Assignment 1

Liebman, Jon (1976) “Some Simple-minded observation on the role of optimization in public systems decision making,” Interfaces 6(4) 102-108.

Summary

This study was dealing with the operation of tools that quantitatively aid the decision-making process in the public sector. The two problems that were discusses were the effectiveness of firefighting organizations, ranging from the alarms to location and dispatch of vehicles, and river basin quality management. A model for the river basin quality simulation commenced in 1964, where Rolf Deininger met the existing water quality standard at minimum total cost. This model and other related aged models had a clear-cut objective function, and constraints were obvious. However, modern optimization solutions have more complexity involved, mainly multiple objectivity and fuzzy constraints.

It was also found that public-sector decision-making modeling incases a higher degree of complexity, compared to private-sector. Public-sector issues are interconnected when it comes to side effects of taking actions, and these effects are in many cases not predictable. Furthermore, decisions and goals never reach a mutual agreement, and as a result, these public goals are often set by the Congress. It is argued that the lack of agreement of goal definitions is due to the absence of “understanding the issue” step in basic problem-solving. The issue can have multiple notations, depending on the diversity of observers (decision-makers). Furthermore, these different observations of the issue cannot be recognizable to a computer-programming language. The motion of conflicting definitions of goals should be made by humans, who are built with a certain morality and set of values. These differences between decision-makers in public-sector problems are not as evident in private-sector issues.

Most cases that are complex issues are due to an inability of decision-makers to set a range of goals, or even to set a boundary. Liebman suggests methodology of modeling techniques that should be used by both the decision-makers and the analysts:

1. One must ask themselves if technological optimization modeling should be used to develop alternatives, rather than choosing one.
2. The decision-maker(s) must directly be involved with the analyst’s model in order to gain a great deal of understanding of the complex interrelationships.
3. Each decision maker should have their own separate model, including goals, objectives, range of alternatives, etc, and should then optimize a solution.
4. If an extremely multifaceted model is being created by an analyst, then one should ask themselves if several simple models should be constructed in order for the public officials to better comprehend the issue.

Conclusively, Liebman discusses that analysts and decision-makers should know how to use optimization modeling, and in what fashion. He also argues whether or not optimization modeling should even be technologically used, and what tools to use for an optimal solution.

Discussion

This article is not supposed to be used as a set of standards for a modeler to follow, but should really be used in an informative sense. Liebman discusses the importance of misusing optimization modeling and the problems that are related to it. Multiple decision-makers and analysts deal with the realm of multiple clients and notational differences in modeling that sometimes the complexities are omitted and in many cases, the non-optimal solution will be reached.

The faults of this work deal were the two examples that were introduced at the beginning of the paper. The firefighting story was not discussed because the author wrote-off the topic as “not wicked”. He had never actually defined what wicked actually meant, although the definition seems obvious. He still did not go much into detail as to the exact reasons why it was not discussed in detail and why it was not a complex issue, versus the river basin quality model. This information would be important for a reader to comprehend if he has no experience in linear optimization modeling. Also, his four proposals for modelers to adhere to could have included examples from the river basin quality model for the reader to grasp the suggestions on another level.

If I were to build on this research, I would discuss this paper with example(s) in a depth similar to what I discussed above. Also, I would include a schematic of methodology that a modeler can go by, similar to the systems approach. Comprised in this model would be multiple questions that will ask the modeler about the quality of the model, which will be used for the modeler to stop and think through the steps that have been taken thus far.

_________________________________________________________________________________

Gradowczyk, Mario H., Jacovkis, Pablo M., Freisztav, Ana M., Roussel, Jean-Marc, and Tabak, Esteban G. (1990) “Water Resources Optimization-Simulation in Argentina,” European Journal of Operational Research 49, 247-253.

Summary

This paper describes the theoretical background of a multi-purpose integrated mathematical model to serve as an optimal design approach for the Rio Negro basin in Argentina. Included in this paper were models for hydrological systems, along with the technical, sociopolitical and economic systems involved with the basin. The global model included three different models: OPER1, OPER2, and OPER4. OPER1 was used as a preliminary design with linear programming techniques, OPER2 was a simulation model that tested the hydrological and economic performances, and OPER4 is a mixed integer programming model that optimized the investments and used a cost-benefit analysis during the development of the basin.

The reason that the Rio Negro basin model of the basin was separated into three different models was due to the lack of time and core memory of the computer used. Many constraints in the models ranged from seasonal variation of water supply to economic and political boundaries of the basin. The hydrological constraints of the models that were non-linear were applied an iterative linearized method that was proposed by Major and Lenton (1979). The Net Present Worth of the dam was the main factor that the optimized design was based on; annual incomes were from the energy sold and the firm power guaranteed, along with increased agricultural productivity.

It was found that the model was both accurate and feasible for the optimal design solution of the Rio Negro basin. The client’s engineers were very traditional, and were not exactly trustful of modern mathematical ‘non-continuous’ optimization methods. However, after being presented with these innovative linear programming techniques, the engineers were satisfied with the work and interested in the future trends of the utilization of similar modeling software for future studies.

Discussion

In the concluding paragraphs of the paper, the authors discuss their reasoning behind some of the steps taken in the modeling process. Since their desire for the modeling software was to be autonomous, they chose to create the linear programming software in-house, as opposed to using a bought software package. Also, throughout the study, the engineers and analysts continued to discuss methodology of the models, creating a strong comprehension of the program. According to the authors, this “allowed a very useful integration of ideas and contributed to avoiding the tradition suspiciousness of government agency staff vis-à-vis consultants.”

The faults of this work were due to the lack of technology performances and time constraints. They were only allowed to use the laboratory 6 hours out of each day, leaving simulation runs at a minimum. Furthermore, the computer used for the simulations had very little core memory, which seems like they might have not validated the work to the same extent as more privileged researchers do with similar work.

After reading this paper, it seems that an optimization simulation modeling of a single basin should always be autonomous. I would analyze the model on a detailed extent, and create a general schematic for users to follow if their desire was to create a global optimization-simulation model for an entire basin. However, I would strongly encourage modelers to start from scratch when creating this software. There are too many variations of politics, economics, physical traits of the basin, etc. involved, and complications will arise when re-using a model for an entirely different basin.