Per the popular article I read, he found that ~10% of the judges were consistent in a given competition, but it wasn't the same 10% across competitions. To deci

Per the popular article I read, he found that ~10% of the judges were consistent in a given competition, but it wasn't the same 10% across competitions. To decide what this implies I guess one would have to figure out whether 10% is higher or lower than you'd expect by chance. (That said, it sounds like he chose his definition of "consistent" after gathering the data, so it doesn't sound like it was a very rigorous process. One could say he found 10% of the judges were consistent, or one could say that he defined "consistent" to mean "within the top 10%".)