If I read that article correctly it seemed to be pointing out that sometimes calories are literally mislabeled. That is - a sandwich from a vendor may be labeled as 400 kcal but actually it turns out it was closer to 500 kcal. Casey Neistat did a special for the NYT a couple of years ago about this very problem [1]. The ArsTechnica article you linked did go into other things - but there's definitely a difficulty in dealing with things where labels say one thing and can be 10-15% wrong pretty easily.

[1]: http://www.nytimes.com/2013/02/13/opinion/calorie-detective....