The development and simulation of formal models are widely supposed to have a positive effect on the learning about nearly any social system (e. g. Milling 1991, p. 20, de Geus 1988, p. 73, Morecroft 1994, p. 4: "So, at some level, managers must build models." and "So, managers must experiment with models,..."). To give people an easier and more userfriendly access to formal models, pre-built simulations with graphical user interfaces are used. Thus, the user does not have to go through the difficult and time consuming process of model development himself (compare the seven levels of System Dynamics-Competence formulated by Meadows 1989, p. 636). With the help of so called Management Flight Simulators (MFS) for individuals or Computerized Planning Games for groups, users should be able to gain insights into the dynamic structures of a problem. Although their validity has not yet been proven , there are some hints that these management simulators promote learning (Milling 1995, p. 106).
One precondition for a learning transfer to take place is a homomorphy
between the real world domain and the formal model that is used
in the simulation tool (the management simulation becomes a transitional
object as mentioned by Papert (1980) which makes a learning transfer
possible). A characteristic of reality which has no correspondence
in the artificial world of most management simulators is the proceeding
of time. This sounds paradoxical: those tools, which are built
to improve knowledge about dynamical, time dependent aspects of
reality, do only show any dynamics when asked to go ahead. Only
when the user has provided all necessary input, the simulation
calculates the results for the next period in time. These kinds
of simulations are called event-driven: when the user has finished
data input he or she presses a button and generates an event which
causes the simulation to proceed.
Of course, the above mentioned issue is not really a paradox: one has to keep in mind the aims of management simulations. The user should gain insights in the dynamic structure of the domain. This often cannot take place in real-time. Time has to be slowed down. The users need to have their time to analyze the situation, make decisions, etc. This stretching of time is as important as the acceleration of time (decision-reaction cycles) which is discussed by most authors. Learning Environments have to provide "opportunities for reflection" (Senge and Sterman 1992, p. 147), in order not to let users fall victim to the "video game" syndrom. This aspect differentiates MFS and "real" Flight Simulators; "Management Flight Simulator" therefore it is perhaps not an appropriate term for one-person business simulation games.
In reality time is an independent variable, which certainly cannot be influenced by the user. Thus, management simulations could be more homomorph to reality when they use a kind of real-time mechanism. Note that this can be implemented in two ways (Buchner 1995, p. 55):
Firstly, by introducing time pressure. Users have to decide and complete given tasks within a certain temporal limit which is externally set by the game provider. This method is often used by experimenters to make results of different subjects comparable or if competitional aspects of simulation gaming prevail.
Secondly, the system changes autonomously its state. Here, time pressure is not external but inherent in the task. This offers interesting perspectives on possible usage, task difficulty, and learning chances of business simulations. One can call these types of simulation systems clock-driven or self-proceeding; sometimes they are just called "dynamic tasks" (Brehmer 1995, p. 104).
The task of running a clock-driven management simulation successfully becomes more difficult because mental processes involved have to terminate earlier than in event-driven simulations. Furthermore, autonomous state changes are probably different from ones which are initiated by the user. The user has to distinguish between those system states caused by his or her decisions and those that are caused by autonomous changes of the simulation model (Brehmer 1995, p. 104). And in fact, learning is severely affected when a simulation system autonomously changes its state as a function of time rather than as a result of user input (Funke and Müller 1988). However, there seems to be no qualitative difference on the (counterintuitive) effects of delays when an event-driven or a clock-driven simulation is used (Brehmer 1995, p. 125).
Nevertheless, one can imagine situations where not learning transfer to a real world domain is the main aim of a simulation tool and therefore self-proceeding simulations seem to be an alternative to "conventional" business simulation games:
In consequence, the familiar concept of decision-simulation periods in management simulations seems to be unnecessary (even though, from a technical point of view, one always has time steps in digital computing): continuous simulations can run independently from discrete decision-simulation periods. Decisions show effects based on current data. As in reality, the state of the simulation changes autonomously. These changes in the environment require reactions of the users and have to be considered when crafting a strategy. Thus, "users have to anticipate the system's inherent changes due to the Eigendynamik" (Funke 1995, p. 258).
Such management simulations provide the users with extended strategic
possibilities: action and reaction at the right time are more
important than in "conventional" business simulations.
Possible learning effects are shifting from knowledge about the
structure of the underlying model to decision making itself.
Brehmer, Berndt. 1995. Feedback Delays in Complex Dynamic Decision Tasks. In: Peter A. Frensch and Joachim Funke (eds.): Complex Problem Solving - The European Perspective. Hillsdale, NJ.
Buchner, Axel. 1995. Basic Topics and Approaches to the Study of Complex Problem Solving. In: Peter A. Frensch and Joachim Funke (eds.): Complex Problem Solving - The European Perspective. Hillsdale, NJ.
de Geus, Arie P. 1988. Planning as Learning. In: Harvard Business Review, Vol. 66 No. 2: 70-74.
Funke, Joachim. 1995. Experimental Research on Complex Problem Solving. In: Peter A. Frensch and Joachim Funke (eds.): Complex Problem Solving - The European Perspective. Hillsdale, NJ.
Funke, Joachim and Horst Müller. 1988. Eingreifen und Prognostizieren als Determinanten von Systemidentifikation und Systemsteuerung [Active Control and Prediction as Determinants of System Identification and System Control]. In: Sprache & Kognition, Vol. 7: 176-186.
Meadows, Dennis. 1989. Gaming to Implement System Dynamics Models. In: Peter Milling and Erich Zahn (eds.): Computer-Based Management of Complex Systems. Berlin.
Milling, Peter. 1991. Strategische Planungs- und Kontrollsysteme zur Unterstützung betrieblicher Lernprozesse [Strategic Planning and Control Systems for the Support of Organizational Learning Processes]. In: Peter Milling (ed.): Systemmanagement und Managementsysteme. Berlin.
Milling, Peter. 1995. Organisationales Lernen und seine Unterstützung durch Managementsimulatoren [Organizational Learning and its Support through Management Simulations]. In: ZfB Ergänzungsheft 3/95: Lernende Unternehmen: 93-112.
Morecroft, John D. W. 1994. Executive Knowledge, Models, and Learning. In: John D. W. Morecroft and John D. Sterman (eds.): Modeling for Learning Organizations. Portland, OR.
Papert, Seymour. 1980. Mindstorms. New York.
Senge, Peter M. and John D. Sterman. 1992. Systems Thinking and
Organizational Learning: Acting Locally and Thinking Globally
in the Organization of the Future. In: European Journal of
Operations Research, Vol. 59: 137-150.
ISDC '97 CD Sponsor