This year may see a populist backlash against the role of simulations and big data in policy. But the failures of governance by curve in the wake of the COVID-19 pandemic must not repeat in the case of climate crisis mitigation.
—“The graph on the wall tells the story of it all.”—Depeche Mode, “Everything Counts”
This is a moment of (failed) governance by curve. Governance by curve sounds boring, but it is in fact morbid.
A curve does not sound like a form of governance. But it can provide a frame, an instruction set, a kind of prescription for what governance is needed. Policy is then set out in order to bring reality into conformance with the curve. The curve is a rudimentary representation of a possible, simulated future. The representation sets events into motion.
In the US, a shambolic and halted attempt at governance by curve of the COVID-19 pandemic has had follow-on implications around the globe. Lockdowns informed by simple simulations resulted in an economic meltdown, significantly impacting 1.6 billion informal workers around the globe who lost 60 percent of earnings. Acute food insecurity is predicted to rise from 135 million to 265 million by the end of the year. In the United States, 1.4 million health care jobs have been lost; at the same time, an estimated 13 million people are losing employer-based health insurance. Social services funded by state and local governments are being demolished.
The systemic effects from panicked attempts to achieve the curve may cost far more lives than COVID-19 itself. They are also crushing the very capacities needed to confront the virus and other global challenges, in effect placing crisis above all the other ongoing and future ones. The fact that officials can enact emergency measures such as they did here without even a documented accounting of how they thought about these follow-on effects should be alarming.
Now, you might say—the curve and the projections were right. The unprecedented lockdown measures were the best policy we could have tried, given the scientific uncertainty in March of 2020, and the basic idea of using them to bolster hospital capacity was correct. The problem was that the follow-through was absent, in the form of testing, tracing, and preparation of hospital capacity, such as governments like South Korea or Taiwan were able to achieve. In the US, this was a leadership failure of a government hollowed-out by a narcissist and troll-in-chief, not a failure of the curve itself.
While it has absolutely been a leadership failure—and we must demand accountability for this—the enterprise of governing by curve is also rotten. The curve and its interlocutors also have some culpability for the follow-on disaster, whose full impacts are yet to be seen. That is a hard message to hear in the middle of a public health crisis. It is also a rather easy-to-miss dimension of the problem amidst far more glaring outrages.
But there are two critical reasons that we must grapple with the failure of governance by curve, and not just gloss over it. The first, which I won’t elaborate here, is that we need a politics of response to COVID-19 that goes beyond the curve and its metrics, because simply clamoring louder to actualize the curve at this point is a political dead-end. The second, which I will describe in what follows, is that we are in danger of applying the same type of management to the climate crisis.
Meet three charismatic curves
1. The 2°C curve
Curbing warming to 2°C follows this curve. It is actually quite hard to explain concepts like “net-zero” or “negative emissions” without reference to this curve—it carries a lot of the conceptual work. Basically, what you see here is a point where the world is removing more carbon than it is emitting, towards the end of the century.
“Negative emissions” are spoken of in reference to conceptual figures like these—in the real world, they could be achieved with “negative emissions technologies,” which are technologies which can generate negative emissions if they are used in particular systems designed to realize this figure. Bioenergy paired with geologic carbon capture and storage (CCS), as well as direct air capture with CCS, can be negative emissions technologies if their inputs are low-carbon enough. Afforestation, ocean fertilization, soil carbon sequestration, and accelerated weathering of rocks are other negative emissions techniques, which I’ll simply refer to as carbon removal methods.
The curve is an average of integrated assessment model runs, which have included negative emissions technologies as part of how to curb warming to 2°C. A lot of computing power went into this curve. Many possible worlds are implied there, collapsed into one.
Here are some more curves from the IPCC’s Special Report on Global Warming of 1.5 °C (SR1.5), looking at different scenarios. In this depiction, you get to pick your curve.
Many people like P1. P1 is a version where the world keeps warming to 1.5°C without using carbon capture and storage. However, there are a few challenges with it—not just with actualizing the steep drop in emissions and scale-up of renewables, but also the requirement to afforest vast tracts of land, e.g. 500 million hectares converted to forest, with a reduction of 500 million hectares of pastureland (so, roughly half the size of the US or China). This could be done via agroforestry, but also via plantations, and it could involve voluntary incentives and indigenous stewardship or it could involve forced displacement. The choices about how to achieve this goal are not simple, even though the picture is.
But back to this 2°C curve. Manifesting it implies building a vast infrastructure which is on the scale of the existing fossil fuel industry, but designed to put carbon back underground, and the renewable capacity to run it cleanly—to power the systems for direct air capture or growing biomass renewably, i.e. we are going to build up even more renewable capacity just for this purpose. This is what the −10GT carbon at the end of the century is telling us. It implies a governance system, a social infrastructure, that can execute a response that involves building a gargantuan clean-up operation, and transfer incredible mass from the atmosphere to the lithosphere and biosphere.
