I
read a very interesting article on Albert Einstein recently, and specifically about his intellectual journey from being a strict inductive Positivist, to a sort of Platonic idealist about scientific theory.
There were several fascinating takeaways from the post. The first was the observation that it is impossible to arrive at theory directly from data -- the universe is underdetermined by pure scientific data, as evidenced by the fact that Newtonian physics and general relativity both agree with "the data" the vast majority of the time, despite their mathematics being derived in completely different ways. It would seem that many theories can be true at the same time, and thus you cannot use simple observation to arrive at "the one true theory".
But a little closing remark also caught my eye. Einstein is known for saying that if scientific experiments has disagreed with the predictions of general relativity, then the experiments must have been done wrong, because the theory is correct, the same way that if a scientific experiment suggests that 2+2=5, you know something has gone wrong without even having to look at it. The author takes this a step further and points out that a great many excellent theories would have been lost if their authors had abandoned them at the first sign of disagreement with experiment. We need thinkers with the confidence that their theories are correct, or at least close to correct, and are willing to ignore countervailing evidence for at least a little while as they work out the kinks, or the experiments improve.
It's underappreciated how much of advancement in science comes from intuition and creative thinking in the abstract realm -- the connection of reason and understanding to the "the real world" is a deep and mysterious one, and no matter what the positivists say, the data cannot tell you how to interpret itself.