This is literally the first one. There's only been a single Starship explosion in the upper atmosphere.

And no, that leads to spending decades of time going down paths and intricately designing and simulating every possible detail of a system, only to build them, have something unexpected happen, and then realize that the team never considered X effect in Y, Z, etc conditions, and then have to spend years redesigning everything. (Not to mention that at the end of all that we still had two Space Shuttles explode in the upper atmosphere, but with crews on board).

Design it, build it, test it, and get immediate feedback on it, and then redesign it. One way or another, it almost always has to go through that cycle, and it's a lot cheaper to do it upfront.