The author's point is not that schema changes are hard - but rather that versioning schema changes are hard.