Implementation R2 assembly machine operators had traditionally been free to change the speed, temperature, pressure, and other settings on their equipment. Under Greenlight, standard settings would be determined and operators would be required to shut down their equipment as outlined above, rather than "tweaking" it as they deemed appropriate. Maintenance procedures were also changed. In the past, each maintenance person had an almost unique set of solutions to problems with each piece of machinery. Project Greenlight required that maintenance personnel collectively determine and adhere to standardized "best practice" procedures in correcting all problems. Project Greenlight was rolled out through R2 over the first six months of 1985. One of the first problems encountered was determining the proper centerline and range of values for the process control measurements to be taken. The centerline and ranges were initially set using base-line data collected at the beginning of the project using sampling techniques that became the recommended procedures for operators. The centerline was set at the mean of the mean values and was the initial target, and the initial control limits were set at three standard deviations. Exhibit 3 shows base-line data for pod weight and finger height. Further, he hoped to do more extensive research on customers' needs and their relationship to various process parameters, in order to determine the specification target values and then the capability of each process. While the initial machine settings were being established, all operators were trained in basic statistics, theories and principles of process control, and construction and use of statistical process control charts. Reactions to Project Greenlight The initial reaction from production, quality control, and others at R2 was that the project team was "giving away the store." Machine operators believed that they themselves would act responsibly, but they did not trust their neighbors. Quality control personnel did not trust the operators: they knew the "games" that some operators played to avoid rejecting production, and thought that eliminating QC's policing role would increase the number of defectives passed on to customers. Also, most process engineering technicians felt they were better trained to do the sampling and process control tasks now being transferred to the operators. In addition to getting people to trust the operators, the team also had to change the mindset of everyone at R2. Like most large manufacturing facilities, R2 was grounded in the concept of volume manufacturing, trying to minimize costs by keeping machine utilization as high as possible. Furthermore, operators and maintenance personnel already were very concerned with quality. However, they understood quality in a different way than that embodied in Project Greenlight, which necessitated replacing the old concept with a new one, not simply getting the new quality concept adopted. For example, based on the old concept, operators believed that shutting down equipment when process variation was high, but before defectives were actually produced (as Greenlight rules required), would be more expensive than waiting until defects actually occurred. For operators to adopt the new quality concept, that belief had to be replaced by one that held that such shutting down would be cheaper, not more expensive, in the long run. For their part, maintenance personnel previously made adjustments to equipment that they individually believed would improve the quality of the product produced. Under the new concept, which required standardized procedures, maintenance personnel felt that their tasks were being depersonalized: what had once been an art was now being bureaucratized. Similarly, operators would need to believe that a more stable process could more effectively produce consistent quality than "tweaking" based on their own experience. The general dissatisfaction came to a head when the first few months of results from the auditors and operators were released (Exhibit 4). While the reported defective rate from the operators had dropped from just under 1% to 0.5%, the defective rate from the central process auditors had climbed from just over 1% to levels averaging 10%. Exhibit 5 lists the sample types of defects assembly operators and auditors identified. It was quickly noted that auditors and assembly operators were finding different types and proportions of defects. Part of Rolfs' task was to assess which of these lists was a more accurate representation of the true defects in the process. To do this, he felt he needed to tie the key defects back to actual process measurements. Two process measurements in particular (Pod Weight and Finger Height) had a direct impact on the customers experience of the product. Operators measurements for one machine are shown in Exhibit 6. Pod weight provided a measure of the pod coming off the pod machines, and could be related to excess or insufficient reagent. If the pod weight were too low there would be insufficient reagent to develop the photo. If the pod weight were too high, there would be excess reagent and it would leak out the back. Finger height provided a measure of the height of the small plastic tab that controlled the advance and ejection of the frames in the cartridge: the finger was designed so that only one frame at a time would be ejected from the camera after each exposure (see Exhibit 7). The injection molded box endcap with finger was ultrasonically welded to the box during the assembly operation. If the finger height were too high the frame would not eject, causing a frame feed failure. If the finger height were too low, two frames would eject at the same time, causing a double feed.