Volume 23 – Number 4 October 2010
Session Reports (2)
The Wednesday morning plenary session entitled, Model Formulation Process, was chaired by Robert Eberlein. This session consisted of two presentations: 1) System Dynamics Problem Definition as an Evolutionary Process Using Ambiguity Concept by Ali Mashayekhi (presenter) and Soheil Ghili, and 2) Improvised Facilitation: A Third Leg on the Group Model Building Stool by David Andersen (presenter) and George Richardson.
Roberta Spencer started the session at 8:30 by giving the morning announcements. Roberta entertained the audience with a joke after her reminders about registration and Friday Bonus Day. It should be mentioned that many chairs in Olympia Grand Hall (the main hall of the conference) were occupied.
In this paper, the authors believe that if there is no ambiguity, the difficulty has been solved. Ali gave an example from Iran about the energy subsidy. In this example, the ambiguity is that the removal of subsidies cannot be implemented.
Ambiguity can play an important role in problem definition and formulation. They employed an example of a 'Real Estate Bubble' to interpret their viewpoints. Four different hypotheses were presented in this example and the results of each hypothesis are exhibited.
After Ali's presentation, Robert introduced the next presenter, David Andersen. David has an academic position at the State University of New York at Albany. He received his PhD degree from MIT.
David started his speech by sharing some memories from thirty-seven years ago, when he and George Richardson and Ali Mashayekhi were at MIT together. He said that Ali became a philosopher but he also became a plumber. He tried to explain the ways in which he and George produced this paper. Also, he told some jokes about their relationship during this work.
This paper expresses how a problem could be modeled better by the joint effort of modelers and facilitators. David, in his presentation, many times used his colleague's name to show and explain the process of paper generation. The presentation file includes many pictures of hand written concepts and models and the photos of their group.
The two presentations were finished at 9:15 and the question period started.
Six audience members had questions about the papers that were presented in this session. More questions were asked of David. However, Ali and David answered the second question together. They showed good cooperation with their answer. Sometimes David invited George to accompany him in answering the questions.
Hadi Akbarzade Khorshidi
This session, chaired by Erik Pruyt, contained three papers dealing with various aspects of epidemiological dynamics. The topics covered in the session included the infectious disease as a connected network, the spread of resistance to antibiotics, and the influenza A(H1N1)v pandemic.
Modeling Propagation of Infectious Disease as a Connected Network, by Lianjun An, Young Lee
The presentation introduced a system dynamics model that combined a compartmental epidemiological model with a connected network of geographical locations and airports. Key findings centered on how various mitigation policies could affect the spread of disease. One highlight was the application of the proposed model for a firm to study the possible effects of pandemic disease on its business. This is especially important as influenza may evolve, spread rapidly, and then form an epidemic in a short time window. One interesting point of discussion was how to develop global mitigation strategies for society as the world is highly connected as a small world network.
Simulating the spread of resistance to antibiotics, by Nabil Mikati
The presentation proposed a generic model of the transmission of commensal bacteria within a community exposed to different levels of antibiotics. Analytical expressions describing the conditions for reversibility are derived as well. Key points included 1) cutting back the volume of antibiotics is necessary but not sufficient to reduce resistance frequency, and 2) the biological cost for sustaining resistant traits and the lateral transmission of genetic material plays a critical role. An underlying assumption is that the epidemiology of resistance is a function of the innate characteristics of the bacteria, transmission rates and the consumption of antibiotics. An interesting point of discussion was that in some cases prudent usage of antibiotics leads to decrease in resistance frequency while there are also other cases where reduction does not lead to reversibility.
The Influenza A(H1N1)v Pandemic: An Exploratory System Dynamics Approach, by Erik Pruyt, Caner Hamarat
The presentation proposed a small exploratory system dynamics model to depict the 2009 flu pandemic, also known as the Mexican flu, swine flu, or A(H1N1)v. Specifically, the proposed model was developed in May 2009 in order to quickly foster understanding about the possible dynamics of the new flu variant and to perform rough-cut policy explorations. Of particular importance was the connection between 1) Exploratory System Dynamics models, and 2) Exploratory Modeling and Analysis. An interesting point of discussion was how to set up plausible behaviors and conduct a large scale of sensitivity analysis in a suitable platform.
