All the js asset sizes quoted in the article seem inflated compared to the figures I see when I try to verify the authors figures.