The term ‘knowledge engineering’ was coined in the 1980s to reference the processes whereby knowledge was elicited from human experts in order to develop knowledge-based systems. It was seen as reflecting an alternative paradigm for system engineering in which, for systems which were difficult to analyze in themselves but were subject to human activities, one modeled the human operators' skills rather than the system itself. In the 1980s, expert systems development appeared radically different from conventional systems development, but in the 1990s it is time to re-evaluate the reality and significance of the differences. The growth of expert systems development coincided with that of high-performance workstations, improvements in the efficiency of symbolic programming languages, and the development of graphic user interfaces. Much of what has been attributed to ‘expert systems’ may be seen as a halo effect of these other technologies. More fundamentally, the knowledge acquisition community has moved from an ‘expertise transfer’ to a ‘knowledge modeling’ perspective, in which knowledge is seen as not so much transferred from the expert as built in conjunction with the expert as a means of emulating his or her skill. This paper develops a modeling framework for systems engineering that encompasses systems modeling, task modeling, and knowledge modeling, and allows knowledge engineering and software engineering to be seen as part of a unified developmental process. This framework is used to evaluate what novel contributions the ‘knowledge engineering’ paradigm has made, and how these impact software engineering.