According to the article, this was measured through King-Devick tests. What's the point of using intentionally paywalled tests, other than harming reproducibility?