Climate Science Crystal Ball: Towards a Cyborg Climate Science

I’ve been trafficking in data. Petabytes of it: ocean temperatures, atmospheric halocarbon concentrations, Arctic sea ice extent, untold amounts more. In climate science, we need more data in order to reduce uncertainty in climate modeling, which we want to do in order to have a higher fidelity image of possible climate futures that guide decarbonization action. When I feed data into a climate model, I see numbers and scenarios and a whole lot of code. But I also see the future. It’s like my own version of a crystal ball, a divination practice for the Anthropocene. Our prospects right now look grim, but like most of my colleagues in climate science, I’m motivated by the idea that it could be better, that a livable future is within our grasp. This link between the climate of the past, the decarbonization actions of the present, and the livability of the future connects me to past and future generations along a geologic time scale. If models are crystal balls, climate scientists are the psychics imbuing predictions of the future with legitimacy. But legitimacy is not a given; my prophets are an unruly, distributed bunch.

How do I even know that climate change is happening? Sometimes, it feels obvious: record high temperatures, devastating flooding, wildfire smoke engulfing my city. But this isn’t climate change, really, these are natural disasters. Weather. Understanding myself in relation to an entire complex planetary system is fundamentally destabilizing. After all, nobody experiences the global climate. There’s an impossibility in holding the unfathomable nature of something such as climate change as truth in your mind, an example of what environmental philosopher Timothy Morton labels a “hyperobject” – something so complicated, operating on such a large timescale, that it’s impossible for one person to conceptualize. So how am I so certain that climate change is not only happening, but is such an impending threat to cause me bouts of climate grief and anxiety?

If models are crystal balls, climate scientists are the psychics imbuing predictions of the future with legitimacy. But legitimacy is not a given; my prophets are an unruly, distributed bunch.

Mostly, it’s the science. Science is real, went the rallying cry of Trump-era liberal media, dotting signs at protest marches and in the lawns of affluent suburban homes. Climate change is science at work, itself an empirically validated pattern and scientific understanding of the history (and therefore potential future) of Earth as a system. It draws from an increasingly complex, interconnected apparatus of biogeochemical sensing, computation, modeling, and surveillance, what historian of science Paul Edwards calls in his book A Vast Machine, a history of computer models and climate, “knowledge infrastructure.” Belief in climate change, belief in the apocalyptic forecasts of life in 2100 or the systemic meaning of unprecedented flooding, means believing in the efficacy and accuracy of this apparatus. Constructing the ability to not only understand but predict the impacts of climate change requires a collective effort so complex and multifaceted that it’s nigh-impossible for any one person to take it all in, to know where this data is coming from and what it means. Still, I attribute an extremely hot summer to the burning of fossil fuels, an attribution borne out of faith not only in the data being collected on our oceans and atmosphere, but in the computation and modeling that puts it all together.

Climate models predict the future. They take in parameters that measure the history of the global system — greenhouse gas concentrations, volcanic eruptions, aerosols, literally thousands more — in order to tell us what might happen in the coming years. I think of it as a divination for the post-Enlightenment world, a message from the gods of reason, evidence, and science. The machinations of climate models not only help us to understand what might be done to avert the worst of climate catastrophe, but also how we know climate change is even happening at all. Modeling acts as a nexus connecting individual events — natural disasters, record high temperatures, habitat losses — and the global phenomenon of climate change. It’s how we’ve come to see record high temperatures as not just an unfortunate coincidence, but part of an interrelated system with local effects. Science guides us as we enter into a planetary way of understanding the world.

It also guides our thinking about what the future might bring. Most of this is likely familiar: record temperature extremes, disappearing islands, a world on fire. These are outputs from climate models, which most commonly model a set of five headline scenarios of future developments in global society, economy, and demographics, known as Shared Socioeconomic Pathways (SSPs). Each model adds another wing to the grand apparatus of climate science that is perhaps the most sophisticated example of global coordination in history. This also means that the future of climate science is not a given; we’re entering uncharted territory. The field can be intimidating with its complexity and rigid protocols, but an examination of it as a process reveals that there are still opportunities within climate science to call for a modeling that enables a wider breadth of climate action than SSPs endorse.

