Session Reports - Sunday and Monday
8:45 AM Sunday: PhD Colloquium
The 2007 International System Dynamics Conference (ISDC) was the 25th International Conference and also celebrated the 50th year of the founding of the field of System Dynamics (SD). It was held July 29th – August 2nd, 2007 in The Seaport Hotel at Boston, MA, U.S.A. Preceding this milestone event, for a full day on Sunday, July 28, 2007, exciting presentations were made and lively discussions held on the platform of The Ph.D. Colloquium. While the colloquium has become a regular feature of the ISDC, the 2007 Colloquium was also special in several ways;
Given that this Colloquium was exclusively designed, developed, and delivered by graduate students coordinating well in advance across geographies, it is admirable to see its excellent outcome reflecting superior content and critique. The moving spirit propelling this forward were two outstanding Ph.D. students, Stefan Groesser of University of Berne (Switzerland) and Chintan Vaishnav of M.I.T. (U.S.A.), who voluntarily organized the Colloquium for many months. Their passion for quality SD and personal high values were reflected in the excellence defined by this Colloquium and the degree of responsibility they effectively executed in a globally integrative endeavor.
Professor Jay W. Forrester, the founder of the field of system dynamics, attended part of the Colloquium with interest, even when he had other commitments that day.
The impact of this milestone Colloquium will be felt for sometime in both, the emerging SD ideas, hypotheses, and theories that were presented and discussed as well as the benchmark this event has set for future colloquia. Full Report
Usman A. Ghani
Participatory Methods -Multiple Presentations Throught the Conferenc
At annual conferences, we have come to expect a number of papers on participatory methods and issues related to modeling work based upon stakeholder groups. The Boston Conference was no different. A superficial examination of the program revealed eight presentations, ranging from new scripts for group-modeling work to experimental assessment of the effect of individual vs. group structure upon model-based decision making. This session report highlights these contributions and calls for interest in the formation of a new SIG or Chapter focused upon this line of research. Full Report
9:00 AM Monday: Plenary Session: Strategic and Project Dynamics
Rogelio Oliva presented Hazhir Rahmandad and James Lyneis and they exposed two important aspects under the strategic management framework.
The session was opened by getting deeper on the explanation of the structure that generates the policies formulation in the organizations, Hazhir told us about the capabilities of the organizations classifying them as level zero and capabilities of level one, where the zero level capabilities are those that allow the survival of the organizations and the level one capabilities are those that modify or alter the level zero capabilities, both capabilities are responsible for the overall performance of the organization. Through some graphics, Hazhir left us clear the importance of balancing the investment in both capabilities for achieving equilibrium, on one side to facilitate the efficient operation of the business, and on the other to optimize the efforts dedicated to improve the structure. Before closing his presentation Hazhir explained us the effect of the competitors once the equilibrium has been reached, as a matter of conclusion he emphasized the importance for investing in the level one capabilities, with doing so we are facing a worst before better effect but in the long term that would help to eliminate the myopic policies.
As a product of the strategy formulation arises businesses projects; the analysis of the performance behavior of these projects has been an application in which system dynamics has had great success, James initiated its presentation showing the facts on the development of Project Dynamics as well as formal studies, as its application in diverse industries and its growth and benefit leverage for the services firms. Through the session we were faced with the Project Dynamics history, emphasizing those outstanding works that were prizewinners; James mentioned us the facts that established the precedent for the development of the two areas of Project Dynamics, first the body of knowledge in order to improve the project management field and second the need to justify the decisions taken by project managers in terms of a legal dispute. James show us the theory underlying Project Dynamics; and we realized the difference among the performance planned, the real performance and the effect that has adding new resources (and of course communication channels), the rework that arose by the learning curve among many others to conclude with the upcoming of the all these factors as feedback of this vicious circle. Towards the end of his presentation, James explained us the generic steps to apply Project Dynamics to a legal case, as well as the application of this method not only for legal justifications, but for taking advantage of other opportunities such as to improve the project management cycle and so on. James ended up with some recommendations for the growth of Project Dynamics.
