Pedro Domingos

Note that if the uncertainty about climate effects of CO2 increases (for any given best average estimate) then the expected economic costs are *higher*.

It's the average estimate that's too high. And uncertainty does *not* necessarily increase economic costs, which could even be negative (e.g., higher agricultural productivity, lower deaths from cold, etc.). (BTW, Gaussians are symmetric - you know that.)

Based on my non-expert reading of the data and research, I disagree. So do most climate scientists. Symmetric uncertainty increases expected costs when costs are convex. (Jensen's Inequality) as with climate change (4 degrees hotter is more than twice as damaging as 2 degrees)

And it decreases expected costs when they're concave, which by basic climate science is easily the most likely. (The greenhouse effect of CO2 is logarithmic, so adding more to the atmosphere produces decreasing warming - are you really disputing that?)

The uncertainty I’m referring to is about how much warming we can expect. There are scenarios where the warming or other bad outcomes are amplified and others where they dampened. For instance, even a small tail risk of a runaway greenhouse effect would be catastrophic.

There are tail risks in both directions, and there are tail risks if we attempt to decarbonize the economy. A slightly increased chance of a runaway greenhouse effect is not enough to make the expected cost of CO2 convex instead of concave.

What are the ecological tail risks in trying to decarbonize?