He's not saying that he's not worried about FPE. What he's saying is that misunderstanding the physical system being modeled is far more likely to cause problem

He's not saying that he's not worried about FPE. What he's saying is that misunderstanding the physical system being modeled is far more likely to cause problems, and that it's much harder to spot than an FPE. So your mathematical model requires much more scrutiny for that reason. Basically if you don't spend much time questioning your assumptions about the world your system exists in, then don't even bother accounting for FPE because it's going to be overwhelmed by errors that you actually chose to introduce by your modeling assumptions.