Evaluating the Evaluators: Problems with Outside "Neutrals"