Multi-winner Voting: a question of Alignment — LessWrong