Thursday, July 19, 2012

Scientific Computer Modeling Method 101

By Roy Tucker

I published this in May of 2009,  but with all the information that has come out in the last three years I think this is worth another look.  RK

I'm a bit dismayed about how computer models have come to be more important than actual observations and so I offer a formal statement of the Scientific Computer Modeling Method.

The Scientific Method

1. Observe a phenomenon carefully.

2. Develop a hypothesis that possibly explains the phenomenon.

3. Perform a test in an attempt to disprove or invalidate the hypothesis. If the hypothesis is disproven, return to steps 1 and 2.

4. A hypothesis that stubbornly refuses to be invalidated may be correct. Continue testing.

The Scientific Computer Modeling Method

1. Observe a phenomenon carefully.

2. Develop a computer model that mimics the behavior of the phenomenon.

3. Select observations that conform to the model predictions and dismiss observations as of inadequate quality that conflict with the computer model.

4. In instances where all of the observations conflict with the model, "refine" the model with fudge factors to give a better match with pesky facts. Assert that these factors reveal fundamental processes previously unknown in association with the phenomenon. Under no circumstances willingly reveal your complete data sets, methods, or computer codes.

5. Upon achieving a model of incomprehensible complexity that still somewhat resembles the phenomenon, begin to issue to the popular media dire predictions of catastrophe that will occur as far in the future as possible, at least beyond your professional lifetime.

6. Continue to "refine" the model in order to maximize funding and the awarding of Nobel Prizes.

7. Dismiss as unqualified, ignorant, and conspiracy theorists all who offer criticisms of the model.
Repeat steps 3 through 7 indefinitely.

Reprinted with permission. RK

No comments: