| Complexity
and Organizational Structure by Emily F. Breuner |
|
|
Chapter
6 |
|
The test of a first-rate intelligence is the ability to hold two opposed ideas in mind at the same time and still retain the ability to function. - F. Scott Fitzgerald
|
|
| If there is one generalization of decentralized organizational structure that can be drawn from a comparison of these two organization, it is this: on several dimensions, it is the interaction of opposite forces that makes them resilient, robust, and successful. Just as a positive and a negative charge keep a molecule together, opposite forces seem to be the stabilizing forces in both the Internet and Visa International. Competitive forces are used to manage cooperative inefficiencies, while cooperative activities are used to build more effective but less efficient solutions. Hierarchical architecture is used to make the organization more flat, and those who earn the most respect have the least incentive to force a solution. It is the "yin and yang" that makes these organizations work, not the linear, bi-modal thinking that has shaped organizations in the past. I have found two theoretical systems especially applicable in making sense of these apparently paradoxical organization structures: complexity theory and Weick's theories on the social psychology of organizing. I briefly summarize these theories below and go on to analyze both the Internet and Visa International in light of them. |  |
| Important Analytical Lenses There are many fascinating aspects of both the Internet and Visa that suggest that they can best be viewed as biological analogs, or biologs. Indeed it seems the world will never be without an Internet or a Visa card again; they have literally taken on lives of their own. In describing his vision of Visa, Hock often uses biological references to the brain and other organisms. He also refers to this type of organization as a "Chaord," a combination of the maximum chaos possible with the minimum amount of order.(1) These concepts are congruent with the emerging science of complexity theory, which appears to be a useful theoretical framework for understanding why and how these two organizations have so successfully organized to create platforms that have effected fundamental change around the world, especially when complemented by Weick's theory on the social psychology of organizing. |
|
| Complexity The emerging field of complexity theory deals with the relationship between chaos and order in complex, adaptive systems. By nature an interdisciplinary field, it encompasses biology, physics, psychology, economics and a wide range of other fields. Some have called it the Theory of Everything, and at the very least it is the search for the Theory of Everything, a single unifying rule that explains (but will probably not predict) the behavior of all things. |
|
| Complexity has gained wide recognition, as some of the world's most famous Nobel laureates have validated this new science by founding or becoming involved in the Santa Fe Institute. A non-profit, multi-disciplinary research center in New Mexico, the Institute is meant to be a forum for scientists to exchange ideas from their areas of expertise in order to facilitate the development of complexity theory. | |
| Complexity theory is based on the belief that it is the interaction of the behavior of many individuals or "agents" that causes dynamic change in any system, and the sheer number of interactions means that the evolution of any system is unpredictable. This idea was born when Henri Poincare realized that if a system consisted of even a few parts that interacted strongly, it could exhibit unpredictable behavior. "For example if three planets orbit around one another influenced only by the force of gravity, it is impossible to predict the motions for a long period of time, even if the positions and velocities of the planets are know with great accuracy."(2) This strikes at the very heart of all the beliefs we have established as a result of centuries of linear, Newtonian, scientific thought. Not only are we not able to control the world around us, but we are not even able to use the laws we understand to predict it! | |
| "Investigators began to wonder if they could apply the emerging theory of complexity to adaptive systems...Adaptive systems like other complex systems, consist of many relatively independent parts that are highly interconnected and interactive...An adaptive agent can be anything from a single cell organism to human society; it must be capable of forming and changing strategies. For instance, an aspen will orient its leaves to capture the most sunlight, a strategy that is encoded in its genes, and can change through mutation or recombination of genetic material. Complex systems can learn and adapt to changing conditions in the environment with which they interface."(3) | |
|
The following simple example illustrates a complex adaptive system:
|
|
Other important aspects of complex adaptive systems:
|
|
| Of course, in the simple example above, the frog's adaptation changes the environment in ways that affect other agents apart from the bird, such as the algae and the insects. These agents also adapt, and soon the number of possible adaptations and their interactions becomes infinite, and thus the evolution of the system becomes completely unpredictable and chaotic. Yet the world is not chaotic and systems seem to stabilize. How this order evolves from chaos in adaptive systems is the subject of complexity theory. | |
| A key insight of this theory is that the loose relationship between many parts of the system that creates stability. Agents that are too dependent on one part of the system have a greater chance of failure, while agents that are slightly dependent on many parts of the system are insulated from the risk that one change in the environment will cause extinction. Thus, a network of dependence causes stability. | |
| While there is chaos, systems of all types have a propensity to reach local equilibrium or order before evolving in some new direction. Our instinct until recently has been to analyze a part of a system in isolation to look at the forces that drive it to equilibrium, which is a very linear and Newtonian perspective. This is exemplified by the economists' standard statement at the end of every proof---"all things being equal." | |
