The irony in discussions about climate engineering is that, while society considers its merits, the process itself is already in full swing. With vast amounts of heat-trapping molecules released each day into the atmosphere, humans are deliberately altering the planet’s climate in unpredictable ways. The magnitude of the resulting climate change is worryingly uncertain. Even more uncertain are the physical, social and economic side effects of global warming. There is every reason to believe that, by and large, they will be harmful.

Why, then, is the idea that future generations could use a little science and engineering to deliberately cool the world so controversial? The answer, of course, is that the cure could be worse than the disease.

Adding sulphate into the high atmosphere, for example, is one of a broad range of geoengineering techniques proposed in response to the warming driven by greenhouse gases. If the technique helps to destroy the ozone layer or increases drought risk in vulnerable regions, then there is a strong argument not to do it.

Scientists are not solely responsible for the problem of global warming. And many argue forcefully that they should be wary of simply replacing one evil with another. Even scientists who are directly involved in geoengineering studies often admit that they do not like the prospect of their research becoming a real-world necessity.

There are some aspects of geoengineering on which all can agree. It should not distract from efforts to curb emissions. An effective political agreement to radically reduce greenhouse-gas emissions, such as that being discussed this week at the United Nations climate-change conference in Lima, must take priority over speculative notions to instead tinker with the atmosphere to meet climate goals.

In fact, geoengineering practices that do pose significant further risk to the environment must be prohibited, if necessary by international law. After all, no single nation — let alone any faction of science — can assume the right to deliberately modify the physical set-up of the planet.

Large-scale and possibly irreversible atmospheric interventions are clearly beyond what is scientifically and ethically justified. But apart from behemoth plans (which nobody is seriously promoting), there are many more limited climate-engineering options that do deserve serious consideration and study. To that end, leading scientific societies are this week discussing a set of guiding principles for responsible field experiments (see page 20). It would greatly enhance the credibility of the field if it could adopt such a scientific code of conduct.

The geoengineering option that seems simplest — scraping carbon dioxide from the air and permanently locking it somewhere secure — is already being intensely investigated. Carbon capture and storage technology is now widely considered to be safe, but technical and financial challenges limit its wide-scale adoption. Because the world’s appetite for fossil fuels has not yet peaked, it is as important to encourage and fund research on the carbon capture side of the technology as on the carbon storage aspect. But whether this technology will really help to fight climate change depends on political governance, such as whether it becomes standard in the international energy sector to fit new coal-burning plants with carbon capture equipment.

In its last report, the Intergovernmental Panel on Climate Change (IPCC) left little doubt that some form of geoengineering (or ‘negative emissions’, in IPCC language) will probably be needed to meet the goal of limiting global warming to 2 °C. Having delivered its fifth full climate assessment report since 1990, the IPCC is considering adopting a new role in the future. If the group were to switch to more-focused, trimmed-down reports, delivered on demand, a special report on climate engineering might be the perfect place to start. Meanwhile, researchers should work fast to clear the way for more responsible research, even if responsible action means that its results will never be needed.