"The exception proves the rule." In this usage, prove means to test. If the model can't explain such anomalous behavior as an increase in CO2 and a decrease in temperature, then the model is fatally flawed.

It's as simple as this: if the model can't accurately reflect what actually did happen in the past, then it's useless at predicting what might happen in the future. There is no simplier way to explain it. You're the one in denial here, not me.

Validation is essential for modeling. One of the projects that's going on in the space surveillance community is an improved atmospheric model to help increase the accuracy predictions for reentering satellites. This work is pain-staking, tedious, and important. Today, it's hard to narrow down a reentry period to less than a 6 hour window when you're more than a day out. The developers are taking steps to validate the new model against actual reentry events through different solar cycles. The sun plays a big role in the speed of orbital decay.

This is how modeling is done properly and scientifically. They're not just creating a model and expecting anyone to believe the results. The model must prove more accurate than the previous atmospheric model or it won't be used. But then, perhaps the space surveillance community isn't as gullible as the global warming true believers who seem bent on accepting the results of any model that agrees with their views regardless of the model's maturity.