Starting simple versus starting complex is different than making the same kind of thing repeatedly to refine the act of making it and the end quality.

Iterating on a project is definitely a great way to learn about it, refine how you make it and accrete complexity. So it can but doesn't necessarily have to encompass both ideas.

I also think that whilst it's tempting to stretch the anecdote to fit all sorts of ideas it's particularly unhelpful when trying to discuss the idea it actually references.

But if you threw away all the code after each version, I doubt you'd get much done. Sure the code might be perfect, but you can do only so much in a short timeframe.

Perhaps it would be best to iterate fast early on in the project, throw away a couple of weeks with PoCs and then go for the big one.