Most system development efforts have traditionally been one-shot affairs. A temporary team is put together. Technicians draw ideas from the people in question. While all eyes are fixed on a limited budget set up for the project, system elements are designed until the money is gone. Then they are gone. As often as not, the people in question, the ones that are to use the system, find that they can think of all kinds of enhancements to the system that they hadn't thought of before. Having been awakened to the possibilities, they can see more and more of them.
Sternly and with a professorial air, the technologists basically say, "Tough luck, you had your chance." In most cases, a good deal of time will go by before you get them back. By that time, your zeal may well be gone. You and your team may be gone to other activities as well.
All of this, ironically, contradicts a basic benefit that computers were expected to provide. What did Church and Turing say in their concurrent papers on the subject in 1936? That recursive thought was the characteristic of human thought that would need to be provided by computers to be useful. Reflection. That is the thing. Maybe it is the third or fourth time thinking through an issue that the really good ideas come. Maybe the tenth. This is how we learn. By truncating this process, we shouldn't be surprised that systems are not more relevant. Furthermore, we shouldn't be surprised that people do not invest more effort in thought than they do. Why should they bother?
With fluidity and the opportunity to loop through the creative process, the people of the organization can rise to new heights each time. In each case, they can also expand the scope of the system, adding more and more requirements of the organization until they have all been brought in. Of course, as that level is reached, incentives to invest more care into the capabilities of the system increase significantly. The system becomes far more useful, in fact, even fun.
When an anomaly is found, it can be studied and then incorporated into the system by those who know the system best. Those who do not can walk through the generative taxonomy trees without a lot of technical training as long as they are authorized to do so. The only thing that users will encounter is better answers, better solutions.