That was a good read. However, it seems like the section on geo engineering does exactly what the rest tries to warn against: predicting behavior using simplified first order models.
I'm biased towards believing the conclusion the author draws, but the reasoning here is suspect.