User testing has clear limits. The context of a user test is usually something like, "Assume that you have bought one of these, and assume that you're trying to do X. See if you can do X with this." If you're assuming those things, you can't test them.

This is in contrast with what startups do, which is just launching something to see if the market gives a shit. Apple is a good example of that, too. Look at their first product:

https://en.wikipedia.org/wiki/Apple_I

This computer didn't have a keyboard, a monitor, a case, or a power supply. But it was a clear minimum viable product: people bought it and used it, while making it clear they wanted more. A year later, they launched the Apple II:

https://en.wikipedia.org/wiki/Apple_II

These days Apple would never launch something so rough and experimental as the Apple I. Instead, as with MP3 players, smartphones, and smart watches, they were content to let other people pioneer the space and then swoop in once the market was proven.

Magic Leap is trying to be as slick as 2010s Apple, but that cuts them off from acting like 1970s Apple. That could work if they're correct on all their key design and market hypotheses. But it's a giant gamble, and unlike 2010s Apple, a failure could kill them.

This used to be the case and might still be compared to most other products, but that value keeps diminishing more and more.

I would certainly welcome an increased focus on "it just works" again as opposed to "it looks flashy". I enjoy using Apple products a lot for various reasons but it became hard to overlook some issues, especially when it comes to software.