Image by Critical Carbon Computing Collective. CC-BY-NC 4.0.
How climate policy is guided by tech
Computers lie at the heart of climate policy. Computationally-intensive models identify the problems that policy tries to address, and computer-generated data define the means by which increasingly technical solutions are imagined to address those problems in the broader world.
Models are an invaluable, but never neutral, tool. They can drive the formation of policy priorities and directions in ways that preclude wider political debate and ignore the social and material challenges of realizing specific futures. The role of 1.5 and 2°C targets in the 2015 Paris Agreement is a case in point. The climate modeling community could tune their models to show a pathway to these targets, but only with unprecedented and, to most, implausible amounts of ‘negative emissions’ (technologies to remove carbon dioxide from the atmosphere and sequester it, largely underground). Realizing this feat of geoengineering using the technology contemplated in existing models would require a landmass the size of India dedicated to creating and burning biomass with carbon capture and sequestration—a geopolitical (if not geophysical) absurdity. Yet, this key detail was largely lost on policy makers, who instead proceeded with the simple message that 1.5 and 2°C were possible. In this way, climate modelers did more than just inform policy—they shaped it.
This problem continues. Since the Paris Agreement, climate policy has often been organized around “science-based pathways” that outline specific targets and technologies by which a nation, industrial sector, or company can align its emissions profile with a 1.5 or 2°C future. The Information and Communications Technology (ICT) sector has its own such pathway, calling for a 45% reduction by 2030 in order to give industry a chance to coordinate with these goals by the end of the century. Yet, these private sector pathways are not apolitical. The model-derived predictions that undergird these optimistic projections often make implausible assumptions about policy outcomes, or “greenwash” complications that arise from economic growth models.
More recently, and in light of these targets, many nations have embraced industrial policy as a climate necessity, with further consequences for the role of ICT in the imagined green economies of the future. The EU Green Deal, which aims to make Europe the first ‘climate neutral continent,’ treats tech as a green growth pathway, arguing that digital technology is key to reducing the carbon footprint. The Korean New Deal recently committed 60 trillion won in ICT spending alongside its commitment of 70 trillion won in green tech. In the United States, the Inflation Reduction Act (IRA) of 2022 is a landmark $370 billion effort largely focused on advancing clean energy by subsidizing and re-nationalizing the production of renewable electricity and electric vehicles.
A world designed in anticipation of the 2 and 1.5°C is one that increasingly relies on computational metrics and innovations. As a result, nations and industries today are committed to meet what may ultimately be unmeetable targets and are pursuing them with more hubris than ever before. Assessments of climate policy today require not just a healthy skepticism towards techno-optimistic promises, but a deeper awareness of the ways models, targets, and predictions shape climate policies from design to implementation.
How tech is guided by climate policy
The business decisions and climate trajectories of big tech companies, in turn, are also shaped by policy efforts—internal and external, formal and informal. Policy debates are important areas of activist engagement and worker struggles. However, despite the upwelling of climate bills, there is currently minimal existing legislation (especially outside of Europe) that requires explicit climate action from the tech sector. In the United States, the largest impact governments have historically had may be as small as setting efficiency standards for government data center procurements.
There is currently minimal existing legislation (especially outside of Europe) that requires explicit climate action from the tech sector.
Nevertheless, the mere threat of regulatory action is often effective. Many U.S. tech companies pursued internal carbon accounting efforts in anticipation of 2009’s Waxman-Markey bill, even though the legislation ultimately did not pass. Large international corporations increasingly proactively include some form of climate monitoring and mitigation efforts in their internal management evaluations, simply because it’s easier to use a unified framework across global operations than adopt a piecemeal approach to different national requirements. In this way, small wins and ongoing struggles can have cascading effects on climate action.
