Reply To: Questions regarding DPF.

[vansig]

Brian H wrote:
Retro-casting is useful for generating speculations and eventually hypotheses. Actual predictions require full disclosure of model variables and algorithms, plus input data set, which are then frozen (no touch, no fiddle) for the duration of the prediction. A reasonable fit fails to disprove the forecast model; a failure invalidates it completely (since the specific reason can only be guessed at — for which you need a brand new square one re-prediction and test.) Enough failures to disprove, despite best (honest) efforts, and you MAY begin to put some reliance on the model.

None of the Warmist scenario-games meet any of those criteria.

Stunt FOIA requests do not really expose a lack of full disclosure.

Let’s lock in Hansen’s predictions made in 1988. There were three scenarios, A, B, and C.
Observations are now out of agreement with A; but B and C are still good.
— http://www.logicalscience.com/skeptic_arguments/models-dont-work.html