Chapter 2: The Conduct of Scientific Investigations

• Lemmingineering n: the process of engineering systems by blindly following techniques the masses are following without regard to the appropriateness of those techniques.

2.1 The Principles of Scientific Investigation

Engineering as a discipline has at its foundation the application of the principles of science to solve real problems in a real world. Empirical investigations serve to drive the engineering process. Structural engineers can design and build a good bridge because they understand the structural properties of the elements they will use to build the bridge. The structural properties of the materials used in bridge building were developed over years of measurement, testing, and analysis.

The engineering discipline presumes the application of an underlying science based on hypothesis, measurement, and experimentation. Software engineering is a very young discipline based on the tacit theoretical foundation of computer science. That software engineering is an engineering discipline is still an aspiration in part because computer science is not yet a science. We have yet to master the process of validating our theories of computation and software development in the laboratory.

The current best practice in software development appears to be software lemmingineering. We tend to follow the experts without ever asking if we want to go where this leader is taking us. We tell ourselves that if everybody else is taking a certain path, it must be the right way to go. However, we do not even ask if the leaders know where they are going. The path to our current state-of-the-art of software development is littered with examples of the application of lemmingineering methodology. Consider the example of computer-aided software engineering (CASE) tools. Not long ago, every software development organization had to have CASE tools or they were considered clearly out of the mainstream. There were major worldwide conferences on CASE tools. Everybody who was anybody was using CASE tools because everyone else was using them. What was missing was the empirical validation that CASE tools somehow enhanced the software development process or the products developed under CASE tools were superior to those that were not. The use of CASE tools just felt right. In addition, everyone was doing it. Thus, there must be some good in it.