Many questions and responses about how to perform reliable sensitivity analysis were raised after the presentations. Participants feel that a great interaction and communication had taken place in a pleasant atmosphere.
The parallel session on modeling approaches contained two papers about methods to analyze mental models and one paper that investigated how group model building can contribute to enterprise resource planning.
The first paper by Martin Schaffernicht and Stefan Groesser was entitled Automating the comparison of mental models of dynamical systems and presented a software tool to compare mental models. The paper focused on the conceptual architecture of the tool, its main data structures and proceedings. The software tool automatically calculates element distance ratios, loop distance ratios and model distance ratios to compare mental models. For this purpose a series of preparatory activities have to be implemented. The paper applies the software to compare five idealized mental models.
The second paper, Automated assessment of learners' understanding in complex dynamic systems, by Birgit Kopainsky, Pablo Pirnay-Dummer and Stephen M. Alessi, evaluated a software which performs measurement and analysis of understanding via verbal protocols. The software had been used in a wide range of application domains before. The paper shows that it is also useful in the context of dynamic decision making. Data from two dynamic decision making tasks are used to demonstrate how participants' understanding compares to experts' understanding in the tasks, how it changes over time and how it correlates with task performance.
In the third paper, Improving operations management by synthesizing participant knowledge and system data, Etienne Rouwette and Jack Vennix explored the utility of group model building to combine client involvement and identification of system improvements in enterprise resource planning systems. Group model building in an enterprise resource planning setting seemed to result in positive effects such as learning, improved communication, consensus and commitment. On the basis of three cases, the paper identified five points in which the generic group model building approach needs to be adapted in the enterprise resource planning context.Birgit Kopainsky
Urban dynamics has always been a classical research topic for the System Dynamics Society. Although there was only one presenter at this session, a significant number of participants were attracted and debates lasted more than twenty minutes.
The paper, The Relevance of Urban Dynamics to Singapore's Success Story: Lessons for Moving beyond the Crisis, was coauthored by John Richardson and Elizabeth Ong Ling Lee from the National University of Singapore. Elizabeth presented the paper and was joined by John later during the Q&A period. They reviewed the development policies that had secured the well-being of Singapore's people since independence and then compared the lessons of Urban Dynamics. Singapore successfully overcame the water and land shortage problems because the country is sensitive to the need for resiliency in the face of challenges. They concluded that the urban dynamics model would be a valuable tool to urban decision makers to move beyond the Crisis.
A few critical questions were raised. The urban dynamics model has been developed within the regional context. Labor and capital are free to move. These assumptions do not hold true for a city state like Singapore. Therefore, the applicability of the urban dynamics model to Singapore is questionable. In addition, the participants urged the presenters to build an urban dynamics model for Singapore to examine the arguments they presented.Xu Honggang
System dynamics has a long tradition of being used in the analysis and communication of climate change issues. Climate change is also closely related to sustainability and both terms are, at times, used synonymously. This parallel session includes three interesting presentations on climate change and sustainability. The presentations covered specific climate change and sustainability issues in New Zealand and Latvia and a generic simulation model of carbon circulation and methane feedbacks. The approaches that were used involved qualitative group modeling, stock and flow simulation, and broad qualitative stock and flow maps to guide national sustainability development. The presenters shared useful insights on climate and adaptation and demonstrated how system dynamics is used in the subject area.
Full ReportTopics and presenters:
The presentation by Cavana and Adams focused on the Emissions Trading Scheme (ETS) of New Zealand. The research question was: "How will carbon emissions trading effect New Zealand forestry?" Several stakeholders participated in a group modeling exercise. Based on the research question, a wide range of issues were identified and then clustered using hexagonal post-its. The key variables were then identified and mapped onto a causal loop diagram. The authors identified that the ETS is reliant on economic analyses and its success is dependent on public perception of economic success.