Absent state action, the tech industry continues to form its own internal standards. Sometimes this proceeds through formal standard-setting bodies, such as the Green Software Foundation’s proposed software carbon intensity specification or the International Telecommunication Union’s efforts to establish metrics and pathways that could lead the industry towards a future compatible with the Paris Agreement. These valuable efforts produce social coordination and consensus within expert knowledge communities, but do not yet forge strong links to actual business decisions or broader citizen engagement. Additionally, internal metrics may not always measure the right thing; an emphasis on efficiency, for instance, can downplay overall trends in emissions and energy use.
Competition has also played a large role in directing sustainability efforts within the industry. From Greenpeace’s Clicking Clean report cards to the escalating carbon neutral/negative pledges from big players in the industry, external consumer perceptions and internal rivalry has led to escalating climate ambitions within ICT. Further competition to attract tech talent, which is increasingly outspoken on climate issues, has intensified these concerns within management.
Yet, without stronger policy signals, these efforts can only go so far. As backsliding and evasive accounting threaten the credibility of current corporate climate pledges, and as the labor market in tech cools, it’s not clear that previous internal drivers of voluntary climate action will be maintained. It is also the case that existing state regulations can be bent for the industry, as seen in exemptions to air pollution standards contemplated in Virginia, where a large number of energy-intensive data centers are concentrated.
Overall, we can’t wait for the industry to write its own rules, nor trust that they will be sufficient. Within tech worker coalitions and the broader climate movement, we must push on both fronts for stronger internal measures and firmer external regulations.
Image by Critical Carbon Computing Collective. CC-BY-NC 4.0.
As both consumers and producers of carbon accounting tools, the information and communications technology industry (or ICT) today powerfully shapes how corporate and institutional climate impacts are monitored. Since the 2010s, big tech companies have been early and influential adopters of carbon emissions accounting and pledges, establishing these techniques as normative practices and evidence of corporate social responsibility. At the same time, the sector stands to benefit from controlling the tools and information that other sectors and organizations need to reduce their carbon emissions. How can we strike a balance between getting more granular data and building more advanced models, and recognizing when the production of new environmental data and models may lead to near monopolies, and might actually narrow our pathways to meaningful action?
ICT and Carbon: A Two-Way Street
The ICT sector–like all industries–has material impacts on the climate system. As data infrastructures, machine learning models, and smart systems expand, these impacts are growing. It’s difficult to estimate the global picture, and efficiency and substitution effects can be complex, but ICT is probably responsible for somewhere between 1.4 – 3.9% of annual emissions globally. To do its fair share in the climate struggle, the sector will have to decrease these numbers.
At the same time, digital technologies are essential to monitoring emissions and constructing the complex models through which we understand climate change. These include skills and tools like advanced data analytics, blockchain ledgers, and the Internet of Things. Counting carbon is a fraught endeavor because of the many intersecting systems and counterfactual scenarios one needs to consider; it makes sense that we’d want digital tools to help. But as networked sensors and machine learning tools play an expanding role in assessing and even predicting carbon inventories, in spite of the growing carbon costs of producing these tools, something of a paradox may be forming.
Simplistic methodologies, bad data, and incorrect assumptions about future events are also enduring problems in carbon markets: the national and voluntary trading of carbon offsets and certificates. Offsetting can further perpetuate colonialpatterns of resource control, and a single-minded focus on carbon can undermine issues like energy equity, biodiversity, water access, and local self-determination. What’s more, dubious offsets can distract from more credible strategies for addressing climate change, such as verified removals or operational reductions against industry-specific baselines. As a result, even those who work to become carbon neutral or net zero can end up causing lasting harms.
The problems with carbon markets don’t mean that carbon accounting should be abandoned; the contributions of various actors to the composition of the atmosphere do need to be tracked. As Nobel Prize-winning economist Elinor Ostrom has demonstrated, monitoring behavior is both necessary for managing environmental resource commons and insufficient on its own for sustaining such commons. Environmental management is effective when monitoring is accompanied by other factors, such as an alignment between rules and local conditions, and the ability for everyone affected to participate in governance processes.