---

How does one accomplish the historically unprecedented task of sensing an entire planet? Climate science involves a constantly re-negotiated entanglement of state, military, and market actors. Due to the benefits for national and market stability, the field shares incentives with governmental institutions; namely, to surveil and therefore monitor behavior across the world. Uncertainty is bad for business, and bad for national security, too. So climate science has always been political, and will only continue to be so. It’s a field that takes place mostly outside the academy: data is collected in the Amazon rainforest, funding is authorized in the chambers of Congress. Some may see in this a reason to discredit the legitimacy of climate science, but Wendy Hui Kyong Chun shows us that uncertainty (whether from unreliable political actors or unstable Earth systems) should be addressed as enabling rather than disabling. When moving from climate models as evidence to models as hypotheses (emphasis on the plural), as goes Chun’s argument, we open up space to speculate on possible future scenarios, beyond SSPs and their inherent assumptions. Thus, I argue for an explicitly political climate modeling, one that takes into account a wider variety of possible futures and perspectives, including those beyond the frameworks and constraints of global sensing systems.

By leveraging climate models as speculations, we loosen our tendency to rely upon “hard evidence” on the wretchedness of our futures and instead argue directly for a livable future. The economic, social, political, and climatic benefits of taking action together make for a much more compelling argument for climate action than the simple technical fixes that comprise decarbonization action in most scenarios. Models offer the technical piece of this argument. Take, for instance, the Climate & Community report on transportation decarbonization pathways, in which the authors do the math to show how changes to the organization of how people move in America — public transit subsidies, walkable cities, high speed rail — will result in not only less carbon emissions, but more mobility and economic opportunity for the average American. Further, changes to the organization of our transportation networks can reduce lithium demand by up to 92 percent in comparison to the most lithium-intensive scenarios, according to the report. The Climate & Community report speculates on what a less car-centric future could look like, and then uses technical analysis to back up its hypothesis that such a future is one of the most equitable and least-extractive scenarios available to us.

But we can go even further. While most models simulating scenarios in IPCC reports require large amounts of time and expense to run, emulators simplify aspects of climate models to vastly reduce the parameter space of the simulation being run, thereby greatly decreasing the expense needed to speculate on climate futures. Some emulators simplify entire climate models, but most specifically emulate particular processes in the Earth system such as sea level rise or regional sea ice extent, thereby enabling tweaks to key variables without re-running entire simulation processes. Emulator-based climate modeling promises a wider variety of parameters with which one can alter the scenario at hand, thus becoming a tool for further speculation on possible climate futures. Early stage emulation tools, such as En-ROADS by Climate Interactive, demonstrate how, in the future, emulators may enable anyone to adjust for various actions taken on global or even regional scales to understand possible decarbonization pathways. Climate science is always going to need more up-to-date data on Earth systems, but a move towards emulation demonstrates that rethinking the processes by which we come to understand climate change can help us to break from narratives that claim that all we need is more data or compute power. We can and should be constantly reconfiguring our tools to hypothesize on the most just pathways to fight climate change.

---

As someone immersed in the complicated world of climate modeling through my day job¹, I still find myself falling into wishful thinking. I want climate models to be the crystal ball that makes unequivocal the steps we must take to avert the worst of climate catastrophe. I want the models to reinforce my own beliefs about what it will take to take on climate change. I want to hear that we have no choice but to restructure our society, redistribute wealth, and greatly reduce consumption habits in the Global North, and I want our global leaders to listen. Maybe my secular background has me searching for something, such as climate science, to put my faith in. Maybe I’m searching for a foothold of stability amongst an increasingly insecure global system. Maybe I don’t know where to direct my climate grief in a way that makes me feel part of a movement towards progress.

