| Complexity
and Organizational Structure by Emily F. Breuner |
|
|
Chapter
8 |
|
The Application of Information Technology to CoordinationThird Order EffectOne of the goals of coordination science is to examine the way in which information technology can impact coordination within organizations. Looking at how information technology has been used within Visa, we can see that indeed Malone and Rockart's third order effect has been realized. Neither entity would be possible were it not for the astounding advances in information technology. However, they have not used information technology to more tightly coordinate different organizations, but rather they have used information technology to loosely coordinate many organizations; both entities are designed such that the amount of coordination between subassemblies in the structure is minimal, but the number of subassemblies that are able to coordinate with each other is astounding. |
|
| What are the implications here for the design of collaborative applications? There is a tendency to try to design applications that will allow us to control more parameters of a system or systems. In complexity theory terms , however, this causes organizations to become "overconnected" in that they tie themselves too closely to the fates of other organizations, and spend precious time trying hopelessly to reconcile many changing parameters across the organizations. |  |
| In one sense, material requirements planning (MRP) applications are an example of this problem. MRP systems are elaborate systems in which all parts of a manufacturing process are modeled and then used to predict and manage the flow of materials through the manufacturing process. They require tremendous investment, but their value has been questioned because they are so difficult to maintain and there are so many variables to identify and track in the systems. Conversely the kanban "pull" system, in which each process sends a message only to the one directly before it when materials are needed, is incredibly simple and is based on small amounts of standard information passing between local, separate processes. Until recently we have seen only the second order effect of the use of IT in coordination, as IT has been used to automate and increase the amount of coordination information flowing in a system, as in MRP. | |
| On the other hand, Visa and the Internet seem to be displaying the third order effect of the use of IT in coordinating organizations. Rather than using IT to try to control more and more coordination parameters within these two organizations, they both use IT only to pass large amounts of transactional data between the "subassemblies" of the system, more like kanban systems. Nonetheless, they have succeeded in coordinating the behavior of thousands of institutions and millions of people. It may be that the third order effect predicted by Malone and Rockart is not simply more coordination-intensive organizations, but organizations that require a small amount of coordination on a scale never before possible. | |
| Collaborative Applications How can IT help disambiguate information and mitigate the "inefficient" consensus process? Collaborative applications that can speed up Weick's "cycles" that help clarify the issues might add significantly to better inter- and intra-organizational coordination. Video conferencing and electronic whiteboards will no doubt speed up the board cycle at Visa because it will cut down on travel and the need to have a set meeting schedule. Similarly, I assume that the six month to two year RFC process exists to make sure that every person affected by a proposed standard has a chance to "hear" about a standard under consideration and make comments. "Knowbots" or intelligent agents that can be programmed to alert people to issues of interest might also cut down on cycle time on the Internet, potentially shortening the RFC window. In these organizations where reduced transaction costs have already been realized, the best application of IT will be the one that gives the ability to manage this usability dependency, that is, make equivocal information usable. |
|
Designing the Decentralized OrganizationThe Internet and Visa International have both succeeded in distributing authority, responsibility, and control. I have analyzed the key characteristics that have enabled each to succeed. How can these ideas be applied to the design or redesign of other organizations? The following is by no means a recipe for success. However, the astonishing similarities between the Internet and Visa suggest that some of the following are indeed requirements of a successfully decentralized organizational structure. |
|
| Purpose It is important to define a broad purpose for which the organization exists in the first place. Most people would say that their organization has a mission statement and leave it at that. However, the purpose needs to be more abstract than most mission statements, leaving room for the organization to as innovations and adaptations move it in different directions. Most mission statements are too narrowly defined, and do not leave room for evolution. |
|
| Bottom-Up Design It is clear from both examples that the emphasis of each organization is on what Weick calls the subassembly, or what Handy calls subsidiarity. Neither organization was designed from the perspective of a central control point that delegates power and authority outward, but rather as separate autonomous units which are increasingly aggregated until there is a collection of cooperative activity at the top. Therefore, any decentralized organizational design effort should take a bottom-up approach in order to achieve success. |
|
| In particular, one should use Weick's ideas of subassemblies to try to find the naturally stable unit in the organization being designed, and then identify these subassemblies' "Prisoner's Dilemma." That is, careful conceptual analysis should be done to identify all the motivations for competing, and all the motivations for cooperating. Once the incentive to cooperate are defined, the subassemblies "Tragedy of the Commons" must similarly be identified. | |
