New evidence that the climate system may be especially sensitive to the build-up of greenhouse gases and that humans are doing a poor job of controlling their effluent has animated discussions around the possibility of offsetting the human impact on climate through ‘geoengineering'. Nearly all assessments of geoengineering have concluded that the option, while ridden with flaws and unknown side effects, is intriguing because of its low cost and the ability for one or a few nations to geoengineer the planet without cooperation from others. I argue that norms to govern deployment of geoengineering systems will be needed soon. The standard instruments for establishing such norms, such as treaties, are unlikely to be effective in constraining geoengineers because the interests of key players diverge and it is relatively easy for countries to avoid inconvenient international commitments and act unilaterally. Instead, efforts to craft new norms ‘bottom up' will be more effective. Such an approach, which would change the underlying interests of key countries and thus make them more willing to adopt binding norms in the future, will require active, open research programmes and assessments of geoengineering. Meaningful research may also require actual trial deployment of geoengineering systems so that norms are informed by relevant experience and command respect through use.