The upside was they got lots of data to improve the next rocket, which they already did. It's ready to go again, just waiting for FAA approval.
Or they could have done that the first time with a flame trench and a water deluge system that prevented the energy of the rockets from damaging the rockets dooming the first stage and making concrete rain on the neighboring wildlife reserve.
Elon could have verified that hypothesis with a full static fire test AND be able to see the damage to both rockets and pad from an actual launch without destroying a rocket.
This is why SpaceX grew from the start with an immune system to prevent Elon from ruining too much stuff.
I believe he’s better observed in Twitter, which is a company that didn’t develop an immune system to deal with him.
It seems that while the explosion would have been avoided with a deluge or flame trench, it could have also likely just been mitigated with a thicker concrete pad or a vent system.
Iterative implementation works very well for many kinds of software development projects. When an iteration involves 200 tanker trucks carrying fuel and oxidizer and dozens of rocket engines and results in destroying launch infrastructure and dumping debris into wetlands and inhabited areas, one might conclude "iteration" has made the jump from sound engineering principle to management fad dogma.
Wouldn't that be called working from first principles? Proving it doesn't work can be quicker in the long run than over-designing something based on what you think won't work.