|Rogelio Oliva||Hazhir Rahmandad||James Lyneis|
11:00 AM Monday: Parallel Session: Formulation and Parameterisation
The first parallel session for Formulation and Parameterization for the 2007 International Conference of the System Dynamics Society was held on Monday, July 30th, at 11am, and was chaired by Yaman Barlas. The parallel session featured three presentations. The first presentation was given by Gönenc Yucel, on a paper co-authored with Yaman Barlas, titled "Pattern-based System Design/Optimization." The second presentation was given by Rudolf Kulhavy, on the paper titled "Bayesian Analysis of Stochastic System Dynamics." The third was presented by Erling Moxnes on "The unavoidable a priori, revisited, or deriving the principles of SD." The novel contribution of the first paper is the development of a 2D pattern recognition algorithm and an optimization heuristic for searching for parameter values that yield model behavior similar to the desired one in terms of pattern characteristics. The second presents a particle filter algorithm for estimating state variables and unknown parameters in an SD model, using sequential Monte Carlo approximation. The third paper confronts criticism of the system dynamics methodology and presents SD as consistent with advanced forms of quantitative analysis such as optimization under uncertainty and Bayesian statistics. Full Report
Jeremy B. Sato
11:00 AM Monday: Parallel Session: Project Management Dynamics
The session was chaired by Ginny Wiley and consisted of three presentations:
Project Controls to Minimize Cost and Schedule Overruns: A Model, Research Agenda, and Initial Results, by Timothy Taylor, David Ford and James Lyneis -- David Ford, professor at Texas A&M University, gave a wonderful presentation of an on-going research of applying a system dynamics approach to examine project control policy for the purpose of minimizing cost and schedule overruns. The project dynamics model consisted of a "rework cycle," controlling feedback loops, ripple-effects and secondary effects. The model focused on supporting the project control decision for labor and systematically evaluated the effects of three common control options: work overtime, change workforce size, and change work intensity (work harder/faster). Contrary to the conventional mental model which assumes there is no impact on rework when taking control measures to meet the schedule, realistic scenarios need to reflect the negative impacts on project rework resulting from fatigue, pressure, and inexperience workers. The research team applied the model to a test project with 500 tasks to be completed by 20 people and a rework fraction of 20%. The preliminary results demonstrated how the common control measures may backfire and cause unintended increases in costs and delays to the schedule.
The Dynamics of Software Testing, by Anas Tawileh, Stephen McIntosh, Brent Work, W.K. Ivins -- Anas Tawileh, Ph.D. candidate at Cardiff University, U.K., developed a dynamic model for gaining insights to and investigating the impacts of software testing. The model focused on the software development phase, thus requirements and O&M phases are not part of the scope. Tawileh methodically illustrated the use of model-using simulations on a hypothetical project. For example, the model can be used to evaluate the impacts of different levels of people review by comparing the "costs" of testing against the "savings" from reduced errors. The model can also be used to evaluate different testing strategies that utilize different mixes of people review, automated analysis and execution testing and tools. Tawileh also imparted the audience a valuable insight from model simulation: In planning for software testing, it is important to consider the full impacts of testing on project costs and schedule by taking into account not only testing staff, but also to consider the increased error discovery and increased rework to correct these errors.
The Dynamics of ERP Success, by Margaret Fryling --Meg Fryling, graduate student in SUNY Albany, shared her experience and preliminary analysis of the dynamics of implementing the commercial package of enterprise resource planning (ERP) software. ERP systems have been touted to offer benefits of increased efficiency and a higher level of customer service. However, the success rate is around 33%, and 90% of ERP projects are late or over budget. For example, the implementation at the University was originally estimated to be 2 years but has become 7 years. One of the issues was the lack of appreciation for additional work resulting from initial customization during the planning stage. With every quarter update or software upgrade, the initial customization portion requires additional efforts. Therefore, the manager needs to carefully balance the desire to increase user satisfaction through customization and the additional efforts needed for customization through the life-cycle of the system. The research question is "Is the Information System success framework appropriate for ERP implementation?" Fryling approaches this by examining feedback behaviors.
11:00 AM Monday: Parallel Session: Mental Models and Managerial Performance
The session "Mental models and Managerial Performance" evoked a good response from participants at the conference. Despite an afternoon session, both, the presenters and the audience of this session were quite vibrant. The session included presentations on a wide range of topics which were held together with a common theme – mental models. The presenters (Luc Cassivi, Bill Lawless and Shayne Gary) seemed satisfied with their presentation and with the feedback they received from the audience.