| What does complexity teach us? Many systems can, in fact, be thought of as complex adaptive systems. The key drivers of such systems are the cooperative and competitive strategies of the agents within the system, and because all these strategies are interacting in unpredictable and uncontrollable ways, a system's behavior and evolution are also unpredictable and uncontrollable. Thus, management of such a system should try to capitalize on these forces in ways that are consistent with the system, rather than trying to control them. These ideas obviously have implications for both the Internet and Visa which seem easily characterized as complex adaptive systems. | |
| Subassemblies and Cycles In his book The Social Psychology of Organizing, Weick explores the psychology behind the development of organizations. The book is rich with important information, but two of its concepts are especially applicable to understanding the Internet and Visa, especially when taken in conjunction with complexity theory (which Weick seems to have anticipated in some ways). |
|
| Subassemblies (or Agents) Weick(4) tells the parable of Hora and Tempus, two watchmakers that were interrupted frequently in their work by people calling to order watches. Hora was able to survive and prosper while Tempus soon went out of business. Why? |
|
| The idea relates to complexity theory's concept of agents in that it involves the interactions of many sub-units. The watches made by each man contained 1000 parts. The watches Tempus made were constructed so that if one of them was only partly assembled and he had to put it down, it fell apart and had to be reassembled. Thus, if Tempus got a phone call before a watch was finished, he had to start all over. | |
| Hora's watches were just as complex as Tempus's, but they had been designed so that they could be put together using subassemblies of ten parts each. Ten of these subassemblies in turn could be assembled into a large subassembly and ten of the larger subassemblies could be assembled in the 1000-piece watch. Whenever Hora was interrupted, then, he lost only a small portion of his work. | |
| The parable's crucial implication for our concern with organizing is its suggestion that the time required to create a large organization from simple components depends critically on the number and distributions of the potential intermediate stable forms. This is a vast simplification but it does give us a tool to use in comprehending complexity...The notion of stable subassemblies implies that when a set of subassemblies are aggregated, there will be strong ties within the subassembly and weaker linkages between subassemblies. The bond of most subsystems, in most organizations, should be relatively loose. This means that both stability and adaptation are achieved with less interdependence, less consensus, less mutual responsiveness than we usually assume. | |
| ...The combination of stable subassemblies composed of double interacts and of loose coupling among double interacts is attractive when pondering organizations because it suggests the conditions under which evolution can occur quite rapidly, adaptation can be preserved and adaptability can also be maintained. Subassemblies and loose coupling provide the potential for flexibility as well as stability. (p 110-112) | |
| Rules for Assembling Interacts (Dependencies) Weick then goes on to describe rules for assembling double interacts which are defined as follows: Processes contain individual behaviors that are interlocked among two or more people. The behaviors of one person are contingent on the behaviors of another person(s), and these contingencies are called interacts. The unit of analysis in organizing is contingent on response patterns, patterns in which an action by actor A evokes a specific response in actor B (...this is an interact), which is then responded to by actor A (this complete sequence is a double interact).(5) |
|
| It easily follows then that by constructing these double interacts, which in coordination theory would be labeled dependencies, Weick is suggesting rules for developing coordination processes among agents, or what he calls subassemblies. He goes on to suggest that the way this coordination is structured depends heavily on the characteristics of the information being processed by the subassemblies: | |
| (In assembling double interacts into subassemblies and processes,) we presume that the actors use the following meta-rule (a rule about how to choose rules). The greater the perceived equivocality present in the input, the fewer the number of rules used to assemble the process...However if the input is judged to be less equivocal, there is more certainty as to what the item is and how it should be handled; hence a greater number of rules can applied in assembling a process to deal with this input.(6) | |
| I have interpreted this passage as saying that when the types of problems coming into the system are fairly well structured, many rules can be codified that will handily deal with them with a process needing relatively few steps or cycles (i.e., time). However, when there is uncertainty about what kinds of problems an organization is going to deal with, few rules can be specified, but many steps will be invoked as means of defining the problem space. (7) | |
| Both complexity theory and Weick's concepts of subassemblies and the equivocality of information are important in analyzing Visa and the Internet and understanding the success of their respective governance structures. Each of these perspectives is applied below to several aspects of that success. | |
| Agents of the Complex Adaptive Systems that are the Internet and Visa In order to use these bodies of theory to look at the Internet and Visa, one must identify the "agents" or "sub-assemblies," not an easy task with respect to the Internet. In one sense, the agent is the end-user, since it is the interaction of all end-users that most effects what goes on in the Internet. However, at this point in time, end-users' ability to gain access to the Internet is determined by the cooperative actions of the organizations that supply network access. With respect to Visa, the card holders and merchants also are important features, but the structure is explicitly designed with the view that the member bank is the natural unit of stability. Thus the following analysis is done with the Internet access providers and the member banks as the agents. |
|
