There’s a point at which you learn more from actually building something and putting it through its paces than simulating. It’s a tough balance to strike , no argument there. Simulating until you’ve covered every conceivable edge case and failure mode is ludicrously costly and time consuming. Relying entirely on yeeting shit and seeing how it fails risks missing the edge cases. But so far, I’ve seen little reason to doubt that SpaceX has found a working balance between simulation and practical testing. They’re certainly progressing faster than the industry historically has and the F9 has had no failures, even partial ones, in over 200 flights. That’s a track record that most launch vehicles can’t meet. It’s definitely possible there’s a 1/1000 flaw in the Falcon 9, but until it actually happens and they lose a rocket and/or a payload (gods willing it won’t be crew), it’s nothing but a hypothetical “but what if…” scenario.