Bayesian logic makes extensive use of inductive logic, and Bayesian techniques can help to determine allowable climate sensitivity values, as often the estimated climate sensitivity values are based on data that are sparse, noisy or biased, and/or all of these. Using Bayesian logic, estimates of climate sensitivities based on observations/measurements can be compared with previously projected values, and observing a difference, the projections can be "corrected" by arbitrarily subtracting off the discrepancy. When new data are collected, these too may disagree with the new predictions, and if so, another "correction" is applied, and so on.

Kelly and Tan, 2011, built a preliminary integrated assessment model to test how fat tails (for temperature change PDFs) due to uncertainties in the climate system affect economic growth, and examined whether, and how fast, uncertainties could be diminished through Bayesian learning. This research found that if the climate system is close to equilibrium then Bayesian learning will quickly provide evidence against fat tail uncertainties; while if the climate system is not close to equilibrium (implying that true climate sensitivity is high) then Bayesian learning occurs more slowly, and the implications of a fat tail on the demand loading distribution would have a significant increase in the risk of failure. Weitzman, 2009, argues that fat tails reflect the "deep structural uncertainty for the low-probability, high-impact catastrophes" and that: "… It is inherently difficult to learn from finite samples alone enough about the probabilities of extreme events to thin down the bad tail of the PDF because, by definition, we don't get many data point observations of such catastrophes". Therefore, I find it disturbing that recent climate sensitivity measurements have not lead to an elimination of the fat tail for the climate sensitivity PDF.

Kelly, D.L. and Tan, Z., (2011), "Leaning, Growth and Climate Feedbacks" 2011 Camp Resources XVIII, University of Miami, August 15, 2011.

Weitzman, M., (2009a), "On Modeling and Interpreting the Economics of Catastrophic Climate Change," Review of Economics and Statistics, 91, pp. 1-19.

Weitzman, M., (2009b), "Additive damages, fat-tailed climate dynamics, and uncertain discounting"; Economics - The Open-Access, Open-Assessment E-Journal, 3, pp. 1-29.

What does the Bayesian interpretation of probability tell us about reductionism? The key to the Bayesian interpretation is the notion that, if probabilities represent our states of knowledge, measurements update these states of knowledge. Thus knowledge is gained in an incremental manner which is the essence of reductionism. Thus probabilities, in a Bayesian context, are absolutely reductionist. Reductionism does not preclude the existence of what might be called emergent phenomena (such as a Black Swan climate change event), but it does imply the ability to understand those phenomena completely in terms of the processes from which they are composed (given a suitably sophisticated well calibrated Earth System Model). Thus with sufficient Bayesian learning possible Black Swan events (which may be predictable by the "Butcher" but cannot be forecast by the "Turkey") can be revealed to society at large (hopefully in time to do something about it).