This is a false choice, and much of the discussion here is arguing for or against a strawman. Pick the right algorithm, and your efficiency battle is mostly ove

This is a false choice, and much of the discussion here is arguing for or against a strawman. Pick the right algorithm, and your efficiency battle is mostly over. Code it cleanly and have good tests for it. Then, when you discover where the actual performance problems are, you can fix them, transforming clean, correct code to faster, clean, correct code.