So this curve is not a curve of something that will happen, or that is likely to happen. Don’t read it that way. It’s an aspiration. Some people are treating it as an eventuality. I’ve seen these curves projected at conferences, with the experts assuming that we are now going to build this, because the IPCC said we need to. The whole prescription here follows from making reality conform to the curve. But know that achieving this curve would be the most ambitious planned endeavor a global society has ever mounted. Meanwhile, one of the world’s largest economies can’t organize enough cotton swabs to adequately detect a microorganism. Which brings us to the next curve.
2. The peak-shaving curve
In part because that decarbonization curve is so hard to deliver, and not particularly likely given the constantly-shrinking timeframe, this curve has appeared on the scene. This is the so-called “peak shaving” diagram of a stratospheric aerosol intervention, in which particles are injected into the stratosphere to reflect a fraction of incoming sunlight, thereby cooling the Earth. In this formulation, it can “buy time” for decarbonization, and shave the peak off of the climate impacts—it’s conceptualized as temporary harm reduction, though the time axis here is likely a century or two, at least.
This has been described by many researchers I’ve spoken with as the best-case scenario for how to do a solar geoengineering intervention. It also functions as an argument packaged into a curve.
One challenge with this curve in particular is that if the curve isn’t brought down—but the solar geoengineering line is interrupted—warming goes rapidly upward to a level commensurate with the amount of carbon dioxide in the atmosphere, in what’s known as the termination effect or a termination shock. This means that beginning the intervention but failing to conform to the curve bears unique risks.
3. The Western story arc
Here we have the plot diagram for much of Western literature: the rising action-climax-falling action curve, a basic form that has been observed since Aristotle’s Poetics. If you took a literature or drama class in school, you may have encountered this curve, and perhaps had to fill in worksheets like these.
I don’t think the similarity between this form and the climate prescription curves is a coincidence, especially since the storyline puts us up top, at the climax, at the hero’s moment of action, inviting actors to claim that role. Much of climate politics consists of filling in this worksheet.
What science and policy seem to be doing with these curves that modelers have made—even though they look like “science” and are described in those terms—is trying to put a familiar narrative structure onto our terrible situation, one in which it works out okay in the end. They are not “science,” they are a story. I don’t have a way of testing that hypothesis, so it’s an intuition at present.
In reality, there never is a real end, and much damage is irreparable. This isn’t something that US society in particular seems to have a narrative form for.
“Flattening the curve”
Trying to make reality conform to the curve is an understandable narrative drive, but an incredibly simplistic way to govern complex systems.
Why not try for the curve? Because that focus obscures all the things that are outside the curve, which are also things we care about.
In the case of the 2°C scenario, we might miss the effort involved—technical, economic, social—in building the massive carbon removal infrastructure. (I am actually all for building up much of this infrastructure, though I also think the curve masks legitimate questions about the amounts and allocation of significant residual emissions).
In the case of peak-shaving stratospheric aerosol injections, we might miss the work involved with both carbon removal and mitigation, and the risk of failing to follow through with them. This is perhaps the key concern that people have with the idea.
And in the case of COVID-19, we might miss the collapse of a global economy—a complex system which is entangled with culture, health, and security, and not really just about an abstract “economy” at all. Let’s talk a bit about this other curve we’ve come to know so well: that of COVID-19 mortality (or infections, or hospitalizations).
Introducing this curve, and acting upon it, has not been handled well. There are three particularly glaring ways that the curve has been a tool of confusion.
First, the focus of the temporality of this curve was misunderstood, particularly in the US context.
Trump’s pivot towards taking the disease seriously was related to the release of the Imperial College report in mid-March, with its projections of up to 2.2 million deaths, told in curve form.
As Neil Ferguson, lead author of the Imperial College report, said to the New York Times, “we don’t have a clear exit strategy... we’re going to have to suppress this virus—frankly indefinitely...” The original report has curves that stretch into 2021.
But the administration cropped out the second part of the timeline in much of its graphics, and launched a 15-day initiative, despite the lengthy temporalities of the multiple curves.
The curve that politicians have been displaying is only a piece of a larger story. (Since I gave the talk this article is based on, other commentators, like David Wallace-Wells and Nate Silver, have also been making observations on what the models can’t see, and critiquing the assumptions of symmetry in the curves).
The result is a public that’s extremely confused about the aims of sheltering in place and the overall plan, and feeling like the goalposts have been changed—which is dangerous in a conspiracy-laden media ecosystem. Cropping the curve might have been considered an easier form of science communication, but it was a mistake.
