I think year to year correlation would obviously be one thing to look at, but a low correlation doesn't *necessarily* mean that ERA is less useful. It could mean that pitching performance is subject to fluctuations. If you take whatever pitching statistic you like to use, WAR, ERA+, WHIP, it would be interesting to see how well they correlate with ERA, and how well they correlate year to year (some of that may be in the article that I can't see).
Regardless, maybe you were just using hyperbole, but it would be silly to say ERA is useless. If you take the top 10 pitchers in ERA any given year, I guarantee you they will in aggregate have had better years than the bottom 10 pitchers.