Gravar-mail: Automated grading of research performance clearly fails to measure up