There has been widespread application of system dynamics modeling to software development/software project management. The paper by Araujo et al. - Improving the Software Development Process: A Dynamic Model Using the Capacity Maturity Model, was on a similar theme. They discussed about the SEI Capability Maturity Model (CMM). Their study focused upon identifying the benefits associated with implementation of CMM Level 2 practices for the smaller software development firm. The five levels of the CMM model were outlined for the benefit of the audience. A Powersim-based simulation model was presented that the authors claim would result in a better understanding of the impact of alternative management policies and practices.
In the second presentation, Bill Lawless presented a nonlinear approach to game theory. The presentation discussed two issues - the arbitrariness of valuing cooperation greater than competition in determining social welfare and the lack of interdependent uncertainty. The presentation was quite different from the other two as it focused on mathematical approaches. Nevertheless, many found it to be interesting and it evoked a good response from the audience.
Shayne Gary’s presentation (co-authored with Bob Wood) was a sequel experiment of the one presented at the 2006 conference. In an experimental study, the authors show that native performance of people in dynamic decision-making tasks is quite poor. Further, they aim to explore how this deficiency in human decision-making could be improved. They hypothesized that providing decision support aid (in the form of causal loop diagrams of the task) to the participants would improve the accuracy of their mental models and improve performance. Their results indicate that task complexity and decision aid support are significant predictors of mental model accuracy, and that decision aid support moderates the relationship between task complexity and mental model accuracy. Many in the audience could relate the two experiments conducted by Gary and Wood and felt that such fundamental studies were vital in understanding our decision-making process.
2:00 PM Monday: Parallel Session: Information Security
The first presentation of the Information Security parallel session, titled "An Experience Using System Dynamics to Facilitate an Insider Threat Workshop," was given by Andrew Moore from CERT, Software Engineering Institute and CyLab at Carnegie Mellon University. In his presentation, Andrew gave a brief account of the evolution of insider threat modeling and how system dynamics was utilized to facilitate and support interactive discussions in a series of workshops conducted as part of the research. The workshops, and the supporting model, focused mainly on the insider IT sabotage. In order for the developed model to be validated against empirical data, 64 real world cases of insider IT sabotage were identified and analyzed for relevance to the task at hand. Not all of these cases proved to be useful, as only 30 of which included sufficient information to allow practical modeling.
Several interesting observations were made during the modeling in the workshops. Annotations of arrows in influence diagrams were found to cause confusion for those less experienced with system dynamics. They were removed from the diagrams to solve this issue. Another point that was observed is the question of learning and how it can be facilitated using system dynamics tools and techniques. Many questions were posed in this regard: are we learning the right lesson? Are we learning and changing? What did we learn about using of system dynamics in to communicate complex concepts to an audience of business and IT managers?
|Andrew Moore||Slide from Andrew Moore's presentation||Navid Ghaffarzadegan|
The second paper, A Preliminary Model of The Vulnerability Black Market, presented by Jaziar Radianti from Agder University College in Sweden described the authors’ attempt at modeling the dynamics of the black market for information systems vulnerability. The presenter gave a quick introduction to the definition of the vulnerability black market and stressed the importance of understanding its dynamics in order to deal with it more effectively. Afterwards, the historical evolution of vulnerability disclosure policies and its impact on the vulnerability black market was examined and the preliminary modeling activities and findings discussed. Comments from the audience focused mainly on the difficulty of obtaining reliable information to validate the model in this particular setting. Software development firms are usually reluctant to disclose information about the vulnerabilities of their own applications, and the black market is secretive by definition, and interacting with its members may become a significant challenge.
Navid Ghaffarzadegan’s presentation: "How a System Backfires: From a Redundancy Solution to Redundancy Problems in Security" was the last in this session, and tackled the very interesting counter-intuitive problem when the implementation of more guards and countermeasures may backfire and actually decrease the security of the protected system. Navid’s argument is based on Sagan’s hypothesis and aims to provide an explanation using system dynamics. He attempted to examine the causes of the backfiring of redundancy solutions, and suggests that the problem may arise from one of several possible reasons: insider threats, social shirking or overcompensation. The modeling and simulation findings denote adequacy of the controlling capability (the managerial abilities in controlling guards) as a high leverage policy lever to reduce the possibility of a system backfire. It was suggested to the presenter that social shirking is largely a cultural phenomena that differs greatly between different cultures. Also, the audience challenged an assumption that nations strive to match the power of terrorists, while in reality, they try to remain superior to the enemy’s power at least 3-5 times.
