It's not that pain points are impossible to predict. It's that you'll get some predictions wrong. False positives will cost you from the start and false negatives will bring you problems when they surprise you.
Instead of planning for future predicted problems, plan for change. Embrace the fact you know little about the future and that both the problems and the way you tackle them will always be mutating.
What you're saying is basically "the only constant is change", which is fine and dandy, but also kind of a useless truism. You are attempting to predict the future by building a product/project. You are assuming that the thing you build will continue to be useful and valuable into the future. By the very act of building, you are asserting some knowledge about the future.
Sure, you shouldn't go crazy and build a complex and expensive solution you don't need today and may never need, but it's disingenuous to claim that a lack of perfect clarity into the future justifies a complete lack of planning for the future.
Instead of planning for future predicted problems, plan for change. Embrace the fact you know little about the future and that both the problems and the way you tackle them will always be mutating.