When we develop a product, whom do we develop it for? How should we go about choosing the right design when different groups present conflicting requirements?
I got the chance to tackle this real-world challenge as part of a team chosen by NASA to create the architecture for a multi-decade program that would include exploration of the moon and Mars.
The team, composed of professionals from MIT and Draper Laboratories, had knowledgeable experts on space hardware, astrodynamics, and other aspects of space exploration. Yet, the question of how to select the “best” set of options remained elusive for two reasons: the project benefits were delivered to many constituencies and were not necessarily monetary.
I decided to study the problem, and the work became my SDM and aeronautics and astronautics thesis, which was awarded the SDM Best Thesis Prize in October 2007. The challenge was similar to a multi stakeholder problem, yet we did not have the advantage of clear representatives from each stakeholder group or clear need statements to solve, thus we were unable to use traditional multi-stakeholder analysis tools. Instead, we created a new framework, one that could perform a quantitative analysis of multi-stakeholder problems.
|Figure 1. The diagram presents the value feedback|
loop by which the Value Creating System supplies
some of the needs of stakeholder groups, and
those groups provide value back to the VCS.
We sought to formulate a model that could provide guidance in a broad range of circumstances where a system needs to satisfy multiple constituencies with nonaligned objectives. The model we prepared is based on three principles.
1) Any Value Creating System (VCS), including the specific NASA program, receives resources from its stakeholders, which receive value in return. While different stakeholders have different and sometimes conflicting needs, the more satisfied each stakeholder is the more resources that specific stakeholder is likely to supply. Since each of the alternative architectures satisfies different needs, the architecture selected by the VCS will affect the resources that the VCS receives through a feedback loop. Some architectures will increase the amount of resources provided to the VCS and thus allow it to grow faster.
|Figure 2. This mathematical formulation represents the|
value feedback loop. Vectors are used to represent the flow
of value elements and matrices to transform
one type of flow into another.
2) These loops can be represented as a tiered multi-attribute decision analysis model, with the VCS needs on the top. We can see that the flow is actually composed of many individual loops, passing through one specific resource, one specific stakeholder, and one specific need. Now, we can take advantage of the fact that VCS participates on every loop and group the needs that feed into the VCS. Then we can group the suppliers of each of these needs, simplifying the problem into several tiered multi-attribute decision analysis models. The top layer would be populated by the resources needed by the VCS, the second layer will include the stakeholders supplying those resources, the third will have the relationships between stakeholders, and the fourth will contain the needs of each stakeholder.
These multi-attribute functions can be combined and operated using standard matrix operations, and thus the tiered multi-attribute model can be evaluated by simply multiplying matrices.
3) The use of a stochastic model to propagate uncertainty. Since our model analyzed a social system, certainty was impossible. By using a stochastic model, we captured the information provided by the lack of agreement.
Based on these three principles, we assembled a model that worked as a feedback flow of vectorial quantities, using standard matrix math. The model produced two main results:
There is a tradeoff between growth and stability
By focusing on satisfying a few stakeholders that control the most critical resources, a VCS can maximize the flow of resources back to itself and achieve faster growth. This is what we called a “leadership” strategy, which alienates several stakeholder groups but rewards the ones with most power and provides a higher “value feedback” to the VCS.
Alternatively, the VCS can use strategies that keep a larger number of stakeholders satisfied, even if those strategies do not maximize the flow of resources back to the VCS and thus its growth. These are what we called “consensus” strategies, which we deemed more stable, since a shift in power among stakeholders affect in a lesser way the flow of resources to the VCS.
This tradeoff between growth and stability is in our understanding the basic diagram of political decision making. Leadership-oriented strategies aim for short-term results, and thus “bootstrap” the value creation process before a shift in stakeholder power happens, but risk failure. Consensus-oriented strategies look for stability through broad support, but risk not delivering results to any group, and thus not generating enough value to continue operating.
Uncertainty is valuable as information
In some instances, the model will not distinguish between alternatives. I believe that making the decision makers aware of the lack of information is in itself valuable, since it will signify the need for additional information or further study of the parameterization used to develop the model.
In conclusion, by using three simple principles, we formulated a mathematical construct that shows a tradeoff between short-term growth and long-term stability. Further interesting work could explore the accumulation of resources at the VCS and the indirect propagation of value through interaction between stakeholders.
Our work was supported by a NASA grant and received invaluable input from MIT faculty members, including Jeffrey Hoffman, Ed Crawley, and Pat Hale, as well as from fellow students from SDM and the MIT Department of Aeronautics and Astronautics.