Modeling: what we have yet to learn

This week Bill Gates noted that we need better modeling tools for software in order to solve some of the persistent and hardest problems facing the planet. I agree, but I would suggest to him that we not only need better tools. We also need better concepts and techniques for modeling. Some research on modeling human-machine systems by my friend Andrew has indicated that modeling techniques still lack meaningful representation in complex systems [1].

I have reviewed a number of papers this year analyzing Gantt charts and their capacity to improve project management. In each case, Gantt charts tend to get used as eye candy with no impact on improving project outcomes. The idea is to have a representation a group can use to jointly visualize complex data and understand it [2], but these current representations are still inadequate. So, simply making software that add digitizations of current concepts will not be enough. We need new concepts.

Ideas from agile project management are providing some of these new concepts. We see nico-nico charts and kanban burn charts getting some usage and making a difference anecdotally. Certainly, I have added some from my own research, particularly around the understanding of what to monitor and when intervention will be needed [3]. Understanding what a leader or leadership group needs to monitor can inform what entities and events need to be modeled. So, for example, in my research I can juxtapose the common designs in software such as Microsoft Project with the implications of my work. In Microsoft Project the dominant design focuses on tasks on a schedule. They have budgetary implications and resource requirements. As a result, a leader can see models of the current time and budget status of a project and get some information about the scope completion. The underlying concept here is the triple constraint (time-budget-scope). The point of monitoring it for the leader is to head off any trouble and keep the project on target.

Unfortunately, my work and the work of a variety of other researchers shows that these indicators only show up once problems have fruited. Triple constraint violations need to be monitored but only secondarily. Primary concerns such as group relations, capabilities, interference, and technological breakdowns are no where to be seen in the status reports in Microsoft Project. I would suggest that this is a good place for us to start in our modeling improvements, and I must admit, I helped do some testing and improvement on the Team Foundation Server underneath the current iteration of Microsoft Project for IS projects back in 2005. So, I too am guilty. Much work is needed in this area, and as Gates points out, we need more workers who understand modeling and can at least attempt to build better models. Thus, I emphasize modeling as a method to design better systems and teach it to undergrads and grads in my courses.

[1] A. Burton-Jones and P. N. Meso, "Conceptualizing Systems for Understanding: An Empirical Test of Decomposition Principles in Object-Oriented Analysis", Information Systems Research, 17 (2006), pp. 38-63.

[2] E. Bendoly and M. Swink, "Moderating effects of information access on project management behavior performance and perceptions", Journal of Operations Management, 25 (2007), pp. 604-622.

[3] D. M. Thomas and R. P. Bostrom, "Vital Signs for Virtual Teams: An Empirically Developed Trigger Model for Technology Adaptation Interventions", MIS Quarterly, 34 (2010), pp. 115-142.