Interesting article. I was unaware of such developments.

But what I am puzzled about, is that it seems to me that you are saying the error they report is only saying how much the models disagree with each other, whereas I am interested in how much the models disagree with the reality.