The Design of Innovating Machines: A Fundamental Discipline for a Postmodern Systems Engineering

Systems engineering has been the Rodney Dangerfield of engineering disciplines: it gets no respect. On the one hand, everyone now claims to “do systems,” but few are willing to specialize in it (thinking that doing so is by definition an oxymoronic enterprise). Moreover, those who do specialize in systems can’t seem to agree on what it is. Fortunately, a growing number of engineering programs in the US and around the world are looking at systems engineering anew with an eye toward building a consensus on what this paper will call postmodern systems engineering. Indeed, the MIT Engineering Systems Division Symposium is an important event toward building that consensus, and our debates today may shape the future of systems engineering research, teaching, and practice in important ways for years to come. Although the consensus has not yet emerged, it is increasingly clear that it will not be your grandmother’s system engineering. Where once system engineering was largely a Cold War tool for planning, executing, and controlling large-scale space, military, and industrial projects necessitated by the space race, various arms races, and industrial needs, the postmodern consensus seems destined to view systems engineering in a context of democratic, post-industrial capitalist economies (Fuyukama, 1992), where human needs and influences are well integrated into the engineering process. The last phrase of the last sentence in the preceding paragraph is a tough challenge. To better account for human needs and influences, systems engineering must retool itself and build new models and disciplines. Discussing the full range of disciplines and models needed is beyond the scope of this paper; however, here I suggest that postmodern systems engineering requires a better understanding of the theory and processes of human innovation. Systems engineers of times past failed to anticipate the innovative capability of their users, adversaries, and colleagues. By better understanding processes of human innovation, postmodern systems engineers will be better able to (a) harness the innovative capability of their own organizations and (b) anticipate or, at least, better understand the unintended innovative ability of those who will use or try to defeat the system designed. Creating an engineering understanding of human innovation is a daunting task, and if we were to start de novo the odds would be quite long; however, more than twenty years ago, I likened the operation of genetic algorithms (GAs) to certain processes of human innovation (Goldberg, 1983). Genetic algorithms are search procedures based on the mechanics of natural genetics and selection (Goldberg, 1989). At the time, my aim was to give a plausible explanation of GA power to new readers in an effort to connect with those who might otherwise find the operation of GAs somewhat suspect. Since that time, I’ve come to realize that the connection between GA mechanics and innovation is much closer than I originally surmised. And in that story lies the threefold purpose of this paper and the book upon which it is based (Goldberg, 2002):