2021-07-02
This rather begs the question of whether "models" are science, could be science, or can never be science.
I would suggest that to qualify as science, they should have to be incorporated into the scientific method.
The basic scientific method involves several stages, more or less as follows:
- Consider a problem
- Collect facts and data about the problem
- Formulate a hypothesis
- Make predictions using the hypothesis
- Test the predictions
- Repeat as required until it's all good
- Promote the hypothesis to a theory
- Publish everything to enable others to independently review, replicate and validate
No models in sight so far . . .
But wait! The problem usually involves a system of some kind (for example, a gravitational system) and it is often possible to create a model that implements a (possibly simplified) version of the hypothesis about how that system works, and can them imitate the system (up to a point), and can make predictions about how it would behave in specific circumstances.
So a model can be useful as a way to express the hypothesis in the form of a set of equations, and/or rules of operation, which may then be implemented as a computer program.
So we could add point 3A to the above list:
3A. Create and test a computer model that implements the hypothesis to simulate the system under investigation:
The computer model should then be tested by the person that defined it to make sure that the computer is faithfully behaving according to the specification of the model (the equations, rules, etc which are described by the hypothesis).
After this the computer code may be released for testing by the person who defined the hypothesis, who should test its performance against both the hypothesis and the real system.
If the model is required to make predictions then it should be used to make predictions about the system that can be validated against the performance of the actual system (for example, the gravitational solar system) over time.
Then it can be published (including all the computer code) for peer review and validation by independent peers as part of the total package.
Only once all these steps have been undertaken can the model really be acknowledged as a scientific representation of the workings of the system under investigation. Until then it is simply a computer encapsulation of a hypothesis which remains unproven (even if it was created by a 'scientist'!).
So bearing all that in mind:
- the AIER offers us a critique of climate modelling under the title Climate Models: Worse Than Nothing?
- this site previously offered a critique of the Imperial College model used by Professor Ferguson to advise the UK and American governments in the early days of the Covid crisis.
Please review these offerings against the criteria offered above in steps 1 through 3A through 8, and form your own conclusions.
You may wish to keep a soothing cup of tea to hand . . .