Monitoring behavior is both necessary for managing environmental resource commons and insufficient on its own.
So what would make carbon monitoring data better?
One way to think about this is to remain attentive to financial incentives and business pressures in carbon accounting tools. Because of increasing regulations and new state and corporate climate commitments, carbon data is increasingly lucrative. This can incentivize investment in research and new monitoring tools, but it can also incentivize unregulated markets for datasets and environmental dashboards which may or may not actually lead to reduced emissions.
Carbon accounting relies out of necessity on a hodgepodge of sources, raising concerns about methodology and certainty. Trust, accountability, and iterative development are necessary to overcome these limitations, but they can be hard to secure in the context of proprietary software, cloud platforms, and opaque machine learning models. Public standards and nonprofit or open-source development can provide useful alternatives. Accounting systems can and should also be designed with long term maintenance, accessibility, sovereignty, and surveillance concerns in mind, and with an eye to the climate impacts of the digital accounting systems themselves.
What else should tech actors looking to address climate change keep in mind?
At the end of the day, climate change is more a political problem than a data problem. Carbon accounts can provide necessary snapshots of operations and important heuristic tools for companies, institutions, and governments. But in and of themselves, carbon inventories and net-zero campaigns do little to guarantee emissions reductions, let alone holistic shifts towards environmental and social responsibility beyond the ICT sector. Bringing the public into carbon governance, and creating transparent forms of monitoring and vetting data, will help bring accountability into the tools of accounting.
Image by Critical Carbon Computing Collective. CC-BY-NC 4.0.
Computers lie at the heart of climate politics. Computationally-intensive models frame and constitute how governing bodies and public perception alike understands the problem of climate change, while a growing number of digital tools and systems are positioned as the solution to the climate crisis. The urgency of global warming isn’t in dispute but how the computing tech industry is setting the terms and horizons for mega climate change projects should be understood and recalibrated.
This inquiry attends to the material realities of computing technologies–including labor, supply chains, and digital infrastructures–and to the widespread faith that digital solutions and technological management are the best and only responses to climate change. Even when carbon reporting mechanisms and software tooling intended to reduce GHG emissions are most effective, it is unclear if:
1) carbon cost numbers are accurate across fully supply chains
2) measuring and reporting emissions leads to substantial reductions in GHG emissions (measurement is not the same as action),
and 3) an outsized focus on decarbonization casts aside or, at the very least, deprioritizes other crucial environmental and social factors.
We call this paradigm carbon computing. It describes the massive enterprise of using computers to manage the climate system—from measuring individual impacts to adjudicating global climate politics. It is an increasingly core, if under-recognized, feature of climate management and governance. As an enterprise, it makes use of technological expertise and raw computing power to document and archive the past, manage the present, and anticipate the future. It incorporates computing technologies that range from simulation software and supercomputers to accounting sheets, blockchains, and machine learning systems. It is also, at times, paradoxical, as digital networks and infrastructures make up a growing share of global energy and carbon budgets.
We call this paradigm carbon computing. It describes the massive enterprise of using computers to manage the climate system–from measuring individual impacts to adjudicating global climate politics.
Understanding carbon computing requires in-depth historical context, and deeper critical and ethnographic engagements with these dynamics. Techno-solutionist approaches alone won’t cut it. It is imperative to include and learn from other perspectives and forms of knowledge, and carefully weigh what society stands to win–and lose–every time we turn to digital tools and networks for solutions. This belief in inclusive, contextualized, and critical engagement with digital carbon narratives motivates our collective.
Our Project
The Critical Carbon Computing Collective (4C) is a group of researchers, academics, activists, and artists working to contextualize and demystify the proliferation of technologically-oriented proposals that currently govern knowledge and resources in the climate system. Two core questions drive our work: (1) How is computing tech implicated in the unsustainable resource use that contributes to climate change? and (2) What role should computing tech play in strategies for environmentally just futures? In developing our responses, we do not take “tech” or technology to be a stable category or even one with a single, self-evident definition. We do, however, emphasize the role of the Information and Communications Technology (ICT) sector, given that industry’s increasing climate costs and simultaneous dominance in climate accounting, reporting, and management.
