Bodybuilding hard seems to conflate 2 types of hard tasks: move-a-mountain-with-a-shovel hard, where something is time consuming, such as having to hand-label m

Bodybuilding hard seems to conflate 2 types of hard tasks: move-a-mountain-with-a-shovel hard, where something is time consuming, such as having to hand-label millions of images for machine learning datasets, and quit-heroin hard, where something is against human desires, such as convincing people to abandon their current way of doing something for something else that is a short-term loss but a long-term gain, for example switching to a more performant but incompatible CPU architecture. Math hard, as the article says, is not really defensible, since copying is easy. Move-a-mountain hard is more defensible along the time/money axes, competitors need to throw time/money at the problem to catch up. Quit-heroin hard is the most defensible, since customers will loath to pay the switching price regularly. Of course, it's not as black-and-white as this. All products have a mix of the three, and its where the product lands in the mix that determines the promotion strategies that the company should take.