Brad Sallows
Army.ca Legend
- Reaction score
- 8,847
- Points
- 1,040
Science is all about doing repeatable experiments in an attempt to prove that a theory is wrong.
Not entirely, or even mostly. Experiments look for things theories predict. If a prediction is wrong/contradicted, the theory is wrong or incomplete (requires modification).
For example, suppose a climate model (which is not a theory, but its internal workings are at least based on some bits of theories) predicts a future value of some defined measure (eg. a temperature). The only practical experiment is a natural one - to wait and measure (and it can't be repeated, except as a set of natural experiments - start with many predicted values for different locations and different times and later measure them all). If a measure doesn't match a prediction, the model is wrong, and has to be modified.
Predictions will often have ranges of uncertainty, and computer model outputs will with very, very few exceptions have ranges of uncertainty, so failure to match exactly is not disproof. The go-to excuse when a measure fails to confirm a prediction closely is that the measure falls within the range of uncertainty of the prediction (which is legitimate, if it does).
[Snark begins here.]
Climate models have a remarkably consistent history of predicting temperatures higher than what are subsequently measured, with large uncertainties. The modellers and theorists seem to be having difficulty adjusting theories and models so that measured values tend to fall on either side of predicted values, with narrower ranges of uncertainty. The models aren't comprehensive, and the masses of data that people should be gathering to put into them, no-one seems very interested in gathering to the point of doing all that hard work themselves. Instead, they excitedly jump into the policy arena, and here we are.