It is not how long it is, but how you make it long – Waiting lines in a multi-step service process, (2001) System Dynamics Review 17(3), 333-340 ________ , Follow the laggard? – not all bottlenecks are created equal

Genesis. I have always been interested in waiting lines because waiting in line for me is like hell on earth. I take great pains to avoid waiting lines. My academic interest in queues, however, started when I tried to simulate waiting lines in a closed system such as a ski resort. Little did I know that I was trespassing on hostile territories. It turned out that queuing theory is a very well developed field with voluminous literature. It is hopeless to go through all the literature just to find out what has not been covered yet. All was not lost. A casual reading of the voluminous literature on queuing theory taught me that a queue must be present even in a steady-state service process. I then inferred that the greater the flow capacity (number of customers served per unit time), the longer must be the steady-state queue if the service process is not to be interrupted. And the minimum steady-state queue should not be regarded as a waiting line at all. Hence the title of this paper. I also noticed that Naor’s seminal paper that started the whole literature on regulating queue length with prices was based on a single-step service process. And this assumption was followed by almost all later contributors. Since I just finished a simulation on Goldratt’s scout hike, I knew well that the most binding bottleneck in a multiple-link process does not alway