2:00 PM Monday: Parallel Session: Innovation and Growth
1. Untangling the Origins of Strategic Innovation. A System Dynamics Approach, by Carmine Garzia -- Innovation in strategic positioning enables companies to redefine the way to do business delivering more value to customers and achieving superior competitive performances. A simulation model is presents how the firm’s ability generates and implements innovation in strategic positioning. Three factors determine this. Firstly it is related to the ability to govern the technical innovation process. Secondly strategic innovation processes occur in firms characterized by a certain degree of entrepreneurial orientation. Entrepreneurial orientation is deeply influenced by the introduction of organizational innovations that allow the release of the entrepreneurial energy embedded in the organizational structure. In other words, is provides the answer to the question "how can a large organization develop and maintain a culture of innovation." And thirdly, technical innovations become strategic innovations only if top managers are able to manage a process of integration through which they integrate new technical initiatives into the company’s strategy. After the presentation, questions about the external validity of the model occurred. It became obvious that the model is purely conceptual at this time. It is intended to increase the validity in further steps.
2. Achieving Competitive Advantage in a High Turnover, Dynamic Market: A Causal Model Approach, by Thomas Clark, Mary Jones -- The presentation shows a dynamic perspective to establish a causal model of how organizations can achieve and maintain competitive advantage in highly competitive situations where unique resources do not exist. The study is grounded in the resource based theory of the firm, and the model is developed from research surrounding this theory and from an extensive fifteen month in-depth examination of an organization. The results suggest that keys to achieving a competitive advantage arise from the ability to establish a strong competitive culture, detailed execution of organizational processes, and establishment of a constant sense of urgency in preparation and execution of plans. The model was, at the moment of the presentation, only a qualitative model with the following sectors: competitive quality of the culture, team coaching environment, coaching staff structure and other similar structures. The discussion eluded that the simulation model will follow soon, but that the quantification of soft variables are highly challenging the modeling process.
3. To Cluster or Not to Cluster: A Simulation Study of Managerial Practices for Innovating in Small and Medium Entities, by Martin Kunc -- The presentation is about the relationship between firms and clusters. One of the several reasons for the increased interest in clusters is the positive association between networks within geographical boundaries and knowledge diffusion. The presentation tried to answer the following questions: What will happen if firms in the cluster lack of skills to absorb information? What will happen if clustered firms do not have rich interactions? The presenter showed that manager of Small and Medium Enterprises need to decide either to emphasize the inter-fim interactions generated inside a cluster or to follow a standalone process without clustering in order to obtain the additional benefits. The discussion elicited that the model creation process was especially difficult when it came to the model calibration phase. Even though, the model is a conceptual, it provides interesting insides in this field of strategic management.
2:00 PM Monday: Parallel Session: Process Improvement
This session was chaired by George Roth and included three presentations: Improving Maintenance Operation through Transformational Outsourcing , by Jacqueline Ye, The Good, the Bad and the Mediocre: Creating Insightful Stories on Process Improvement , by Markus Salge and From Continuous Improvement to Centralized Information: The Life and Times of a Systems Thinking Intervention , by Jason Jay, George Roth.
|Jacqueline Ye||Markus Salge||Jason Jay|
This session was attended by approximately 20 - 25 people. All of the speakers were energetic and articulate. The session was introduced and moderated by George Roth. Full Report
Michael D. Okrent
2:00 PM Monday: Parallel Session: Equality & Public Policy
During this session we could enjoy, first, two interesting presentations that added to our understanding of how certain public policies affect two important pillars for more equality in the future: the empowerment of women and the chance of quality education for the poor. In the same frequency, the third presentation explained the counterintuitive results of Mandatory Arrest Policies.
Richard Dudley, started by challenging the assumption that the reason for so few women in leadership positions (in the USA) is merely the absence of enough talent in the pipeline. After explaining his approach (for data collection and analysis) and the structure of his model he suggested that an alternate explanation is that women’s movement throughout the employment chain is limited because they are tagged as homemakers.
Schaffernicht explained how Educational Quality and Equality in Chile are so hard to reconcile. He realized a qualitative analysis that allowed better understanding of how a given institutional setting, an already unequal distribution of wealth and the maximizing choices of the three groups of agents (schools, teachers and parents) result in two strong reinforcing loops that brings good teachers to work for the best and more expensive schools. He showed how, unless serious changes are realized in the education policy with more involvement from the government- the new Education Bill just approved (that demand the right of quality of education for every child) will remain an utopia.