Akpinar and Saysel took a hard system dynamics approach in their study. They developed a system dynamics model in STELLA that integrates several components of the climate system. The model includes the carbon cycle, impact on heat radiation and temperature feedback. The model also incorporated the feedback between permafrost melting and emission of methane. The model is freely available on the conference proceeding website.
Yang Miang Goh
Chaired by Valerie Gacogne
The session was fairly well attended. It included three interesting theoretical and experimental applications of control theory to different types of supply chains. A summary of the papers is given below in the order of presentation.
Supplier Capacity Decisions Under Retailer Competition and Delays: Theoretical and Experimental Results, by Paulo Goncalves, Santiago Arango Aramburo
The focus of this paper was to find optimal capacity investment decisions when a single supplier is faced by a sudden increase in demand due to retailer competition. The authors compared the results of different decision heuristics obtained from system dynamics (SD) simulation and from experiments (performed on a management flight) with an optimal cost function obtained analytically. The results showed that the simple rule of the SD model and the decisions taken by participants led to over investment and poor supplier performance. Excessive costs originated mainly from change in capacity or from maintaining backlog depending on whether order inflation is low or high respectively.
Understanding Supply Chain Replenishment Decisions, by Rogelio Oliva, Luis Herrero, Santiago Kraizelburd, Noel Watson
Chaired by John Richardson
In this session, the first speaker was Tony Phuah. He presented a single author paper titled, Can people learn behaviours of stock and flow using their ability to calculate running total? An experimental study. Tony Phuah's paper won this year's "Barry Richmond Scholarship Award." Congratulations Tony! The second presenter was Heather Skaza. She presented a co-authored paper named, Assessing the effect of systems simulations on systems understanding in undergraduate environmental science courses. The last presenter, Ignacio Martinez-Moyano, presented a co-authored paper called, Stock and Flow Failure: Initial Stock and Priming.
Tony Phuah's work focused on the understanding of stock and flows. He tested the effects of two treatments; "graphical integration guidance" and "running total and reflection." According to the experimental results the "running total and reflection" group did not show statistically significant improvement. This result was not expected. However, the "graphical integration guidance" group improved significantly after the intervention. Nice work! Thanks Tony.
Heather Skaza's presentation focused on the effect of simulations on understanding and learning. Heather and her colleague divided undergraduate students into two groups in an environmental science course. In the pilot study, both groups were identically treated expect for the intervention; the experimental group received simulations in completing three of their assignments and the control group received text descriptions of the same assignments. Results were not conclusive; initially, the experimental group showed a greater improvement, but later the control group managed to improve their performances and catch up with the experimental group. I personally believe this is partly due to the same debriefing that both groups received at the end of each task. I told Heather that the need to be fair to the students made it hard for them to effectively compare the two treatment groups. This is a very important and challenging study. We wish success to Heather and her colleague in the continuation of their study.
Similar to Tony Phuah's presentation, Ignacio Martinez-Moyano's talk was also on understanding of stock and flows. In a well-defined experiment, Ignacio Martinez-Moyano and his colleague tested two hypotheses: "The existence of an initial stock value in a stock accumulation question will improve the performance of the participants" and "people will perform better with a simpler version of a stock accumulation question." According to the initial results of the study, providing an initial value seems promising. I believe this is due to the decreased cognitive load on the participants; although an initial condition is not necessary in answering the questions, once it is provided, there is no need for people to try to judge if it is important or not. Important work! Ignacio, thanks for the presentation and good luck with the continuation of the study.