| Diversity It is important to note that neither of these agents are highly dependent on the other agents in their respective systems. Each is autonomous and free to develop its relationship with its customers independent of the other agents in the system, and few have tied their fates closely to others'. This provides some stability. It also means that as each develops in a certain direction, they do so without endangering the survival of the group as a whole. For example, if BayBank develops a Visa product that causes BayBank's collapse, it neither damages Visa or the other members. Conversely, the fact that these independent agents are working to innovate at the same time means a higher likelihood that one or more of the diverse solutions or adaptations to problems in their environment will be successful. Thus, diverse and divergent solutions create niche strategies and give organizations like Visa and the Internet a greater chance for survival as a whole. |
|
| Discontinuous vs. Incremental Change The other important aspect of this concurrent, diverse solution or adaptation development process is its tendency to create incremental rather than discontinuous change. For example, a centrally planned product development process defines a problem and then engineers a solution. Specifying in advance all the facets of a problem and then coordinating the development of the solution is efficient if the conditions that impact the problem are static. However, when you are trying to design solutions for an ever changing environment, and the problem definition and solution engineering take sufficient time that the problem itself has changed, and the engineered solution no longer solves the problem. |
| By relying on the diversity of agents to create emergent, incremental solutions, the organization can adjust along with the environment. Planned solutions lock an organization into a trajectory of change that may not fit the unpredictable evolution of the environment in which it "lives." An example of this is apparent in computer hardware. By investing in IBM's engineered solution for computing, that is their mainframe computing environment complete with proprietary software, companies locked themselves into IBM's vision of the future. Those that have invested in more open systems can adopt new software solutions that have been incrementally developed over time by many vendors in many niches. This is an oversimplification, but it does illustrate the difference between a "built" or a "grown" solution. | |
| It is true that "grown" solutions are not perfect or even elegant. Usually they are merely adequate. However, they tend to be applied more quickly than engineered solutions, and with application comes feedback. The feedback serves to refine the problem more and thus enable the production further incremental solutions. Each new "release" solves the problem even more adequately and includes adjustment to new environmental conditions that have arisen in the interim. In this way, incremental solutions can be more effective, if not more efficient, because they allow adaptation to the unpredictable evolution of the environment in which they are applied. These ideas also have been applied to software development; prototypes of software systems are tested with users in order to gain feedback about the fitness of the solution, as well as to gather information about changes in the environment that have occurred since the last prototype. It is also clear that the Internet follows a similar strategy; each protocol has many releases, or versions, which solve problems that arise out the evolution of other technologies in the system. | |
| Both the Internet and Visa use decentralized forces to develop incremental improvement. Both are solution-driven(8) in that they do not try to engineer solutions, but rather choose among possible solutions developed by their respective "agents." The IETF uses the follows the maxim "we reject kings and concentrate on working code."(9) Similarly, Dee Hock tells of choosing "serviceable technology"(10) over elegant solutions. When deciding on which technology to use to encode information on the Visa card, many technologies superior to the magnetic stripe were available, yet not all countries had the electronic infrastructure in place to make use of them. Thus Visa decided to go with an incremental solution---magnetic strips---rather than embedded chips with the understanding that they would most likely switch to embedded chips at some point in the future. | |
| This decision seems inefficient given that embedded chip technology was already acknowledged as the long term solution. However, it was an effective solution, given that it encouraged the better data communication in the entire system in the short term, lowering transaction costs in the process. In addition, there was time to learn more about how data embedded on the card could be used in the future, and how the ability to do so might impact the desire to do so, changing the trajectory of the embedded chip's development. This topic is illustrated further in the next section where I discuss the use of cooperative and competitive forces. | |
Cooperation versus CompetitionThe most important forces working in opposition at Visa and the Internet are the forces of cooperation and competition. Both are set up to enable their respective agents to develop both competitive and cooperative strategies as they react to their environments. This ability is important because it allows for the entire entity to develop very effective solutions while at the same time preserving the agents' autonomy and incentives to compete with the same agents with which they cooperate. It is the way these forces work together, the way they both attract and repel the agents, that make the Internet and Visa so successful. These positive and negative forces push and pull the systems together, achieving Hock's desire for minimum order and maximum chaos. |
|
| Prisoner's Dilemma/Tragedy of the Commons One way of thinking about the cooperative and competitive forces at work in Visa and the Internet is to identify the Prisoner's Dilemma and the Tragedy of the Commons that face the agents in each organization: |
|
| Visa Member banks want to cooperate at Visa so that the network of cardholders and merchants is as large as possible, as is the potential for transactions and interest charges earned. However, each bank wants the other banks in its area to fail or be excluded from membership in the Visa card program so that it can have a local monopoly on Visa cards and therefore more cardholders and member institutions. Therefore, the banks have an incentive to compete, but each bank will also be better off if it cooperates. |
|
