I will give an example. Did we need years and years of funding for a lab to work on an obscure programming language for multiprocessing that basically only that lab ever used? Probably not. How much of funded science is just useless waste for a group of people to play with things like this? A lot, I would speculate. There isn't really a good way to spot what's useful and what isn't but let's not pretend academia is a purely selfless institution.