Some perspectives on uncertainty-based derivative-free optimization