Estimating the odds on a high climate sensitivity

Climate sensitivity, or the final equilibrium warming resulting from a doubling of carbon dioxide levels, cannot be measured directly, since the real climate system will never be subjected to a carbon dioxide doubling and then allowed to come into equilibrium. Because we can neither observe sensitivity directly nor find observable quantities that are directly proportional to it over the full ranges of values that are consistent with current observations, any estimate of the probability that a given greenhouse gas stabilisation level might result in a "dangerous" equilibrium warming turns out to be dependent on subjective prior assumptions of the investigators and not purely on constraints provided by actual climate observations.
In contrast, we can observe the strength of atmospheric feedbacks, or the change in top-of-atmosphere energy flux in response to a surface temperature change, much more directly than climate sensitivity itself. The net strength of these feedbacks is directly related to the inverse of the climate sensitivity, or the range of stabilisation concentrations consistent with a target temperature rise. Hence policies that focus on a maximum temperature rise, accepting uncertainty in the stabilisation concentration that may be required to achieve it, are better informed by climate observations than policies that focus on a target stabilisation concentration, accepting uncertainty in the resulting long-term equilibrium warming.