I think the key here is the word "fact". Scientific models are models. They are good models if they correspond with measurements that we make in the real worl

I think the key here is the word "fact". Scientific models are models. They are good models if they correspond with measurements that we make in the real world. The are even better models if they make a prediction that we can test in the real world. However, a model is never a "fact". "The FSM did it" is always going to be a possibility, even if it is an incredibly remote one. We use scientific models because they are useful not because they are fact. Even if our model does not always match our observations it can still be a useful model in some circumstances. It's very important when talking about science that we understand that it isn't fact. It is a model. It might be a good model. It might be a bad model. It might be a useful model for our application. It might be a poor model for our application. That's it. When you see a headline that says "X is Y" and it's about a scientific model, what that means is "If our model is consistent with the real world then the model predicts that we should find that X appears to be Y in the real world". But that's a bit too long to put in a headline and you would have to do it for every scientific conversation. It is extremely unfortunate that on the first day of junior high school when you start doing "proper" science classes that they don't sit people down and explain what a scientific model is. Mostly I think it's because the teachers don't know, because their teachers never taught them.