It's really amazing to read and see all sorts of good points from both sides, and know that if it weren't for the benefit of hindsight, I'm not really sure which side I would agree with.
But I really strongly get the feeling that Andy had a great opportunity thrown into his lap, but by looking at MINIX as a "non-serious" hobby (as Linus implied) that he only did for fun and to unwind, and by not really listening to his user base (make it free, fix filesystem race conditions, add POSIX compatibility, etc), he really threw the opportunity away for it to grow. Which is fine if growth was never the goal.
And sure, a project owner doesn't have to say yes to every feature, and they actually probably shouldn't. But you have to let what you're creating evolve into something that meets needs, and turn into something different. Look at the first page of commits on Semantic-UI's repo. It was called Origami (then Shape) and was meant for doing 3D stuff in JS, before he renamed it to Semantic and turned it into a JS-based version of Bootstrap. Terrible example and doesn't really apply here but I remembered coming across that fascinating bunch of commits the other day and wanted to share it with someone.