(5) Truer insights from false models:
The point of a scientific model is not to be true, but to be false in interesting ways, leading to new questions and refinements - models that are wrong in different and more productive ways.
A perfect model would be pointless. It would be an evolutionary dead end in the history of science.
But models can be false in uninteresting ways: they can be plain wrong. Wimsatt lists the varieties of being wrong that are actually productive.