Re:Climategate Inquiry Clearly Ignores Wrongdoing
by yummyfajitas on Thursday July 08, 2010 @01:36PM
(#1498082)
Yes, this should reduce our confidence in the fact that the "scientific consensus" is sound. Many arguments are made that "the science is settled".
Such arguments are sound, the logic behind them is the following: intuitively, we start out believing that if there were evidence against our hypothesis (in this case, unprecedented warming), with a high probability the scientific process would have discovered/published it. Since evidence against this hypothesis has not been published, we conclude with high probability it does not exist. The lack of evidence against it increases our confidence that the hypothesis is true.
Climategate has reduced our estimate of the probability that the scientific process would have discovered/published evidence against the unprecedented warming hypothesis if it existed. Therefore, such evidence may exist, but may not have been discovered due to bad behavior by Mann/Jones/etc.
I'll geek out a little. Let H be a hypothesis and let E be evidence against it. I.e., P(H | !E ) < P(H), and conversely P(H | !E) > P(H). Now let O be the fact of E being observed/published/becoming known to us. We can compute:
Before climategate, we believed that P(!O | E) is close to zero, so therefore P(H | !O) is approximately P(H | !E) > P(H). After climategate, we must increase our estimate of P(!O | E), which in turn reduces our estimate of P(H | !O).
Such arguments are sound, the logic behind them is the following: intuitively, we start out believing that if there were evidence against our hypothesis (in this case, unprecedented warming), with a high probability the scientific process would have discovered/published it. Since evidence against this hypothesis has not been published, we conclude with high probability it does not exist. The lack of evidence against it increases our confidence that the hypothesis is true.
Climategate has reduced our estimate of the probability that the scientific process would have discovered/published evidence against the unprecedented warming hypothesis if it existed. Therefore, such evidence may exist, but may not have been discovered due to bad behavior by Mann/Jones/etc.
I'll geek out a little. Let H be a hypothesis and let E be evidence against it. I.e., P(H | !E ) < P(H), and conversely P(H | !E) > P(H). Now let O be the fact of E being observed/published/becoming known to us. We can compute:
P(H | !O) = P(H | E) P(E | !O) + P(H | !E) P(!E | !O)
Bayes rule turns this into:
P(H | !O) = [P(H | E) P(!O | E) P(E) + P(H | !E)P(!O | !E) P(!E) ] / P(!O).
Before climategate, we believed that P(!O | E) is close to zero, so therefore P(H | !O) is approximately P(H | !E) > P(H). After climategate, we must increase our estimate of P(!O | E), which in turn reduces our estimate of P(H | !O).