The authors appear to define "Open Access" as something along the lines of "any human with an internet connection can download a human-readable copy of the paper". They are talking about what Peter Suber called "gratis" OA: the removal of price barriers.[http://www.arl.org/sparc/...]

Usually, as in the BBB declarations and in national and institutional mandates, the definition of OA used is of "libre" OA: the removal of permissions barriers. As science is ever more dependent on "big data", computer processing, literature mining, and the synthesis of work dispersed over many journal articles, it is this definition of OA that is actually more interesting. This is not an obscure definition or distinction and I'm surprised the authors didn't include it in the article.

Hopefully next year's paper on the OA situation will give us the numbers regarding this important distinction, and on the spectrum of permission barriers (licenses) within OA.