Those mistakes chiefly being starting at the low end and not doing anything to make the platform particularly more interesting than what was already out there. Users don’t talk about budget devices and developers don’t talk about mediocre platforms, and so the positive dev-user feedback loop needed to kickstart a platform never materialized.

I think a third totally new platform could be established even today, but it’s going to take smart planning and flawless execution to make happen.

Mobile OS's have the worst case of the chicken and egg problem. Nobody will buy a device with your OS unless there's apps for it, and nobody will make apps for it unless there's users.

How do you solve this problem? Mozilla is too small to do it, and even Microsoft has seemingly given up.

Canonical tried pretty hard.

The reality is that you're going to fight with the two richest companies on Earth, on a field where commercial relationships with OEMs (phone manufacturers) matter more than anything else - and OEMs like to talk with big, reliable companies, not scrappy startups. Your competitors have a 15/20-year head-start, in a field where such head-starts are practically unsurmountable. Your OS will be judged on its first few releases, which means you better have it on a beefy flagship phone; except no decent OEM is going to bet their flagship on your unproven OS, so they're going to give you a second- or third-rate model instead, which will make your OS feel slow and glitchy. Oh, and you can't make any real money from the whole effort, because otherwise they will go Android.