth can be limited to a “much bigger human impact” of climate and environmental change, as Alex suggests.
3. I don’t buy, I’m afraid, Alex’s “refutation” of Malthusianism. The assertion that Darfur is “overpopulated” doesn’t stand or fall on the number of deaths arising from famines in the region. In other words, a region’s population doesn’t need to collapse for the region to be overpopulated. We can quite reasonably operationalize “overpopulation” using a variety of other metrics, including morbidity, land degradation, per capita agricultural yields, trends in average income, etc
4. Note the “absent-factor-as-necessary-cause” explanation of the famine in this passage:
“The reason for famine in 1984-85 and food insecurity subsequently was technological: farmers did not apply efficient fertilizers and utilize small-scale irrigation. This was not through ignorance. Darfur’s farmers have a remarkable entrepreneurial spirit. James Morton’s study of Darfur documents examples of how rapidly Darfur farmers adopt new crops, new fertilizers and new irrigation techniques. But they need either public sector investment to make these inputs cheap and reliable, or good markets to ensure sufficient returns. In the 1980s, Darfur was deprived of both: roads were so bad and fuel prices so high that the cost of transport meant that it simply wasn’t worthwhile to produce more than could be sold locally. The economic incentives simply did not exist to expand production.”
Here, again, Alex argues that the absence of something (in this case, economic incentives arising from public sector investment and good markets) caused an event. But something that doesn’t exist — or never happened — can’t be a cause.
He more or less implicitly acknowledges this concern in the following sentences:
“If technology and market access are held constant, then drought and desertification can cause food crisis and famine. But in 1980s Sudan, the situation was often worse. Government policy and the activities of traders and commercial farmers closely associated with the ruling elites meant that rural people were often deprived of the meager assets that they possessed.”
Here we come back to real events and policies as causes. Partly for this reason, I find the next section of Alex’s essay very convincing. I wonder, though, whether the “depredations of traders and commercial farmers” were made more severe by worsening environmental scarcity. In our work we have identified a process of “resource capture” by elites that is often accentuated when key resources become increasingly scarce, because the economic rents from monopoly control of the resources rise — exactly the kind of phenomenon Alex refers to in the next sentence:
“Worst of all, a handful of major grain merchants sought to profit from speculating in the cereal market in 1985-85, withholding stocks from sale until the price climbed high.”
5. In the following passage Alex introduces temporality of causes — that is, where the causes enter the causal process in the time-line of the causal chain.
“[The] underlying causes of the famine were impoverishment associated with neglect, alongside the inadequacies of existing technology and marketing infrastructure, the immediate cause of the famine was drought compounded by the government’s denial that a problem existed.”
As I’ve noted in some of my work (including the appendix to chapter 5 in Environment, Scarcity, and Violence), the variable of temporality (or what I refer to as “proximity”) is often conflated with the variable of causal power or weight. In the above sentences Alex seems to be doing this: he makes a distinction between “underlying” and “immediate” causes (a distinction of proximity, in my terminology), at the same time that he appears to be implying the underlying causes were more important. (It’s possible, though, that I’m reading too much into this sentence.) Note, once again, the “absent factors” as causes: inadequacies of existing technology and marketing infrastructure and government inaction.
6. I think Alex’s next section about adaptation and dislocation in response to the famine is terrific. I have no quibbles here. In fact, this kind of mediated, multicausal, and multi-stage causation is exactly the kind of process that I identify repeatedly in my work, including in my accounts of various cases in Environment, Scarcity, and Violence. The essay weaves a compelling account here.
7. I agree that the fourth hypothesis is misguided, for the reasons Alex adduces. But the failure of that hypothesis doesn’t weaken the general case that drought has been a contributing factor to the crisis in the Sudan, and more specifically Darfur, in a multicausal, interactive, nonlinear process.
8. Alex is right that it is premature to attribute the drying trend in the Sahel — and especially the drought in the mid-80s — to anthropogenic climate change. But even if the drought itself was not human-induced, it provides an instructive example of the human and social impacts of potential future anthropogenic climate change.
9. In the following sentence in Alex’s concluding paragraph, he appears to return to additivity:
“Climate change causes livelihood change, which in turn causes disputes. Social institutions can handle these conflicts and settle them in a non-violent manner — it is mismanagement and militarization that cause war and massacre.”
Here he is asserting that “mismanagement and militarization” are the true causes of the violence. I would accept that these are indeed causes, but only in a larger constellation of causes that interact with each other across time. One component of that constellation is climate.