Good point, the author clearly misses the point that tests themselves are code and thus can have errors in them as well. Code coverage is just a vanity metric if you're tests don't actually test the correct thing.