I think what he was saying is that past successes are not necessarily indicative of future successes. I do think this simple observation can be used to derive understanding.

I'd like to increase the volume of my balloon as much as possible. I blow it up a little bit. It increases a little bit. I blow into it hard. It substantially increases. I keep blowing and it invariably keeps getting larger and larger. There is an absolutely perfect and undeniable correlation between how much I blow into that balloon and the measured volume. In fact it's not even a correlation, it is a clear causal relationship. This leads to an empirical conclusion that is completely wrong.

In our society those that say to keep blowing are going to end up succeeding. Not only do their predictions work out, and quite profitably so, but they're also optimistic and promise untold volumes to come. The understanding would be to consider that in fields where meaningful tests cannot be readily performed, empiricism should never replace logic. And illogical but empirically supported concepts should be treated with extreme skepticism. Logic can be flawed and proven to be wrong. Empiricism can be equally flawed, yet the danger is that it can also be impossible to prove wrong until you're left picking up pieces of rubber.