The importance of some uncertainties is dependent on the deployment scenario envisioned. Our chosen scenario is just one possible version of SAI, but is a useful central case of well-managed SAI, with hemispheric balance of injection, a gradual ramp up of injection magnitude, and no interruptions. The decision-relevance of engineering uncertainties depends on the start date of SAI. We assume 2035 as a start date not because this is when we think deployment should or will begin, but because this is around the earliest that deployment could proceed.
To make the scale of this project tractable — and to best utilize Reflective's internal expertise — we narrowed the scope to include only technical (i.e. physical science and engineering) uncertainties. Societal, governance, and geopolitical uncertainties will be indispensable parts of any eventual decision-making framework for SAI. We encourage other groups to repeat this process for those uncertainties, and we will consider ways of incorporating these aspects into our database in future.
The database is a living resource, and we welcome the expert community to give us feedback which will help us to continually refine and expand its content. We are also undertaking significant engagement to solicit expert input through 2026, and will submit a paper for peer review detailing our process and findings.
Reflective will analyze new research on a rolling basis to determine if updates are merited. Any updates made as a result of new literature will be documented monthly along with the changes made from community feedback.
This resource is designed to help those inside the SAI research community align on and prioritize solving the most decision-relevant questions, and to help give those outside of it a clear view of the current scientific understanding. Please see our About page for more information.
This project serves as an initial foundation for a transparent, prioritized, stage-gated SAI research roadmap. We are now beginning the process of refining and expanding the database towards a second version.
We drew on team expertise, literature reviews, and expert consultations to arrive at these categorizations, and conducted the consultative process described here.