A roadmap for testing the proposal

Start where errors can be found cheaply

The first task is radiative closure: independent calculations should agree on what sunlight is transmitted, reflected or absorbed, where the reduction occurs and how uncertainty propagates. Setting the proposed attenuation to zero should recover the ordinary reference calculation. A result that depends on a unit error or a numerical grid is not ready for atmospheric interpretation.

Chapter 25 then discusses laboratory thermal and cloud-chamber studies. These would test restricted mechanisms under controlled conditions. A bench-top temperature response would not establish outdoor cooling, and a chamber droplet result would not establish rainfall. Each method answers only the question it can actually reproduce.

Challenge models with observations and each other

Historical eclipses offer real atmospheric cases for model testing. The book proposes separating development cases from independent evaluation cases, while checking radiation, temperature, wind, cloud and recovery. Reproducing one attractive temperature curve would not demonstrate that the whole causal chain is represented correctly.

Chapter 24 introduces ASES-MIP as a proposed model intercomparison programme. Its concept is to give different models traceable forcing, matched untreated controls and declared diagnostics. Disagreement is valuable information. Related model versions would not count as fully independent evidence, and an ensemble average would not establish real-world safety.

• Freeze primary hypotheses, outcome measures and exclusion rules before confirmatory runs.
• Distinguish exploratory studies from pre-specified tests.
• Keep null, adverse and contradictory results in the public record.
• Publish enough methods and data for independent reproduction.

Every larger exposure needs a new justification

Later stages in the proposed pathway would require their own evidence, environmental review, technical controls and legitimate authority. Success in a laboratory or simulation would not create permission for open-atmosphere work. A safe shutdown mechanism would also need to be distinguished from demonstrated environmental recovery.

Chapters 22, 23 and 26 ask how ASES could fail, which numerical assumptions are fragile and how independent critics could expose weaknesses. The correct end point may be a narrower hypothesis, a different design or no further intervention research. A result that closes an unsafe or unworkable pathway is still useful knowledge.