You're not differentiating between an iterative design process, during which failures are expected, and a monolithic one where the end product is final. When Starships blow up, Musk can build new ones and try again. When the probe crashed, that was the end of the mission and there was no more Mars Climate Orbiter. It was a one-off. This is a common problem in spaceflight.
The Webb Space Telescope is another example. If it eventually does launch and something goes wrong, we'll have paid >$10 billion for absolutely nothing. These aren't trial and error because the designs are never improved based on the errors. They're just scrapped. SpaceX works in a fundamentally different way.
They are not that different from other space startups.
> "We succeeded in launching the rocket," Zhang told the media. "The experience we gained from evaluating the rocket's flight conditions will help us remodel the rocket as well as advance new rocket research and development."
When's the last time that someone tested a series of satellites (or launch vehicles), with a number of up-front failures? Prior to SpaceX, of course. To me it looks like that sort of iterative trial-and-error design process died with the end of the space race.
For instance, looking at LandSpace - that quote relates to the launch of Zhuque-1, of which there has never been a repeat despite a maiden flight in 2018, and Wikipedia says that a second flight ever happening is doubtful. The company's second launch (and probably any after that, too) will likely be with a completely different, liquid-fuelled design.
THAT's what failure normally looks like in aerospace. Whoops, we fucked up, and so now that design never flies again, or the company goes bankrupt, or everyone panics and goes through a massive design review because everything was supposed to be perfect and now obviously it wasn't.
The Webb Space Telescope is another example. If it eventually does launch and something goes wrong, we'll have paid >$10 billion for absolutely nothing. These aren't trial and error because the designs are never improved based on the errors. They're just scrapped. SpaceX works in a fundamentally different way.