We might be able to slow down climate change by injecting sulfur dioxide aerosol into the stratosphere using a fleet of specially-designed aircraft. Once up there, the aerosol would mimic the effect of a major volcanic eruption, cooling the planet. This is called stratospheric aerosol injection (SAI), and it would give us something like a thermostat for the whole planet.

This system would not perfectly counteract the effects of greenhouse gases, and would probably have side effects, such as effects on precipitation in some parts of the world (although these are still debated among scientists ). But perhaps a more pressing issue is political. The effects of stratospheric aerosol injection cannot be limited to just one country or world region. It's either we modify the climate of the whole world, or not at all. So who gets to control the global thermostat? Different preferences for where to set the thermostat, concerns about side effects, and worries about the use of geoengineering as a weapon or form of political leverage, will make this a very high-stakes discussion. A "climate clash" in which rival countries implement their own SAI projects without coordination could exacerbate the negative effects of SAI, while limiting the positive benefits.

While SAI would represent a new step in humanity's relationship with the environment, it would not be the first time we have modified the earth’s systems on a large scale. One good example of an earth system intervention technology which already exists is the large-scale management of freshwater systems, using dams, canals, weirs, and similar infrastructure. When the systems being modified cross political boundaries, the politics—which scholars call “hydropolitics”—can be complex and difficult.

Surveying the historical hydropolitics of ten major transboundary freshwater systems, we find a few important things which may have relevance for SAI.

The most important might be that success is far from a given. Of the ten freshwater systems we studied, only about half had outcomes which are both cooperative, and effective at incorporating good science into the river systems' governance. For those interested in placing a ban or moratorium on SAI, the rate of success is even more limited. Only one freshwater system we studied (the Great Lakes) successfully passed an agreement to limit interventions in the system, rather than just to ensure cooperation in existing large-scale interventions.

In cases where cooperation was not achieved, the consequences could be dire: Either de-facto domination of the river system, or unilateral action by a regional power (China on the Mekong or Turkey on the Tigiris and Euphrates), or outright conflict (Egypt and Ethiopia on the Nile). The river systems that have successful agreements tend to be those passing through relatively few countries, such as the Columbia and Vuoksi watersheds; or those where the countries in the watershed have pre-existing positive relations, such as in the Rhine Basin.

Common risks and barriers for the establishment of good freshwater management treaties include a lack of scientific consensus; an inability to get all countries in the basin to sign onto an agreement (China and Afghanistan, who are not parties to the Indus Water Treaty, have both begun damming projects in the tributaries of the Indus), or the lack of a clear "critical juncture”—an incident that brings enough political attention to the problem of governing a freshwater basin.

The single biggest determinants of effective cooperation in a watershed are two-fold: Strong, neutral institutions, and time. It can take decades of both political and scientific work to resolve hydro political disputes. The Great Lakes took over a century from the first technological intervention in the basin (The Chicago Canal) to the establishment of a strong international agreement. The most consistent pathway to success is to start early, creating respected, formal institutions to facilitate cooperation and build a shared set of scientific facts. Then, when the fate of the watershed moves up political agendas, all parties have a strong framework to work from.

For SAI, this suggests some clear paths forward, but also some dilemmas. First, it suggests that we need to start building the capacity to govern SAI sooner rather than later. If we wait until there is a global conflict over this technology to try and established shared scientific knowledge or political institutions, it will be too late. Some level of compensation or guarantees will probably be required for various parties to any SAI treaty, to make it internationally viable. Most worrying: We will probably not be able to make major progress until something—either a climate disaster or maybe a unilateral SAI deployment—forces this issue onto global political agendas.

A major dilemma comes from the fact that institutions to manage international rivers tend to have an internal incentive to increase, rather than decrease, technological interventions. So if the goal is to stop SAI from being deployed, then we probably need institutions to allow us to do so. But those same institutions might be the ones who eventually implement SAI rather than opposing it.

Either way, the risks of this are high, and in many ways stratopolitics—the atmospheric equivalent of hydropolitics, will be very difficult to get right. Scholars working on this area should not assume that SAI deployments will be scientifically rational in all scenarios. Building strong, objective institutions now to manage these deployments might be a good idea—but this will only work if these institutions are set up from the start to be able to block SAI as sell as just to facilitate it, to enable real democratic deliberation on this critical question.