Before we discuss queues, we must introduce a little vocabulary. When we describe queueing systems, we refer to the queue, which is the waiting work, and the server, which is the resource performing the work. The pattern with which work arrives, which is usually unpredictable, is known as the arrival process. The time it takes the server to accomplish the work may also be unpredictable. This is known as the service process. We handle waiting work in a certain sequence, such as first-in-first-out (FIFO). This sequence is known as the queueing discipline.
product development has asymmetric economic payoff-functions. Not all deviations have negative economic consequences. Because product developers encounter unexpected opportunities, conformance to plan may not represent the best economic choice. Our control system must enable us to exploit unexpected opportunities by increasing the deviation from the original plan whenever this creates economic value.
We want to control measures of performance that have strong influence on economic success. We can identify these parameters using the economic framework discussed in Chapter 3. If a 10 percent increase in project expenses reduces profits by 1 percent, and a 10 percent increase in unit cost reduces profits by 50 percent, we should focus our attention on controlling unit cost, not project expenses.
Adaptive control systems designed to track dynamic goals are different than those designed to conform to static goals. Because product development has inherently high variability, it is critical to recognize situations where our goals should be dynamic. In such cases, we should strive to constantly reduce the gap between our current state and the economically optimal state, even when this economically optimal state is constantly changing. For example, this is commonly done in consumer marketing, where continuous preference-testing is used to detect market shifts that take place during development.
In a Markov process, the elapsed time between arrivals is exponentially distributed. This simply means that short inter-arrival times are more probable than long ones. This is also known as a memoryless arrival pattern, because each arrival is statistically independent from the next.
we should remember that our control system can add economic value in two ways. It can help us to reduce negative deviations from the plan, and it can attract effort to emergent opportunities, thereby increasing positive deviations from plan.
let me point out one more subtle implication of this approach towards buying information. It implies that there is an economically optimum sequence for risk-reduction activities. Low-cost activities that remove a lot of risk should occur before high-cost activities that remove very little risk.
Few developers realize that queues are the single most important cause of poor product development performance. Queues cause our development process to have too much design-in-process inventory (DIP). Developers are unaware of DIP, they do not measure it, and they do not manage it. They do not even realize that DIP is a problem.
Any subprocess within product development can be viewed in economic terms. The total cost of the subprocess is composed of its cost of capacity and the delay cost associated with its cycle time.