Chaired by Martin Schaffernicht
This meeting has been held since 2005. As in previous years, the attendees included members of the Policy Council. I wish to thank all of the participants for their time and for producing important arguments and suggestions. However, attention has to be given to the fact that most of the issues which have been discussed at this meeting were not new. They had been discussed at previous meetings. This indicates that the implementation of the suggestions has to be emphasized for the upcoming year.
This year's meeting dealt with the following topics:
There are about 600 reviewers. Thread chairs evaluate "their" reviewers' quality. The previous evaluations are taken into account in reviewer assignments to papers. However, a bad reviewing performance does not lead to deletion from the reviewers list: as has been argued before, the reviewing process is thought to help a reviewer be integrated into the community and become a better author.Spread of review quality
The quality of reviews is heterogeneous: some are very useful for the acceptation/rejection decisions; others plainly fail to understand the paper they review. Also, some are helpful for authors and others are useless. Too often, the review quality is too different. In order to maintain a balance, usually the mix of reviewers assigned to a paper is such that a concentration of poorly evaluated reviewers is avoided.
Feedback to reviewers
Currently, reviewers can receive feedback from the authors, but the thread-chair assessments are not fed back to reviewers. It has been suggested that this might be done to give them a signal for future reviews: since it has to be assumed that a reviewer would like to do a good job, this feedback would probably be helpful.
Help for reviewers
The website gives rather precise indications for reviewing; however, this has not prevented poor reviews. In order to go beyond feeding back these evaluations, it has been suggested that reviewers might be invited to a reviewing workshop.
Standardization of conference papers
Currently, there is some discussion concerning a possible standardization of conference papers. This is a delicate matter: how many different types of contributions should there be and how much can be standardized without becoming too restrictive? There are two possible ways to go about standardization: one can define a standard structure for each type of contribution, or one can define sets of attributes each kind of paper must/should have.
In this place, it may be worthwhile recalling some suggestions from the 2009 dialog meeting report (Schaffernicht and Groesser, 2009):
"For the main types of contributions–as of now, these are (1) model based papers and (2) methodology papers–specific guidelines should be defined and made available to authors and reviewers as a checklist for writing a paper and assessing the paper's quality.
1. Since problem-oriented, model-based work seems to be the main type of contribution at system dynamics conferences, papers written about completed research are expected to describe a series of topics (derived from Forrester's view on what can be achieved with an appropriate simulation model; see Forrester. 2007 System dynamics–the next 50 years, System Dynamics Review 23(2/3) 359–370) and other review guidelines of other conferences.
2. Papers about work in progress are expected to describe the topics that have been elaborated so far (according to the above list) and to make explicit up to which point the work can be considered as validated.
Last year's report had an exemplary web-form for this in its supplementary material.
Minimum requirements for papers dealing with unfinished work
It has also been suggested that the conference should allow for work-in-progress to be presented, in order to be inclusive and because for a considerable share of participants, the necessary funding depends on having an accepted paper. However, it has also been suggested that such papers should satisfy some minimum requirements.
Communicating recommendations to authors
The current website gives few instructions to authors, mainly formatting guidelines. It may be helpful to indicate what is expected from papers submitted to our conference. Thus, the same topics used to review can be suggested to authors:
1. Model-based work written about completed research is expected to satisfy the following criteria:
2. Papers about model-based work in progress are expected to describe the topics that have been elaborated so far (according to the above list) and to make explicit up to which point the work can be considered as validated.
Martin Schaffernicht and Andreas Groesser
John Sterman, Andrew Jones, Thomas Fiddaman, Elizabeth Sawin, Travis Frank, The Road From Copenhagen: Supporting International Climate Negotiations with the C-ROADS Simulation
When John Sterman, spoke at the 27th International Conference in Albuquerque, the Climate Interactive Group, which he leads, was preparing for the December 2009 Copenhagen Conference. The audience, mostly comprised of true believers, accepted climate change as a manifestation of global overshoot and collapse. They respected–perhaps even revered–John's work as the best system dynamics modeling has to offer. Most were aware of the C-ROADS Model. Some had used its findings as a basis for similar presentations, workshops or university courses. Yet John's passionate delivery brought to mind the prophet Jeremiah's vivid description of probable scenarios leading to the conquest of Israel and the privations its people would face. In retrospect, I believe John may have been using the Albuquerque venue as a warm-up for presentations he would be giving to less congenial audiences in the upcoming months.
