bait, work off a tip on a undertaker on Tuesday and wwas stunned to discover more about this.

Image SourceThe surface food they are beginningto abc develop the marking of their fishery.

Nothing beats freshly caught seafood for sale profitably!

Fishing for whiting on the beaches that are available for hire. ]]>

I realize that absence of denial is not necessarily conclusive, but would it not be extraordinary that Cook would not simply say, “sorry but that information does not exist”??

]]>]]>The fool will no more be called noble,

nor the scoundrel said to be honorable.

]]>a principle of scientific thought that corresponds to

a kind of utter honesty–. . . you should report everything that you think might make it invalid. . . Details that could throw doubt on your interpretation must be given. . .must also put down all the facts that disagree with it . . .give all of the information to help others to judge the value of your contribution

The Annan and Hargreaves paper (published online in 2009, but only in 2011 in the printed journal) compares the effects on ECS inference based on the likelihood function from Forster & Gregory (2006) of three priors for ECS: uniform, Webster (Expert) and Cauchy. All of these are informative priors, and all will bias ECS estimation to a greater or lesser extent. There is no justification for using any of these priors, IMO, and their influence means that none of the resulting ECS PDFs properly reflects the data and method used and the assumed error distributions.

The Gaussian error assumptions made in Forster & Gregory (2006) corresponded, as they pointed out in their paper, to an almost uniform-in-climate-feedback-parameter prior, which has the form 1/ECS^2 when expressed in terms of ECS. That prior – which has a shape quite unlike any of the three informative priors – was uninformative for their study, hence a uniform-in-ECS prior (on the basis of which their results were restated in AR4) was highly informative. I blogged about AR4’s distortion of the Forster & Gregory (2006) results here.

Although AR5 does not properly address the issue of informative vs noninformative priors, it does at least discuss it. And it presents the Forster & Gregory (2006) results on their original basis, corresponding to use of an uninformative prior, as well as the AR4 altered version. The median estimate comes down from 2.4 K to 1.5 or 1.6 K, and the 95% bound from 14.2 K (AR4 Table 9.3) to ~3.5 K.

The median, which is not given in the Annan and Hargreaves paper, is a better best estimate for skewed distributions than the mode, but is unfortunately more difficult to deduce from their graphs.

]]>Mind you, the Climate Scientists have got round the “critical audience” problem in a breathtakingly direct way.

]]>