That illustrates why geo-engineering would first require “atmospheric testing”—an impossibility, because we would have to ruin the Earth experimentally 10 times before we could hope to figure out how to make geo-engineering produce just the desired good effects on the 11th try.
