(I am in agreement with the article) Both Knuth and Hoare would never think that carefully choosing the optimal or near-optimal architecture or algorithm is pr

(I am in agreement with the article) Both Knuth and Hoare would never think that carefully choosing the optimal or near-optimal architecture or algorithm is premature. Unless the number of things you are processing is small and it doesn't matter what you do, you may as well do the easiest thing. Or, if the optimal thing is super hairy, are you sure it will be worth it? They are just arguing for a bit of common sense: identify the most important performance issues, and make sure you work on those. There are all kinds of optimizations that people bring up that are way down the list in terms of impact. 'Nuff said!