The question isn't whether it's necessary, but whether it's profitable. Many programmers have an effect that's a % of existing metrics, for example a performance improvement that increases subscription renewals by 5%. As volume increases, it becomes profitable to hire people who have a smaller and smaller impact.