This post contains the summary and recommendations of a report on a workshop run by the Centre for the Study of Existential Risk^[1] on managing the contribution of Solar Radiation Modification (SRM), also known as solar geoengineering or climate engineering, and Climate Change to Global Catastrophic Risk. The report is linked here. 

In general, I think this will be useful for those interested in SRM (and to a lesser extent climate change) and GCR, as well as those interested in methodologies in GCR work. The sorts of foresight practices we used, both ParEvo scenarios and participatory workshops to iterate on these scenarios, whilst used in GCR contexts, seem uncommon in EA, so I hope this is a useful example of what one output from these processes can look like. Moreover, we tried to unpick the factors and their dependencies to try and understand the complexities of GCR, which also seems to often not be done in EA. I am hoping this will be a useful example that may allow for some critical assessment of methods in GCR studies as well. 

It should be noted the intended audience for the report is not necessarily an EA audience. This audience is important, as many of the recommendations are not suggestions for what is best for an impartial funder to do on the margin, but rather are suggestions for what we deem best for those involved in SRM to do to reduce GCR (thus the theory of change is different to research aimed at an EA audience). The workshop, and my last year of research in general, has had a theory of change of trying to get GCR concerns embedded in the already existent SRM community, and this workshop report represents one output of this process. Further work on many of the ideas explored here will be released over the next few months/year as the work from my project gets published. 

Summary

This report presents key findings from a workshop on managing the contribution of Solar Radiation Modification (SRM) and Climate Change to Global Catastrophic Risk (GCR) that was hosted by Gideon Futerman and SJ Beard at the Centre for the Study of Existential Risk on March 28th and 29th 2023. The workshop was informed by a participatory futures exercise using the ParEvo technique that explored futures for SRM and SRM governance between 2030 and 2050, which some workshop participants took part in. Initial results of the exercise were shared with workshop participants and full results will be published separately.

• Participants emphasised that SRM can both contribute to and mitigate GCR. However, at present, high levels of uncertainty make it difficult to perform a complete assessment of risk and so this report will not attempt to assess whether the research and deployment of SRM will raise or lower this risk. Instead, it focuses on participants’ exploration of the different pathways to global catastrophes and the role SRM might play in them, the factors they saw as influencing the interaction between SRM and GCR, and their proposals for improving the governance of SRM and SRM research.
• Whilst in most cases it doesn't appear to, participants noted that reducing GCR could sometimes conflict with other priorities. For example, downplaying the importance of SRM and avoiding its politicisation could be important for reducing GCR, but also obscures issues of power and reduces democratic control over SRM. Therefore SRM-relevant actions may interact with GCR in many ways; not just contributing or mitigating risk but also changing its nature. 
• Participants identified many ways in which SRM may interact with GCR. Discussions of possible pathways towards global catastrophe typically involved interstate conflict, termination shock, or catastrophic climate impacts, while discussion of pathways away from global catastrophe involved the reduction of climate damages by SRM deployment.
• Many other factors were seen as influencing SRM's interaction with GCR. These included the type of deployment and governance, the perception of SRM’s impacts and importance among politicians and publics, securitisation and militarisation, geopolitics, extreme weather, knowledge networks, wealthy individuals and corporations, and developments in artificial intelligence. Whether these interactions are net contributors or mitigators of GCR will depend on how these factors evolve and interact, as summarised in Figure 1.
• All these factors are contingent on human actions, perceptions, and behaviour. Ultimately, social, political, and geopolitical systems will be as important as physical systems in determining whether SRM reduces or increases GCR.
• While it was generally felt that the current knowledge network around SRM has limited influence, participants also believed that there were actions that could be taken to reliably reduce GCR and that this ought to be a consideration in research and policy. 

Below are some proposals to improve SRM research, policy, and governance. These were written by the authors and, whilst informed by the discussions at the workshop, they are not necessarily endorsed by the workshop attendees. Further work is necessary to refine the workshop’s risk identification and come up with more robust research and governance ideas. These recommendations are targeted at those individuals and institutions who may be interested in their actions related to SRM reducing (or not increasing) GCR.