Donna Haraway shows me that this is an example of what she calls the “god-trick,” a metaphorical turn in which I attempt “seeing everything from nowhere.” Haraway critiques the way in which data is treated as objective, unburdened by the context that informs how it lives in the world. But all data has built-in bias, comes from a particular historical moment, and involves actors with their own incentives and motivations. Climate data is no different. When does our wishful thinking prevent us from focusing on the material realities of our ongoing climate crisis? When does it keep us from taking action, from demanding action, from confronting the systems we don’t want to confront? The god-trick is our way of ignoring that the real problems, at this point, are more about social transformation and mobilization than they are about science. Calls for more data, for more precise modeling can serve as distractions from the changes we need to make to our world and to the way it is organized.

Maybe my secular background has me searching for something, such as climate science, to put my faith in. Maybe I’m searching for a foothold of stability amongst an increasingly insecure global system. Maybe I don’t know where to direct my climate grief in a way that makes me feel part of a movement towards progress.

We can and should acknowledge that the god-trick is exactly that: a trick. Models will not give us the once and future climate, really, and no matter how much data we acquire, there will still be other work to do to achieve climate justice. I work to improve the quality and quantity of climate data not because I believe that we will reach a point where we “finally know enough” to take climate action, but because I believe in the ability of this data to help us on our path to a livable future. It’s a lot easier to fall prey to the god-trick, but we cannot live in the simulation. Working with models and data all day, you’re susceptible to end up with your head in the cloud, so to speak, spending all your time on the abstracted meta-scientific plane, far away from the material impacts of changes to the climate. More abstraction shouldn’t distance us from those all over the world facing the daily threat of rising sea levels, longer droughts, and unprecedented natural disasters — it should bring us closer.

Maybe this doesn’t seem like the domain of climate modeling; traditionally, it isn’t. But I want a climate modeling that drives us towards effectiveness, rather than efficiency, an argument Adrienne Buller makes in The Value of a Whale: On the Illusions of Green Capitalism. Climate change is an urgent and all-encompassing threat, but that doesn’t mean we should give up on social transformations in the pursuit of decarbonization. Global warming is not purely a technical problem of balancing carbon concentrations, and so analysis of the outputs of climate models, whether in op-eds or IPCC reports, should make clear the material stakes of climate scenarios they model. What makes for the most efficient decarbonization pathway may not be the most effective, for instance, if it requires the massive deployment of Direct Air Capture plants which use up dwindling freshwater supply, or the extensive mining of rare-earth oxides. While seemingly natural to model the global impacts in the abstract, the local material realities of a Greenlander living near Kvanefjeld — a former uranium mine currently center-stage in a battle in Danish court over its potential as the world’s second largest known deposit of these rare-earth oxides — are at risk of being overlooked in pursuit of the quickest path to net-zero. Climate models can and should surface the now and future hotbeds of environmental justice in their predictions – a parameter that signifies the electrification of transportation, for instance, should be accompanied by analysis of how mining the rare-earth oxides essential to electric vehicle batteries poses a risk to the human rights of local mining communities. Going further, the narratives and analysis around modeling should pointedly identify and support frontline communities most impacted by changes to land-use, in doing so moving from the abstract to the material.

---

Even still, there remain contradictions inherent to climate modeling, which belie a certain set of assumptions about what matters in climate change. Expanding definitions of what it means to understand the climate, particularly beyond the technical, helps us to erode our reliance upon technical specificity that obfuscates pathways to climate justice. Speculation can only do so much, and our desire for easy-outs may lead us to elevate speculations as supposed panaceas, no matter how effective. Climate fiction, for instance, has been elevated as a savior genre, capable of literally writing the blueprint of a desirable climate future. But Kim Stanley Robinson will not save us; we need a wider cast of heroes.