This shortened and simplified curve has made people misunderstand the nature of the problem and its management. It’s not like you flatten the curve and the problem goes away. Flattening the curve pushes the deaths from the virus later in time, and in theory, minimizes deaths by ensuring hospital capacity is not overwhelmed. Which is a good idea, if governments have used the interim to actually do something, like increase testing or build hospital capacity or manufacture protective equipment. But in the US, the time that was bought was then squandered—like a home paid for dearly and then set aflame.
Second, the curve only captures one metric, typically deaths or hospital capacity. It does so without context. Deaths and hospital capacity are obviously important metrics. But the list of other things that policy must also contend with is long, and much harder to render as data: deaths from canceled or postponed medical care and vaccination programs, mental health impacts, lost livelihoods. Policymakers have the direct responsibility for taking care of those dimensions domestically, but they would also be wise to consider the international ramifications, both in terms of security and humanitarian interests.
Yet there’s little evidence that these dimensions of lockdowns were anticipated or considered.
If you read the community pandemic mitigation guidelines that the US CDC put out in 2007 and updated in 2017, they don’t say anything about the potential for mass unemployment. They suggest that maybe some people can telework, and that workers should consider saving some money in case their workplace is closed. The question remains: Did policymakers expect a global economic meltdown as an effect of following the unprecedented recommendations? Or was it unforeseen? There is very little transparency about this decision-making process. It is insane that we do not know the answer to these questions. I’m not saying that the decision to close down society was unwarranted; the evidence indicates that it was a reasonable thing to do, at least at first. But it is important to know about the foresight here, because if our institutions are entirely incapable of foreseeing even basic second and third-order effects from the decisions they take based on the simulations, that indicates that either our simulation technologies are not up to the task, or that officials do not know how to incorporate the simulations into governance. Both of those things would be a big problem.
Third—and this is related to the last point—the singular curve doesn’t encourage a conversation about choices in how the intervention is designed and deployed.
Consider this figure from a paper published by Kissler et al. in April 2020 in Science, which illustrates multiple curves with different levels of social distancing (in this version, with no seasonality to transmission). These simulations found that “longer and more stringent temporary social distancing did not always correlate with greater reductions in epidemic peak size,” and that “for simulations with seasonal forcing, the post-intervention resurgent peak could exceed the size of the unconstrained epidemic,” in terms of peak prevalence and total infected. These findings are not well discussed in the press. They should be translated into policy possibilities and deliberated upon.
The lesson here might be “multiple curves are better than one curve,” but even that accepts the contextual limitations. What’s really needed is a multidimensional operating space in which the systemic follow-on effects are identified in broad daylight. First, we need the conceptual language for that.
Governance by curve has some culpability in making things worse. “The curve” here encompasses the form, its genesis in narrow-focused knowledge production, and its communication. The curve looks like an advancement in science, in modeling capacity—but it actually results in a reduction in our capacity to fully grasp the situation.
Is this perhaps an intermediate stage in us learning how to create simulations that help rather than harm us? Later this century, will we get to some later point where we are able to model and account for all the follow-on effects that we care about? Can this moment of failed governance by curve just be chalked up to the growing pains of the simulation?
We can’t repeat this failed attempt at governance by curve with the climate crisis
These curves, with their gentle resolution, are a fairy tale. The resolution for this pandemic won’t be a neat and happy resolution of the curve, in the US. It probably also won’t be that way for climate change, though we have a bit more time—years, rather than weeks or months—which is a bit more time to widen the space for alternative courses of action.
The lesson from the pandemic should not be to resign ourselves to loss and damage, though. Because of inaction at the front end, lack of leadership, corruption of our institutions—we will probably have hundreds of thousands dead in the US. We will also surpass 1.5 degrees globally. Coral is done for. There has been too much delay. There are countless losses that we can’t curve-tale our way out of.
But we can still increase our capacity for systemic thinking, and make it a few hundred thousand lives lost rather than a million; make it 2 degrees or 2.5 degrees instead of 3. There are still choices, and we’re more likely to get to those choices if more people are involved in the simulations describing them.
We need “the other experts” in the room—all the sorts of expertise that are useful in seeing follow-on effects, to collectively grasp the complexity of these socio-ecological systems. But for these simulations to be accurate, we also need non-experts—not just for “social legitimacy,” or for normative reasons, but to actually see the impacts of the interventions in various contexts.
This year may see a populist backlash against modeling, against the role of simulations or big data in policy. We need to not just throw away our simulations because of what happened in 2020, though—we need to make them more populist, more legible, more shape-able. Yet the politics of the moment doesn’t seem to allow for this.
The long-term questions remain, and they will require collective imagination:
What does a collectively designed simulation for complex system governance—climate change intervention, COVID-19 management, sustainable development goals, whatever—look like? What other shapes, beyond curves, can it enable? How can it be built?
Cover image: Joyce N. Boghosian / Official White House
Holly Jean Buck