Monday, 9 January 2012

Aristotle on climate models

Aristotle - from Wikipedia

Science, then, may be defined as a habit or formed faculty of demonstration, with all the further qualifications which are enumerated in the Analytics. It is necessary to add this, because it is only when the principles of our knowledge are accepted and known to us in a particular way, that we can properly be said to have scientific knowledge; for unless these principles are better known to us than the conclusions based upon them, our knowledge will be merely accidental.
Aristotle – The Nicomachean Ethics

Okay, Aristotle didn’t quite get round to climate models, but it’s worth going back to basics when we take a look at the claims made about them. Why? Because they have wasted colossal sums of money and still threaten to blight the lives of our children and grandchildren. As Aristotle seems to have grasped two and a half millennia ago, science is all about demonstration, but climate models have yet to demonstrate anything of value.

So let us take Aristotle’s principle of demonstration a little further and apply it to climate models. It’s a good principle even though now call our demonstrations experiments. In that case, there are three key elements to consider if computer models are to be used for predicting important climate parameters such as temperature.
  1. The output of the computer model  – the prediction.
  2. The state of the climate                   – the standard.
  3. The correlation between them         – the demonstration.
In other words, a climate model that purports to predict global temperatures over the next thirty years must demonstrate significant correlation with the actual climate for thirty years. Or at the very least, it must demonstrate a significant degree of correlation with the real climate over significant period of time.

There is no other test – no other demonstration.

The only area for latitude is the degree of correlation we might count as a successful demonstration. Successfully predicting a number of climate cycles would be good. We may be talking centuries rather than decades for that, but a high degree of correlation could possibly shorten it to a few decades.

Unfortunately, the models have so far failed to predict global temperatures and the current warming hiatus of the past decade. In fact nobody is making a systematic and serious effort to demonstrate their long-term predictive skill. We are supposed to take them on trust, but that just won’t do will it? The basic scientific principle of demonstration is being violated.

When basic principles are involved, such as the principle of demonstration, then I think it pays to be blunt about what these climate modelling guys are doing – the value of their activities for the rest of us who play the game with a somewhat straighter bat. These are my conclusions.
  1. Climate models are useless and every penny spent on them has been wasted.
  2. It is for the modellers to prove otherwise – by demonstration.
  3. But not with our money.


Dembones said...

That is why neither psychology or psychiatry should be considered as scientific. Interesting, yes; worthy of discussion, yes; scientific, no. Human behaviour cannot be tested in a truly demonstrable way without a proper standard. As each person is individual and unique, there is no standard.

A K Haart said...

Dembones - yes - psychology has a problem with demonstration because any tested subject usually has to be unaware of the true purpose of the test so often can't be tested twice. So no demonstration.

Yet as you say, worthwhile things can be said.