I think the biggest mistake that Magic Leap made is that they compromised on their hardware. Like you said, they tried to go general purpose which means low "enough" cost for devs to pick one up and build an app, for enthusiasts to try em out etc. But the thing is, they didn't have best in class hardware. I've tried one and its marginally better than a hololens 2. They could have gone all out with a vision prototype that wowed the world with its capabilities, instead they increased the field of view, slightly and made the form-factor slightly better.