Geoengineering Governance: What Each Position Is Protecting

In 2012, a Canadian businessman named Russ George dumped approximately 100 tonnes of iron sulfate into the Pacific Ocean off the coast of British Columbia, seeding a plankton bloom covering roughly 10,000 square kilometers. His stated goal was ocean iron fertilization — triggering plankton growth that would absorb atmospheric CO₂ and then sink to the ocean floor when the plankton died, sequestering carbon at depth. He had not notified Environment Canada or the International Maritime Organization. He had not conducted an environmental impact assessment. He had secured loose backing from the Haida Nation, which later distanced itself from the operation. The international response was swift and largely appalled — this was one of the largest geoengineering experiments ever conducted, done unilaterally by a single actor with no oversight and no liability framework. Nothing that happened was technically illegal under international law, because no international law covered it.

Ten years later, the governance problem has deepened rather than resolved. The IPCC's Sixth Assessment Report, published in stages from 2021 to 2022, embedded bioenergy with carbon capture and storage — BECCS — into the majority of its modeled pathways for keeping warming below 1.5°C. BECCS works in theory: grow crops or plant trees, burn them for energy, capture and sequester the CO₂ released, achieving net negative emissions. In the models, the technology absorbs billions of tonnes of atmospheric carbon annually by mid-century. On actual land, the technology requires agricultural area roughly equivalent to one to two times the size of India, much of it in the tropics and subtropics, much of it currently occupied by smallholder farmers, pastoral communities, and Indigenous peoples. No governance framework for BECCS deployment at that scale exists. The IPCC scenarios assume it happens; the governance for making it happen justly does not.

Meanwhile, at Harvard's Solar Geoengineering Research Program, the planned SCoPEx experiment — a balloon lofting instruments to the stratosphere to test the reflective properties of calcium carbonate particles — was cancelled in 2021 not by government regulators but by the program itself, responding to objections from a Sámi advisory board and Indigenous groups in Sweden who had not been adequately consulted. The scientists had sought governance before proceeding. The governance process, applied seriously, produced a veto they accepted. The episode raised a question the governance debate rarely names directly: if meaningful consent is operationalized, whose consent is sufficient, and who decides that?

These stories are not about whether geoengineering should happen. They are about something prior — about who holds the key to the planet's thermostat, what process legitimizes turning it, and what recourse exists for those who bear the consequences of decisions they had no role in making.

What the multilateral treaty architecture position is protecting

People who argue that geoengineering governance requires formal international treaty frameworks — negotiated, ratified, and binding — are protecting a specific principle: that interventions with global consequences require global consent, and that consent at this scale can only be operationalized through the mechanisms international law has developed for exactly this purpose.

The argument has both procedural and substantive components. Procedurally, the history of global commons governance — the Law of the Sea, the Outer Space Treaty, the Antarctic Treaty System, the Convention on Biological Diversity — shows that internationally binding frameworks, however imperfect, create accountability structures that voluntary norms cannot. A treaty has signatories who can be held in violation. It has dispute resolution mechanisms. It can assign liability for harm. A code of scientific ethics or a set of research guidelines does not. The multilateral treaty position holds that geoengineering's potential for irreversible planetary-scale effects places it categorically in the domain where binding frameworks are not optional — they are the minimum threshold for legitimate action.

Substantively, the position argues that the most important governance decision is not whether to deploy but who decides to deploy and on what basis. The termination shock problem — the catastrophic warming rebound that would follow sudden cessation of a stratospheric aerosol injection program — means that governance of deployment is simultaneously governance of everything that follows: maintenance, modification, and termination. No current international institution is designed to guarantee continuity of a complex technical program across political disruptions, state failures, and geopolitical realignments that will inevitably occur over the multi-decade to multi-century timescales a deployed program would require. A treaty framework would need to create institutions with that continuity guarantee. That is an extraordinary institutional challenge, and proponents of the multilateral approach are the ones most willing to name what building such a framework would actually require.

What this position costs: binding multilateral governance for geoengineering has never been achieved. The 2023 UN Environment Assembly resolution proposing even a modest expert scientific panel was blocked — primarily by the United States. The states with the most capacity to conduct research have the least incentive to submit to governance that would constrain their options; the states most exposed to geoengineering's risks have the least leverage to demand it. Waiting for a treaty framework may mean waiting indefinitely. The multilateral position can become, in practice, a position against research itself — since research proceeds in the absence of adequate governance and the governance may never arrive.

What the BECCS and land rights position is protecting

A distinct set of voices — smallholder farming organizations, Indigenous land rights advocates, food sovereignty researchers, and a minority of climate scientists — are raising a governance problem that receives far less attention than stratospheric aerosol injection: the governance of carbon dioxide removal at the scale embedded in mainstream climate scenarios.

