We can’t know if climate change models are correct - it’s way too much code, solid bug-checking of it would cost many billions, and without it, the models are m

We can’t know if climate change models are correct - it’s way too much code, solid bug-checking of it would cost many billions, and without it, the models are more scientism than science. The core problem here is that it’s hard to test if you don’t know what is the expected outcome (we don’t know how much the temperature is supposed to increase - we expect the model to tell us that…). In such cases, you don’t do black-box testing (i.e. just writing a bunch of test-cases), but rather have to resort to doing a painstaking and rigorous analysis of the code itself. We’ve done it in the past (e.g. on early space flight code), but the scope of the codebase was way smaller. Also, I suspect that, a lot of the time, people work on the model (tweak parameters, find and fix numerical problems etc) until it starts giving responses which match their apriori expectations. That’s pure scientism (bs masquerading as science).