The point of the article is that the relationship between "amount of exercise you do" and "calories you burn" is probably not linear, as has traditionally been believed.
According to the scientist's theory, you get diminishing returns, where you have to do ever more and more exercise to burn the same number of calories.
But the study they cite to support that view talks about very low levels of exercise (1100 calories a week) in the long term studies. If the diet calorie surplus was comparable in both long and short term, it's not surprising that the results were disproportionately low in the long term compared to the short-term studies: with such a low level of exercise, it wouldn't be surprising if the measurement error was significant. In a longer term study, it's tough to control for what people do the rest of the time.
And in their discussion of the Hadza tribe, they admit that the Hadza could be resting enough at other times of the day to offset the increased calories they expend for food gathering. All that does is tell us that given two populations that exercise about the same amount, the one that eats less will be leaner. That says nothing about whether the Hadza could remain lean on a Western diet if they also exercised more, which is the real question. Nowhere in the article is that question actually directly addressed. The study of marathoners they cite points out that the subjects increased their intake while they trained, which unsurprisingly limited their weight loss.
If you eat the same amount but exercise significantly more, you will lose weight. If you eat significantly less with no increase in exercise, you will lose weight. Nothing in this article supports with data the conclusion that the latter approach is better than the former.
According to the scientist's theory, you get diminishing returns, where you have to do ever more and more exercise to burn the same number of calories.