Barring the possibility of Apple ever laying down clear, comprehensive, stable and reliable guidelines for avoiding rejection (with many real, anonymized examples), it sounds like the MVP approach is the best way to go.

Rob's case is special, though, because his app was never actually rejected. I suppose if he'd been willing to submit a MVP and wait the same amount of time in limbo before putting any more effort into it, he could have saved himself all that hard work. But the aggravation of not getting an answer would have still sucked.