Rather than optimization, they emphasize “hemibel thinking,” which says to look for operational performance that is below the theoretically possible by at least a factor of three, because then big improvements can be achieved with or without optimality.





