Rob Lister said:
Perhaps. Perhaps he has me on ignore as well. He did not address my first response to his initial post. Why don't you? I count myself as a complete novice regarding this and most areas of science so your opinion/references will be well regarded. You appear to be much more informed than am I.
His original link is not new information, although he introduces it as such. It is one of countless papers that present scenarios based on projections and forecasts. In one regard it amounts to futurology - you know the type of thing that reckoned by the year 2000 noone would need to work because robots would do everything.
These scenarios do the same sort of thing based on climatic projections, rather than technology projections as in my example.
The projrections are usually derived from global climate models that produce forecatss for all manner of climatic variables, from temperature to rainfall etc. Then, academics of various disciplines draw up scenarios of what they think would happen under these various states of the world.
You can draw your own clonclusions about the likely accuracy or otherwise of scenarios, but either way they will be dependent upon the quality of the projections on which they are built. So from there we come to the projections and the global climate models (GCMs) which are employed to dervie them.
This opens up a long long discussion on the shortcomings of these models including among other things:
1. the very theoretical impossibility of constructing an accurate model of the global climate. The climate is a highly complex, jointly determined, non-linear system. Mathemiticians would call it a chaotic system. Even with perfect data and information it would be impossible to model.
2. the data on which these models are built have substantial errors. Some parts of these models use least squares techniques to estimate relationships between various variables based on the known theories for thermodynamics. The problem is that poor data leads to inconsistent (i.e. biased) parameters in the estimated relationships - these are the unkown numbers that determine the quantitative relationships.
3. There is large scale endogeneity in the climate system. That is, variables are determined jointly (e.g. temp and rainfall - each feeds off the other contemporaneously). Again, this leads to statistical problems when constructing these models. These problems can be corrected in the construction, helping to make th models track history well enough, but they cannot stop forecasts, outside the data set from being biased.
4.Underlying theory. In order to construct any large scale (or small scale) joinntly determined model, one needs a theory to hang it on (except if using something like a neural net). Climatologists point out that thier understanding of the worling of the global climate are very limited. For example, they are almost in the dark on the nature and significance of solar activity in the climate. Building a model on incomplete knowledge also leads to bias.
So. the point is that all the projections, forecasts, scenarios are prone to so much bias and large confidence limits that they should be used very very carefully when making extremly costly policy like Kyoto.