Most members of the 4C work at large universities, and some of us have connections with the work of industry-funded nonprofits in these areas; we therefore don’t claim to speak from outside of these funding structures or power dynamics. However, we each also have connections to groups and movements not represented in conversations at these institutions and understand the importance of elevating independent and under-represented perspectives. We aim to use our multidisciplinary insights to inform ongoing debates and build connections across work that risks being taken as purely technical. This would be a mistake.
As historians and ethnographers of computing and the environment, we hold that:
Tech and climate change are deeply entwined. Computers and ICT both generate a growing portion of climate emissions and are viewed by industry and policymakers as the solutions to climate challenges.
Dominant framings of computing systems and new distractions from computationally-focused approaches to carbon risk upholding the status quo.
The relationship between energy and computation needs to be critically understood and historically situated.
These conversations need to extend beyond industry and academia.
Coalition Building
This project is forming at a time where there is a larger push from industry, advocacy, academic, and policy collaborators to account for and alleviate environmental harms, some of which are caused by computing itself. Much of this effort takes the form of nonprofit organizations seeking to address the relationship between climate change and the ICT sector. Academic and industry-connected researchers at Climate Change AI examine not only the climate impacts of machine learning in a global context, but the ways that machine learning might be used to mitigate climate change. Green Software Foundation, funded by industry steering partners and hosted by the Linux Foundation, is attempting to build tools and provide resources for decarbonizing the ICT industry, changing the culture of how software is built in order to prioritize sustainability. The Green Web Foundation is focused on building a fossil-free internet by 2030. 4C acknowledges such ongoing partnerships as important avenues for academics, civil society, and the wider public to redirect machine learning applications toward climate mitigation and to rebuild alternative practices of hardware and software development.
4C aims to contribute to these kinds of pragmatic technical proposals by expanding their scope of analysis to consider the historical and ongoing labor, resource, and social struggles implicated in such interventions. Rather than viewing carbon computing as the route to promissory futures or guaranteed sustainable outcomes, our goal is to recenter equity and justice in tech-oriented proposals around carbon in the present.
Generating Resources
For our public launch, we are releasing a set of four short guides designed to break down and demystify common terms, practices, and debates at the intersection of climate and ICT. By making these documents concise and accessible, we hope they will be a resource for circulating through other communities, opening up critical conversations about tech and climate.
Carbon Accounting outlines how the tech sector contributes to (and reinforces) specific ways of monitoring and governing climate actors, and raises the prospect of new monopolies around the tools and data that other sectors need to reduce their carbon emissions.
Policy and Tech discusses how states, corporations, and workers are setting standards, rules, and expectations around climate tech, and how tech companies are maneuvering through these policies (and policy gaps).
Computing Net Zero unpacks the power of language in carbon computing.Words in computationally intensive climate projects are particularly prone to manipulations because of their apparent technical complexity. The computational definition of the term Net Zero depends on the specific context in which it is being used, and calculations can conceal who wins, loses, and profits from combinations of new tech, offsetting, and continued emissions.
Labor emphasizes the roles of organized labor and environmental justice coalitions in advocating for and creating more sustainable computing practices and aims to move beyond carbon emissions reporting as the only mechanism for understanding the relationship between climate action and the ICT industry. Labor provides a lens for including the larger supply chains and marginalized communities that are often left out of carbon-centric discussions of climate change and computing technologies.
In the coming years, our group will publish more of these guides, visual diagrams, reports, and white papers on carbon computing alongside workshops and public programming. We invite those interested in these questions and for future collaborations to reach out: info@criticalcarboncomputing.org.
Contents
