Unlocking Solutions—by understanding coordination problems — LessWrong