Finally, Hovmand showed how Mandatory Arrest Policies implemented hoping to reduce Domestic Violence by causing deterrence, have also caused the increase of women being arrested, since men, with the more times they are arrested, seem to get experience in the legal process and learn to manipulate the system in their benefit.
7PM Monday: Poster Session: System Dynamics - the State of the Field
The poster session took place in the Plaza Ballroom in conjunction with a buffet dinner in a very informal atmosphere that contributed to stimulate the debate among participants.
The session was dedicated to the exploration of methodological issues in the system dynamics field.
In the paper "What constitutes System thinking" a group of researchers from Las Vegas University presented an exhaustive review of SD literature to explore the relations between system dynamics and System Thinking. Margarita Vasquez and Manuel Liz proposed and original constructivist approach to understand the philosophical foundation of system dynamics. The ideal complement to this last paper is the work of two German scholars, Milling and Groessler, "Inductive and Deductive System Dynamics Modeling," that investigates the contribution that other methodologies can give in applying effectively system dynamics simulation results to support strategic and organizational change. Finally, the paper "A Research On System Thinking Based on Three Dimension Thought" addresses the important issues of subjectivity of SD models and suggests a 3-way control to reinforce external validity of models.
7PM Monday: Poster Session: Methodology: Alternative Approaches - Steps of Modeling
The session originally included six interesting works mainly related to modeling steps and alternative approaches to modeling in which authors explore integration possibilities with other approaches like multi-criteria decision analysis. However, two of the posters failed to show up; "Validation of System Dynamics Models: A Taguchi Methods-Based Approach" by Hadjis and Papageorgiou, and "Evolving System Dynamics Models, Model Identification By means of Evolutionary Computing" by Bergbauer.
In the "Uncovering Complex Relationships in System Dynamics Modeling: Exploring the Use of CHAID and CART" titled poster, Alexandra Medina-Borja, and Kalyan Pasupathy were introducing two approaches that may help the modeler at different stages of the model building process, but especially during the structure identification stages. Discussed approaches are introduced to be able to assist the modeler in mining databases in order to reveal some relationships between components and possibly transform these into mathematical equations. To be brief and short, this poster introduces these two approaches as relationship identification and confirmation tools that can be used on databases of raw data.
Just besides the former poster, we had the chance to see the work by Michael Schwandt, Kostas Triantis and James Glenn titled "Intercollegiate Athletic Departments Performance Assessment." The approach discussed by the authors can be characterized as a SD-supported multi-criteria performance evaluation. The first stage of the approach is a traditional multi-criteria performance assessment of a set of colleges using data envelopment analysis. This analysis reveals the set of colleges inferior to the others in terms of performance. At this point a SD model is utilized in order to reveal possible underlying causes of this inferior performance. Apart from the research itself, it was quite interesting to see the athletics budgets of the colleges considered in the study.
The third poster in this session was "Dealing with Uncertainty? Combining System Dynamics with Multi-criteria Decision Analysis or with Exploratory Modeling" by Erik Pruyt. During the session Eric presented the issue of uncertainty and the lack of emphasis on this issue in most of the SD studies, and seemed to have lively and fruitful discussions with the participants. After mentioning the basic concepts of uncertainty, resilience, flexibility, risk and robustness, the poster proposes the utilization of Multi-Attribute Multiple Criteria Decision Analysis as a supporting approach to SD in order to enhance its terms of dealing with uncertainty.
The location-wise outlier of the session (since it was located on the other side of the poster hall) was the poster by Anderson, Richardson, Ackermann and Eden; "Two Group Model Building Scripts that Integrate Systems Thinking into Strategy Workshops Facilitated with Group Explore." Considering the expertise of the authors in group model building, the poster mainly discussing the experience of the authors with two different scripts used in a group model building exercise was quite appealing. Two different scripts (script #1 mainly emphasizing system over-time dynamics, and script #2 mainly emphasizing system structure) and the results obtained by using these scripts was compared and discussed briefly in the poster. The poster seemed to attract a considerable audience group interested in group model building processes.
|Eric Pruyt and his poster "Dealing with Uncertainty? Combining System Dynamics with Multicriteria Decision Analysis or with Exploratory Modelling"||
Left: "Uncovering Complex Relationships in System Dynamics Modeling: Exploring the Use of CHAID and CART" by Alexandra Medina-Borja and Kalyan Pasupathy
[ Previous] [ Return to contents ] [ Next]