Who Are We, and What Are We Doing Here?

Abstract

Many Formal Methods researchers and practitioners seem to treat Formal Methods more as a religion than as an approach to rigorous software engineering. This fervour has a few side-effects: i) There have been spectacular advances in a few areas in Formal Methods; ii) There are a significant number of highly effective Formal Methods advocates - and practitioners; iii) The Formal Methods community at large seems to be condescendingly dismissive of any protestation of disbelief; and iv) Different methods and approaches seem to be judged on a belief basis rather than through evidence based analysis. The essential fact remains though, that after decades of research, Formal Methods are not used much in industrial software development. It is time that we, the Formal Methods community, question the basis of our existence. I argue that we exist to further the use of mathematics and rigorous analysis in the development of software applications, in the same way that electrical engineers, mechanical engineers, civil engineers, chemical engineers further the safe and effective development of a multitude of devices, buildings, manufacturing processes etc. This is clearly not a new thought. It does, however, suggest that we need to examine the link between Formal Methods and Software Engineering more carefully than is currently the case. A definition of engineering from the Academic Press Dictionary of Science and Technology is “the application of scientific knowledge about matter and energy for practical human uses such as construction, machinery, products, or systems”. Engineers use science as the basis for their work. This is not a one-way street. Feedback from engineering as to what are the important scientific problems to be solved is an important driver in scientific endeavours. Engineering work, in turn, forms a basis for the work done by technicians in our everyday lives. Again, feedback is an essential driver for the engineering community. In the modern digital world, Software Engineers should assume the role of the engineer. If we are truly serious about Software Engineering as an engineering profession, we need to consider the roles of Computer Scientists and Software Developers in this context. To be consistent with other domains, Software Engineers should use scientific knowledge as the basis of their work. This knowledge includes the growing domain of knowledge generated by Computer Science, and in particular, the specialized forms of mathematics that are applicable in the digital domain. In addition to Computer Scientists and Software Engineers, we also have Software Developers the technicians of our domain. This is a nice and neat correlation with other engineering fields unfortunately it is not, at this time, an accurate description of the situation. In most countries, the difference between Computer Science and Software Engineering is decidedly blurry. Even when the difference should be obvious (for example, Canada insists that to call yourself an “engineer” you must be recognized as such by a professional engineering accreditation body), it is commonplace to find Computer Scientists playing the role of both engineer and technician. What does this mean for Formal Methods? Are Formal Methods people Computer Scientists, Software Engineers, Software Developers all of the above any of the above? If you look back at what I said about our raison d’etre, and if you agreed with what I said, perhaps you agreed too quickly!