Haraway, in her critique of the god-trick in data science, calls for a series of situated knowledges, by which she means claiming objectivity as a constantly re-negotiated construction, in doing so “insisting metaphorically on the particularity and embodiment of all vision (although not necessarily organic embodiment and including technical mediation).” Bruno Latour has drawn on Haraway to call for an empiricism that cares, particularly in relation to climate change. But how exactly do we do this? First of all, we give up the fantasy of total understanding: reject the god-trick. Secondly, recognize the limitations in our ability to foreground care in current ways of doing and understanding climate science (though moving from efficiency to effectiveness does help in this). It’s hard to care for parts per million in the abstract, but it’s a lot more intuitive to care for a migrating bird population, even more so to care for your child or neighbor.

Parts per million, global average temperatures, inches of sea level rise: these are useful terms (themselves the result of decades of building scientific consensus), but they are not the only ways of understanding the climate. Traditional Ecological Knowledge (TEK) is a robust, heterogeneous system of knowledge production practiced by Indigenous peoples, which offers us some alternative metrics of ecosystem health: how do native plants take to changing water salinity? Are bird migration patterns changing from their historic timelines? It also offers alternative solutions, such as the floating gardens used by farmers in parts of Bangladesh to adapt to flooding. Integrating TEK with climate modeling does not mean simply replacing or integrating qualitative ways of knowing. Instead, it means elevating that which builds towards adaptations and local equilibria, in addition to — rather than at the expense of — the global equilibrium. The global carbon balance sheet will not save us, but by expanding definitions of understanding what climate change looks like, we allow for more possibilities to build and sustain relationships with the planet.

I sometimes find this dissatisfying; I want climate models to work seamlessly with TEK, or I otherwise fall into fantasies of a traditionalist return to nature that rejects global climate data collection entirely. But this is the god-trick seducing me again; I need to get messier. What I’m really looking for is a cyborg climate science: that which embraces both TEK and highly technical climate models, in the process finding new subjectivities with which to understand and care for the dynamic environment around us. Ultimately, Afrofuturism, Solarpunk, Indigenous futurism, and other radical imaginations of what our future can and should look like are not mutually exclusive with the millions of lines of code that make up a climate model. By embracing an ever-expanding landscape of situated knowledge, models can provide the very technical background that we need in order hento actualize and act upon these futures.

The global carbon balance sheet will not save us, but by expanding definitions of understanding what climate change looks like, we allow for more possibilities to build and sustain relationships with the planet.

There’s room here, too, for a building of relations between these other ways of knowing and the Western paradigm of climate science. In an essay entitled Making Kin with the Machines, Oglala Lakota writer Dr. Suzanne Kite relays how Lakota ethics and ontology can be communicated through understanding wakȟáŋ, a complex term that indicates how “non-humans have spirits that do not come from us or our imaginings but from elsewhere, from a place we cannot understand.” There, she’s referring to kinship relations with AI, but Kite’s writing prompts me to wonder about the possibilities for an Indigenous-led climate modeling. What would it mean to make kin with climate models? It’s not entirely my question to answer, but by engaging a cyborg climate science, we can begin to interweave TEK with the terabytes of climate science data.

Kite’s writing speaks to me because it touches on the complexity and unknowability of climate data. Day in and day out, I engage with representations of the non-human, from the motion of individual glaciers to the currents of the ocean, that come from a place we cannot understand. Accepting and embracing that climate modeling builds understanding across unknowability is one way we can integrate TEK with climate science, one way we can stop the god-trick in its path, one way we can foreground effectiveness over efficiency. It can be difficult, at times, to remain connected to the Earth, while spending most of my day looking at a computer screen poring over datasets. But I refuse to accept this as a given; I imagine a climate science that shows us all the futures worth fighting for, and how to get there, one dataset at a time.

Notes

Footnotes

1. I work in philanthropy, on a team of climate scientists building out a program for more effective understandings of energy usage, carbon cycle behavior, and freshwater availability. ↩

---