BECCS is not a speculative technology. It exists. Drax Power Station in the United Kingdom co-fires biomass and operates carbon capture units, generating renewable-credited power. What does not exist is BECCS at the scale the IPCC pathways assume: one to ten gigatonnes of CO₂ removed annually by mid-century, requiring energy crop plantations covering hundreds of millions of hectares. The governance gap here is different from the SAI governance gap. It is not about consent to a single actor turning a dial. It is about the cumulative effect of thousands of land-use decisions — national biofuel mandates, carbon credit markets, corporate sustainability commitments — aggregating to the displacement of communities from land their families have farmed for generations, the conversion of biodiverse ecosystems to monoculture energy plantations, and the routing of food production capacity toward atmospheric carbon management.

The land rights position notes that this governance gap is embedded in international climate policy in ways the SAI debate is not. BECCS is in the IPCC models. Countries count BECCS credits toward their nationally determined contributions under the Paris Agreement. Carbon markets certify BECCS offsets. The governance frameworks that exist — the voluntary carbon standard, national land-use planning regimes, the UN Declaration on the Rights of Indigenous Peoples — are manifestly insufficient to protect communities from displacement by projects that are financially rewarded by international climate commitments. The land rights position is not anti-climate. It is arguing that the pathway to net-zero has land tenure consequences that are not being governed as climate consequences.

Ocean-based CDR — alkalinity enhancement, iron fertilization, kelp farming — raises parallel governance questions under the London Protocol. The 2013 amendment to the London Protocol restricting ocean fertilization was blocked from entering into force for nearly a decade due to insufficient ratifications. The governance architecture for ocean CDR is only marginally more developed than for SAI. Each CDR technology has a distinct land, water, or ocean footprint, a distinct liability question, and a distinct affected community — and the governance debate too often treats CDR as a single undifferentiated alternative to more frightening interventions.

What this position costs: naming BECCS land displacement as a geoengineering governance problem complicates the framing that allows mitigation to proceed. If BECCS at scale is ungovernable justly, and BECCS is embedded in every 1.5°C pathway, then the governance critique threatens not just a supplementary technology but the dominant model for how humanity gets out of the climate crisis. That is a costly implication, and the position sometimes avoids following its argument to its conclusion.

What the scientific community self-governance position is protecting

A third position — held by many researchers working in geoengineering science and some science policy scholars — argues that formal treaty governance is the wrong model for governing research, and that scientific community norms, research ethics frameworks, and voluntary codes of conduct are both more practical and more appropriate for the current stage of technological development.

The argument is not that governance is unnecessary. It is that governance should match the stage of the technology. Geoengineering research is, today, mostly modelling, laboratory work, and very small-scale field experiments measuring stratospheric chemistry at scales measured in kilograms of particles. Subjecting this stage of research to binding treaty-level governance is equivalent to requiring nuclear weapon treaties before the Chicago Pile-1 experiment in 1942 — governance that would, in practice, prevent the research from which governance decisions must eventually be made.

The SCoPEx case is, for this position, instructive rather than cautionary. The Harvard program engaged a Sámi advisory committee. The committee raised concerns. The program paused its experiments and ultimately cancelled the Swedish campaign. This is the scientific community governance model working: researchers with real autonomy choosing to seek consent, engaging meaningfully with affected communities, accepting the outcome. It is not perfect governance — the advisory committee was not a veto-holding representative body — but it represents a mode of accountable research that treaty governance is unlikely to produce faster or more legitimately for small-scale experiments.

The self-governance position also notes a structural asymmetry: the scientists most likely to operate under voluntary codes of conduct and seek community consent are the ones at institutions that take research ethics seriously — primarily universities and public research labs in democracies. The actors most likely to conduct geoengineering research without adequate governance are national military programs, private sector operators, and researchers at institutions with less oversight. Restricting the former through formal governance while the latter operate in a vacuum may simply shift research to less accountable actors.

What this position costs: voluntary norms cannot bind non-signatories. The scientific community's ability to govern itself ends at the boundary of its own members. The Russ George episode happened outside any scientific institution. A wealthy nation that decides to deploy for emergency reasons will not be stopped by a research ethics code. Self-governance is adequate for governing responsible researchers; it is inadequate for governing the actors most likely to cause the harms the governance is meant to prevent.

What the emergency unilateral action position is protecting

A fourth position — rarely stated openly but structurally present in climate policy debate — holds that if warming crosses critical thresholds and multilateral governance has not produced a deployment framework, individual states or coalitions retain both the right and potentially the obligation to deploy unilaterally. This position is protecting a principle that is easy to condemn abstractly and harder to argue with when the alternative is named concretely: if Greenland ice loss has triggered irreversible sea-level rise threatening hundreds of millions of people, and international institutions have produced nothing actionable, is a state prohibited from deploying a technology that could slow the warming driving the loss?

