Thomas comments on 2017 LessWrong Survey