Unsystematic Engineering

Published in IEEE Spectrum Magazine, September 2006

Through the years I’ve often had the experience of participating on panels convened to study some particular problem, usually at the behest of the government. Typically there is some program, project, or issue that involves engineering and science, and is not seen to be doing as well as was desired or anticipated. A group of engineers and scientists is assembled and given the charter to look into this troubled issue. The group gets numerous briefings from people involved in the program and from outside experts. Finally, a report is prepared with recommendations about how to fix this trouble.

In my own experience these final reports almost always contain the elements of three recommendations:

1. Give more money.
2. Put someone in charge.
3. Someone needs to do systems engineering.

Such reports are invariably received courteously by the sponsors, but often with a muted enthusiasm. I can see the wheels turning behind their eyes. “We knew this stuff,” they are thinking. “You people are supposed to give us a technical solution, not tell us how to manage this. You’re engineers, not management experts.”

I have the feeling that the study’s sponsors believe or hope that there is some great new technology that will fix whatever it is, but that never seems to be the case. Almost always the engineers and scientists working on the program are perfectly aware of every applicable technology. But what happens is that all their efforts are poorly coordinated, responsibility is diffuse, and there is insufficient management oversight and systems engineering or architecture at the top level.

This brings up a conundrum that has long puzzled me. If systems engineering is so valuable, why is it so seldom practiced? I have my own opinions, which are not based on any facts or special knowledge of the systems engineering field, but simply on my engineering instincts, which I know can be dangerously misleading. Nonetheless, in the interests of starting debate I’ll proceed as if I know what I’m talking about.

First, academic prestige is based on expertise and reputation in narrow disciplines. Our engineering teachers have gained their authority by knowing a great deal about some special subject, rather than having broad general knowledge – they have depth, rather than breadth. This is the culture in which we engineers are trained. Perhaps because of this cultural indoctrination, we instinctively admire a world expert in something so esoteric as to be inconsequential, while disdaining the generalist who lacks the prestige of specialist knowledge. I admit that I have often myself thought that I could admire a generalist only after he or she had demonstrated expertise in some special domain. Thus the problem – can a systems engineer be created from whole cloth in college, or must the person be grown up first through a specific discipline?

The education problem for systems engineering is exacerbated by a nagging worry about its mathematical foundation. In short, is there a “there” there? We engineers are used to building on the foundation of a relatively small set of mathematical rules, Maxwell’s equations being the prototypical example where everything can be reduced to the application of a small set of equations. This kind of world is ordered and fits very nicely into textbooks and test questions. In contrast, systems engineering is often based on experience and common sense, and we know where common sense fits on the hierarchy of things that justify a high salary. Perhaps it is also true that systems engineering is more of an art form, less mechanical, and harder to teach than the traditional disciplines.

Then there is the whole way we go about engineering large systems. Divide and conquer is the usual approach. Pieces of the system are parceled out to different engineers, different organizations, and different companies. Any possibility of a holistic approach is foregone from the very start. Even systems integrators often have to deal with existing or pre-planned piece parts. So everyone can do their job right, and the resultant system can still become a suboptimized kludge.

Through the years various universities have offered programs and degrees in systems engineering, but the results have not always been judged successful. Moreover, some famous engineers, realizing the need, have gone to academic institutions to teach what they have learned about systems engineering. Unfortunately, it may have been that their skills and experience were not easily replicable and extendable to the larger teaching force.

In recent years a number of well known universities have begun new programs in systems engineering. Maybe now is the time for these programs to become successful. If not, we can always call for a panel to study the problem. I wonder what it would say!