The emergency unilateral position does not usually advocate for unilateral action as a first preference. It argues that the possibility of unilateral action should be acknowledged in governance design — that governance frameworks which cannot contemplate emergency exceptions are frameworks that major states will simply ignore when emergencies arrive. The governance design question then becomes: what conditions would justify unilateral action, what liability regime would apply, and what consultation requirements would be minimum — rather than: is unilateral action ever permissible.

There are historical analogues. The Montreal Protocol's unilateral phase-out of CFCs by the United States and European Community preceded the treaty and helped create the conditions for it. Climate action itself has proceeded largely through voluntary national commitments rather than binding obligations, because binding obligations were unachievable. The emergency position argues that a governance architecture that cannot accommodate state action in crisis is not a governance architecture — it is an aspiration.

What this position costs: acknowledging unilateral emergency rights creates a legitimacy problem that is very difficult to contain. The conditions that justify emergency action are determined by the actor claiming the emergency. The difference between "emergency deployment to prevent catastrophic warming" and "deployment calibrated to the deployer's climate preferences at the expense of monsoon systems in countries the deployer has no accountability to" is real but not self-enforcing. Emergency exceptions tend to expand to fill the space of politically desired actions. This is the strongest structural argument against building emergency unilateralism into the governance framework, even for those who accept it might be justified in some circumstances.

What the polycentric governance realist position is protecting

A fifth position — increasingly prominent in climate governance scholarship — argues that the binary choice between "wait for a formal treaty" and "proceed without governance" is a false choice, and that the actual governance architecture available is a network of overlapping national regulations, scientific body guidelines, bilateral agreements, regional frameworks, and insurance and liability markets. This position is protecting practical progress on governance in a world where the institutional capacity for binding multilateral frameworks has demonstrably failed for every major environmental challenge of the past three decades.

The polycentric argument, developed most rigorously by scholars like Elinor Ostrom and extended to geoengineering governance by researchers including Jesse Reynolds and Duncan McLaren, holds that complex governance problems are not solved by single authoritative institutions but by layered systems of overlapping rules, multiple accountability mechanisms, and redundant oversight. Geoengineering governance, on this view, already has components: the London Protocol regulates ocean interventions. National aviation regulations govern high-altitude flights. Environmental impact assessment requirements apply in most jurisdictions. The UNFCCC provides a forum, however inadequate, for discussion. Building out these components — rather than waiting for a comprehensive framework that may never arrive — is both achievable and meaningfully accountable.

The realist position is also specifically attentive to the private actor problem. The declining cost of high-altitude aircraft and aerosol dispersal technology means that within years to decades, the capacity for stratospheric aerosol injection may be within reach of private actors — a sophisticated nation-state, a very wealthy individual, a private foundation committed to emergency climate action. No polycentric governance framework adequately addresses private actor deployment, but neither does a treaty framework that has not been negotiated. The realist position argues for building practical accountability mechanisms now rather than designing ideal frameworks that do not exist.

What this position costs: polycentric governance that is achievable may not be adequate governance. A network of national regulations and scientific ethics codes cannot provide the consent mechanisms, liability regimes, or continuity guarantees that deployment at scale would require. Polycentric governance may be the best available option while being substantially insufficient — and accepting it as adequate risks foreclosing the political urgency that building adequate multilateral frameworks would require.

Where the debate actually is

The geoengineering governance debate has a curious structure: all five positions agree that governance is necessary, and none of them has produced it. The multilateral position names what adequate governance would require and acknowledges it does not exist. The BECCS land rights position names a governance gap already embedded in international climate commitments and watches it widen. The scientific self-governance position produces ethics codes that bind only willing actors. The emergency unilateral position acknowledges that sovereign states will act outside any framework when they judge the stakes sufficient. The polycentric realist position builds architecture that may be better than nothing while acknowledging it falls short of what the problem requires.

The specific disputes within this general paralysis are worth naming. The consent problem is the deepest: there is no mechanism by which the consent of affected communities — not states, but the communities within states whose rainfall, food production, and coastal security would be altered — can be operationalized in any governance framework currently under discussion. State consent is not community consent. Governments of small island states are not the same as the communities living there, and even if they were, their consent to SAI deployment does not extend to communities in Bangladesh, the Sahel, or Indonesia whose governments have different interests.

The liability vacuum is the second structural gap. No international law currently assigns liability for climate harm caused by a geoengineering deployment — the termination shock caused by abandoning a program, the monsoon disruption caused by poorly calibrated aerosol injection, the ocean acidification that CDR interventions fail to address. The actors most capable of deployment are also the actors with the most leverage to resist liability frameworks. The actors most exposed to harm are the ones least able to enforce them.

What is genuinely unresolvable by better governance design — and important to name — is the prior political question: whether the international community that failed to govern climate change itself is capable of governing the interventions proposed to address it. The geoengineering governance debate proceeds as if better institutional design could produce better outcomes without confronting the structural conditions — economic interests, geopolitical competition, sovereignty norms — that have prevented adequate climate governance for fifty years.