It's exactly apples and oranges being compared here. Languages and libraries or framework overhead is nothing compared to the toll of Big oh complexity. In fact the former doesn't even get multiplied that much as your input is pretty much constant. You start your application once unlike the input to your npow3 algorithm taking in a massive dataset. Also languages can make development more productive, which again is debatable but the slowness isn't comparable to Big O complexity. I am all in for leaner and efficient languages or stacks but I can live with bloated languages as opposed to inefficient algorithms.