Working with multiple problems at once — LessWrong