John's post Copenhagen retrospective delivered in Seoul highlighted many climate change projections that the Albuquerque audience had heard. (Many participants in the Seoul conference had not traveled to Albuquerque and were less familiar with C-ROADS.) Temperatures are rising more rapidly than had been projected. The impacts that are already visible will become more severe. Coastal cities will be threatened, the ecosystems that support present agricultural production systems will be altered and the trend of more frequent extreme weather events will proliferate. "Runaway changes" with unpredictable, irreversible effects are becoming more and more probable. "We are playing Russian Roulette with a revolver in which nineteen out of twenty chambers are loaded," John concluded this part of his talk. "The risks will be borne by our children."
But vital though this message was, it should not be the most important take-away for Seoul participants. The take-away should be about how difficult it is to enact any significant policy change, requiring transformation of deeply held attitudes where the status quo is reinforced by entrenched institutions, a limited grasp of system complexity, myopic time horizons and self-serving powerful actors. In 2009, John drew an analogy between reversing climate change and effecting meaningful civil rights policy. In 2010, his message was that effecting policies to reverse climate change is, for a variety of reasons, the far more daunting task.
Since the early days of the Club of Rome and the IIASA Global Modeling conferences, an aging generation of system dynamics modelers has grappled with the challenges of changing public attitudes and influencing policy makers regarding global-scale challenges. The work of the C-ROADS team demonstrates how much has been learned, by at least some members of our community. The work is grounded in high-quality scientific research. There is a serious commitment to dissemination and recognition that dissemination is fundamentally different enterprise than research, with its own imperatives, technologies and pitfalls. There is recognition that raising public consciousness and influencing policy makers, too, are different, though interdependent enterprises. The work is being carried forward by a talented, multifaceted team with adequate funding and a commitment to the long haul. There is much to be learned from their experience. We can help forward their important work with our approbation, our support and our prayers.
Andrew Ford, Greening the Economy with New Markets: Lessons from System Dynamics Simulation of Energy and Environmental Markets
That I have less to say about Andy Ford's fine paper and presentation should not be interpreted as a reflection on its quality. In fact, the two plenary presentations and the issues they raise are interrelated. Andy is grappling with climate change proposals under consideration at the national level. His paper focused on the CLEAR Act introduced by US Senators Maria Cantwell (of Washington) and Susan Collins (of Maine). "It calls for the imposition of a scientifically based cap on CO2 emissions. The cap would apply to the upstream companies that produce or import fossil fuels. The higher prices of fossil fuels would then work their way through the US energy system, sending improved signals to al business on the value of avoiding CO2 emissions. Legislation such as CLEAR, backed up my high quality modeling work such as Greening the Economy… describes, will necessarily play an important role in reversing the process of climate change, if it is to occur."
Greening the Economy... presenting a model Andy describes as "preliminary" is nonetheless an exemplar of how public-policy oriented system dynamics modeling should be executed and documented. It is the sort of paper I assign to my students with the injunction "if your final project paper looks like this, will have met my expectations." The model, of mid-level size, is presented with a stock-flow Vensim diagram that is easily understood. It illustrates one of the strengths of system dynamics modeling, the ability to seamlessly combine physical and financial stock-flow dynamics within a single structure. It provides the capacity to explore scenarios that fall within the CLEAR legislation's purview and those that extend beyond it. Model results show that the "cap and dividend" mechanism proposed by CLEAR shows promise, but that market volatility is a possibility under some scenarios, raising concerns that may need to be addressed in regulatory regimes that are implemented under the legislation. Further work at the University of Washington is envisioned.