| Once these things are understood, a set of operating principles should be created that both preserve beneficial competition while setting up a structure that facilitates cooperative behavior. The principles will evolve into rules which prevent the subassemblies from either acting in self-interest or preservation, or from taking advantage of scarce resources. | |
| The Role of Hierarchy As we have seen in both the Internet and Visa, successful decentralized organizations are not without hierarchy, and the next stage of design would be to design a hierarchy of subassemblies that foster fair competition and cooperation. The hierarchy serves to aggregate subassemblies so that they can cooperate at a mid-level, mitigating some of the inefficiencies of consensus decisions made at the highest level of the hierarchical structure. The hierarchy nonetheless should have fractal properties in that each level should mimic the structure of the ones above and below. This facilitates the development of solutions at a level that is closest to the problem and allows different, locally optimal solutions to be implemented across the same level of the hierarchy. This fractal structure is also important in reducing risk and increasing adaptability to subassemblies' and assemblies' unique contexts. |
|
| Another role of the hierarchy should be to enable enough of Weick's "cycles" so that problems and solutions under consideration by the organization are clarified to the point that consensus can be reached. The mechanisms for moving information up and down the hierarchy of assemblies and subassemblies should be designed to achieve this clarification explicitly, and meta-rules allowing the hierarchy to adjust to evolutionary adaptations must be defined as well. | |
| Managing Inefficiency Inefficiency is inherent in cooperative behavior because each subassembly gives up some optimal solution in favor of a common one. However, as Visa has done, competitive situations can be used to manage those inefficiencies, just as Visa manages its central IT organization by making its services optional for the subregions. Therefore, anywhere there are cooperative inefficiencies, competitive forces can be used to mitigate the drawbacks of consensual, cooperative behavior. |
|
| Innovation/Competition The responsibility for product, service, and/or process innovation should lie mainly at the subassembly level, but there should be opportunity and incentives at each assembly level as well (e.g., group members may license their product ideas to the rest of the Visa organization). The hierarchy should be structured to reduce the risk of chaotic innovation that might potentially hurt the organization as a whole. Visa is set up this way in that the failure of a product at the bank level may ruin the bank, but will leave Visa largely undamaged. In addition, this structure increases the chance of finding good solutions because of the number and diversity of subassemblies innovating along the same lines. In essence this structure uses market feedback to "coordinate" product development, and also avoids having a central organization that tries to "engineer" a solution instead of letting optimal solutions or innovations emerge at the local level. |
|
| Shared Resource Management/Coordination Conversely, with respects to processes that deal with coordinating the use of shared resources, market feedback has less of a role, and some form of peer review should be used to determine what actors, rules, and behavior best reflect the cooperative aspects of the organization. Federalism teaches us that the focus in the design should be to define the interfaces between cooperating entities, not the operations within them. |
|
| Infrastructure Another key lesson gleaned from the Internet and Visa is that the information infrastructure should mirror the structure of the organization. Designing a decentralized organization and then trying to support it with a centralized "nerve center" information architecture will undermine the success of the design. |
|
Running the Decentralized OrganizationStewardshipThe role of the people at the top of the aggregation of subassemblies is dramatically different than that of the people at the top of a traditional hierarchy. The people at the top are stewards, judging the merit of issues and solutions against the purpose and guiding principles of the organization. They do not create "vision," but rather exercise judgment in evaluating solutions that have emerged from various parts of the organization. They are managing the conditions through which growth can occur, not managing operations. Therefore, the people chosen for the top of the organization must have very different skills and motivations than have traditionally been ascribed to board members and executives, and only people with this level of understanding should be chosen to fill these roles. |
|
| Measurement As the organization evolves, so must the things measured as well as the way they are measured. Traditional "yardsticks" of success won't apply in a decentralized organization as we have seen in the Internet and Visa. Measurement might well be very subjective. However, people seem to need standards by which to compare themselves, so the results that are measured need to be chosen carefully so that the incentives are in line with the purpose of the organization. |
|
| Meritocracy It bears repeating that the meritocratic nature of the Internet and Visa are key components of the overall success of these organizations. Reaching the top of any part of the hierarchy requires good judgment and understanding of the mechanisms that make the organization work. It is important to support the self-policing, self-governing nature of the decentralized structure by having peer reviews in the form of "elected" and appointed executives and board members, whose behavior can easily be reviewed by peers. |
|
| Accountability Finally, a key intra- and inter-organizational force is accountability. Therefore, as the power and authority is distributed, so must the accountability and responsibility be. This applies to legal as well as moral responsibility. The implications of accountability for contracting issues and legal demands are enormous, and it is no accident that these two successfully decentralized organizations both came into being in areas where the law was largely non-existent. Decentralizing authority without also decentralizing accountability will lead to failure. |
|
|
|