I came across the blog Scottish Sceptic at the weekend. At the site, the owner has been compiling a non-polemical summary of the mainstream sceptic view of the science. Unlike here, the statement studiously avoids discussion of policy or politics. I made the following comment in the hope of furthering discussion.
I have had a look through the above, and it appears a fair summary the sceptic position of the science. In general it shows how magnitude and likelihood go in opposite directions. The best corroborated science has trivial implications. The most alarming predictions are basically of the form “If A then maybe B. If B then possibly C. If C happens in a certain way then it could be D. D is an extremely alarming situation” This then gives the headline like
“Leading scientists are concerned we are heading for D“.
Having read quite widely on sceptic ideas, on the subject of climate models, sceptics view them as “black boxes“. This would not be concerning if they followed the normal scientific procedure of rigorously evaluating the predictions with the actual data, and adjusting accordingly. Instead, it appears to be past data that gets adjusted to the models, along with some very fuzzy analysis.
Another point is that sceptics tend to see a scientific approach as questioning, identifying anomalies, and getting ever more precise answers. Mainstream climate science is nearer to a definition of “science is what scientists do”.
That leads to another point. Sceptics tend to demand higher levels of evidence. The mainstream seems to accept levels of evidence that a criminal court of law would reject. “Scientists believe/agree”, or “Climate Models predict” or comments a court would reject such views as either hearsay or unsubstantiated. So in the wider world sceptics are not the ones with the marginal position.
ManicBeancounter 
Brian H
/ 09/05/2012The confidence levels accepted as gold standard in CS are pathetic, similar to the squishy social sciences: 95% is “high likelihood”, and it can barely dream of 98% or 99% levels. While real physics demands 5-sigma or higher levels (about 99.99%). Because data selection and confirmation bias and forms of error are so insidious.
manicbeancounter
/ 09/05/2012Brian, in climate, the science is always worse than you think.
I think it would be great if CS would accept 95% confidence levels – and use qualified statisticians to do the testing. They don’t. The 2007 AR4 Summary for Policymakers has 90% as “High Confidence” – and they fail to reject evidence that is just “more likely than not”. Even these significance tests are admissible only if you pasts stringent tests for bias. In Economics these were rarely carried out. In CS it is not even in the textbook.
Brian H
/ 15/05/2012Yabbut, don’t agree “it would be great if CS would accept 95% confidence levels”. Even in the hard sciences where reductionism and experimental replication (supposedly) rules, many more 9s are required — BECAUSE of the dangers of experimenter error and bias! How much more so in poorly constrained “observational” sciences like CS?
I agree that requiring 95% would sink most CS; but that’s because it doesn’t even rise to the level of slightly plausible speculation in real science terms.