The author mentions those at the bottom of the article, but two problems highlighted still remain:

* There's another intermediary concept (kernel density estimation) between the audience and the data

* They're still likely to misrepresent tight groupings and discontinuities, which will be smoothed out