One thing I've learned from reading /. for years is that one must always read the posted article. Unfortunately I don't live on the computer, so by the time this topic came to me I am no longer able to RTFA. I will read the original text if it is posted someplace.
(By the way, congrats on the kid!)

We use data models for all sorts of shit, for example, 'proving' that the design of the aircraft you're flying in won't crash
Please be clear on this: a computerised aircraft model doesn't prove a thing - it just gives designers a good idea. Building and flying the aircraft without crashing actually proves that it won't crash under the conditions the aircraft was flown. Models are just convenient reconstructions of what we have observed, and only include what someone has actually seen and measured. Anything that wasn't programmed into the model isn't included in the model.

We still haven't figured out all the variables that effect our atmosphere, so none of the models are complete.