PROPOSALS TO IMPROVE POLICY AND GOVERNANCE

• If governance frameworks are to reduce GCR (whether they allow or prohibit SRM deployment) they are likely to require widespread, sustained buy-in. Establishing the legitimacy of, and interest in, such frameworks ought to be a high priority and should include a strong diversity of voices to allow for the broadest legitimacy of any decisions. Governance frameworks that minimise the potential for unproductive geopolitical competition over the technology and its outcomes are highly desirable.
• SRM research can and should be used to foster multilateralism, including voices from around the world. Unilateralism, including with very large research projects, may be dangerous as it likely increases the probability of conflict or termination by reducing trust and increasing competition. Building bridges and trust around SRM through joint research initiatives, such as between the USA and China, could help to encourage multilateralism.
• Researchers should seek to lower the barriers to a wider range of stakeholders participating in discussions on SRM, both from groups and states supportive and opposed to SRM, to promote more equitable discourse.
• Actors in the space should be careful not to overstate the importance, impacts and controllability of SRM. Narrative shaping can help to encourage policymakers to perceive SRM as a limited tool over which no individual state can exert total control and over which loss of some power may be of only limited significance.
• Securitisation and militarisation of SRM may be particularly risky. To avoid this national security establishments ought to refrain from attempting to securitise SRM and militaries commit to playing no role in SRM research. Actions by the knowledge network around SRM, such as narrative shaping,activism or policy advice may have a role in reducing this risk, albeit a limited one.
• While there may be good arguments for bringing discussions around SRM into the political realm, this may also have dangerous consequences in promoting geopolitical competition over SRM. Actors engaging in political discourse over SRM should be mindful of this danger.

PROPOSALS FOR FURTHER RESEARCH

• Foresight activities focused on possible consequences of actions related to SRM (both in terms of suppressing and promoting it) may help bring about positive futures, as would better understanding of whether and how those interested in reduced GCR can steer such actions away from some of the GCR pathways identified here.
• More modelling of plausible ‘wild’ deployment scenarios is needed. Whilst workshop participants noted that there had been a start at this, better modelling, including of regional impacts and impacts on critical systems, would give more realistic implications about these scenarios, including constantly changing SAI deployment patterns, uncoordinated regional MCB, or overcooling; however, considerable refinement and exploration of the possibility space will be essential before such research is undertaken. This modelling should consider not just the median but also the plausible worstcase impacts of these scenarios.
• More research and modelling are needed to explore the impacts, on both Earth and human systems and their interactions, of termination scenarios, and how we might adapt to them. This should include explicit attempts to understand the worst-case scenarios.
• More research is needed on the plausible worst-case scenarios of climate change in a range of warming scenarios to carry out a more complete risk analysis of the interaction of SRM and GCR.
• Exploration of cascading impacts in the above sets of scenarios and other complex risk pathways, should be engaged in. This will likely involve methodologies beyond conventional climate and economic modelling and may require methodological innovation.
• Researchers should work with social and behavioural scientists to understand what public responses and attitudes towards SRM might be, especially the potential for sudden dynamic shifts in public attitudes, which workshop participants noted could be especially dangerous.
• Researchers should work with political scientists and international relations experts to understand the geopolitical escalation dynamics that may lead disagreements over SRM to lead to conflict. These experts, together with legal and policy experts, may also help us to understand the full range of governance mechanisms available for SRM, including how frameworks that could develop around the regional deployment of SRM may be scaled up.
• Attention should be given to the potential, and limits, of attribution studies for assessing the impacts of SRM (using existing attribution work in climate change as case studies). This may help mitigate risks associated with perceptions of SRMs impacts, although whether such science could be developed before SRM deployment is unclear, as is the extent to which this would impact perceptions.

1. ^{^}

   I (Gideon Futerman) do not represent CSER and am not employed by them, although did carry out this workshop whilst I was a visiting researcher at CSER