Juxtaposing this model alongside C-ROADS raises the evocative question of how the team at WSU should set its priorities. The next milestone in their project seems clear: refinement of the model and publication of results in one or more peer reviewed journals. But what then? The experience of the Climate Interactive Group and others in our community points to the challenges a commitment to dissemination poses, and the resources it requires. In academic communities, there is little agreement that the sort of dissemination C-ROADS represents is an appropriate activity for university faculty members, especially those seeking tenure. Once this fine model has moved beyond the preliminary stage and results have been published, Andy's team will face the challenge described so beautifully by Poet Robert Frost in "The Road not Taken."
The workshop on Thursday was lead by Billy Schoenberg of Forio Corporation. It was the aim of the session to show how comfortable it is to create a system dynamics simulation game played by two or more players together in an internet browser. Therefore Forio has made the effort to assure that the game runs synchronized for all players, i.e. the next simulation step does not start until every player has made his decision.
Billy showed every step necessary to get a multiplayer simulation up and running as a web service. The Forio home page heavily uses contemporary internet technologies like Ajax to make this process as convenient as possible by interactively supporting the user in every step. After creating an account for free, it is possible to import a system dynamics model created in any of the established modelers, e.g. Vensim. To begin with, Billy presented an already up and running simulation game on the Forio platform. With the example of a price war game, he demonstrated the functionalities that the users have within the game. There is a chat window, a dashboard showing the performance of the players, and the slider with which they can make their decision. After the players make their decision, the simulation proceeds to the next step.
For creating such a game it is as simple as choosing the button "create new multiplayer game". With the help of a text based editor you can adjust your model and simulation. Then you have to choose the roles and teams of the players involved. Currently it is still necessary to create a role for each single person playing the game. The role describes what the player will be able to do within the simulation (for example, if he or she is allowed to change the market price of a product and/or the amount produced by the company). Accordingly, in this example, the team defines for which company a player is making decisions. It is possible to give the teams different names. Thus the dashboard looks different for each role and each team. This has to be considered in the design process of the dashboard, for example by adding a field that shows the team name.
After a completed simulation game run, it is possible to print the results of the run and to analyze them. Even if there is no documentation of the service, the creation process looks so intuitive that it definitely can already be worked with the product. In case any question occurs, Billy Schoenberg encouraged all participants and interested persons not to hesitate to contact him. Overall, the session gave a very good insight into a highly interesting product that is at an early but mature development state.
Robert Eberlein gave a brief introduction about Vensim software in the beginning of this workshop. Vensim is used for developing, analyzing, and packaging high quality dynamic feedback models. Models are constructed graphically or in a text editor. Features include dynamic functions, subscripting (arrays), Monte Carlo sensitivity analysis, optimization, data handling, application interfaces, and much more. This one and one half hour workshop was basically organized for the beginners, who wanted to get know how to use this software. In this workshop Robert Eberlein provided software to all the participants to install on their computers. The basic idea was to go through the basics of building and analyzing a feedback model using Vensim. The class was centered on the development of a simple variation of Jay Forrester's classic market growth model. Participants were encouraged to follow along on their own computers. The workshop was done using the Vensim PLE (Personal Learning Edition) version. Vensim PLE is software that gets you started in system dynamics modeling and is free for educational use and inexpensive for commercial use. Vensim PLE is ideal for classroom use and personal learning of system dynamics. Participants took a lot of interest in learning this software. It was a very interactive session with Robert.
Muhammad Aman Ullah
The second session on Vensim software focused on an optimization problem. This one and one half hour workshop introduced the process of optimizing dynamics feedback models using Vensim. Working with a simple oil field depletion model, the process of optimizing in order to maximize financial performance was demonstrated. Participants were encouraged to follow the workshop on their own computers. During the workshop a simple calibration problem was presented and the potentials and limitations of calibration using data were briefly discussed.
Muhammad Aman Ullah
The System Dynamics Newsletter is published four times a year by the System Dynamics Society.