| Member banks can cause a tragedy of the commons by claiming Visa membership and thus getting the advantages of its name, but not abiding by the rules and regulations set forth in the By-Laws which guarantee interoperability and manage the overall brand image of Visa. | |
| Internet The Prisoner's Dilemma is similar for the Internet. Each network provider want to cooperate by exchanging bytes because the interoperability of the Internet and access to vast information sources via a singular network is what makes the Internet valuable. However, each wants to gain more of its competitors' subscribers by forcing competitors out of the market. |
|
| The tragedy of the commons is that there is a scarce resource in the form of bandwidth. If a network service provider consumes excessive bandwidth on behalf of its customers, then everyone is worse off. There must be joint investment and stewardship to make sure that bandwidth is increased, used more effectively, and/or equitably allocated. | |
| These forces are played against each other in order to keep these organizations together, mainly in the areas of innovation and standard setting cooperative behavior. | |
InnovationWithin Visa, the member banks are granted the right to innovate around products and services that they offer their customers so long as they conform to the guidelines developed by Visa. Thus, each agent has the power to compete with every other agent, in other words, to develop a competitive strategy as a reaction to its environment. For example, banks teamed up with non-banks to create award programs like the American Airlines bonus Mileage cards, which give Citibank an edge over all other member banks. At the same time, each bank must develop its competitive strategy within the guidelines adopted by the Visa organization as the result of a cooperative strategy. As a result, product develop is not coordinated within Visa, but rather the market forces outside Visa, and is a direct analog of agents adopting diverse adaptation in a complex adaptive system. It is a chaotic, haphazard process that is "controlled" only by market feedback and the requirement that the new product interoperate with other products in the system according to standards cooperatively set by other agents---maximum chaos with minimum order. |
|
| Conversely, cooperation is used to make product development more robust than it is in nature. Part of the Visa International staff is focused on product management and supported by the funds contributed by all members of the organization. These product managers are concerned with brand-wide product implementation, but they are also sensors to the competitive environment facing Visa as a whole. They can react quickly to threats from MasterCard and American Express, ensuring that the Visa brand as a whole is successful. When American Express offered travel insurance, for example, product managers at Visa could quickly put together a competitive service after it was approved by the regional and International Boards. If the responsibility for product development had been exclusively at the member bank level, the time needed for a solution to emerge might have been dangerous to the Visa brand. | |
| There is product management at the regional and national level as well; here product management watches local products developed to meet unique environmental requirements. For example, certain cultures disapprove of credit, and as a result, debit cards were developed and managed in these regional or national organizations. Thus at each level of the Visa hierarchy, cooperative and competitive forces coordinate product development, and thus simultaneously enable chaotic evolutionary development, efficient solutions to local environments, and orderly cooperative reaction to Visa's competition. | |
| Similarly, Internet service providers have the ability to innovate around products and services they offer their customers. Many service providers are now developing security services to differentiate between themselves and their competitors. Some are developing software that makes using the Internet easier. While their relationship with their customer is unregulated, the information that they put out on the net conforms to the standards set by the IAB. Again, there is chaotic competitive product development in the cooperative behavior that ensures interoperability. | |
| It has been suggested that control is not the power to direct innovation, but rather the power to stop it.(11) IBM and Xerox fund various R&D projects that develop diverse solutions but they rarely let the market give them feedback as to an innovation's worth. Instead, forces inside the company evaluate the worthiness of a product by deciding to take it to the market or to kill it. Countless times, IBM and Xerox have failed to market ideas (like the graphical user interface) on which others have built empires. It seems that Visa's product development strategy, while more inefficient than a centrally planned one in terms of resource allocation and economies of scale, more effectively extracts the value of innovation from the market. Similarly, the Internet lets the user community test the fitness of a new application on the Internet. By not controlling innovation, both the Internet and Visa demonstrate their distributed control. | |
| Centralized, Cooperative Functions The use of competition and cooperation to coordinate behavior is not only in evidence in product development but also in the centralized cooperative functions. These functions are subject to complacency and inefficiencies because there is no competition forcing these functional areas to innovate and produce high levels of service. Visa has imposed competitive forces on its cooperative, centralized functions to avoid this problem. While Visa International would like to operate the information technologies at all levels of the regional hierarchy, no member organization at any level is required to use Visa International's service or technology. They have the freedom to buy information systems and management in the marketplace. This both forces the design of decentralized information systems that enable freedom from Visa's central management, and provides Visa's internal organization with the market incentives that encourage innovation and high levels of service. |
|
| Currently there is no analog to this arrangement on the Internet since the there is no centralized organization providing centralized services. Instead, the top tier NSPs are actually in competition with each other, providing the same incentives to innovate and provide service. Should a central organization be formed as part of the commercialization of the Internet, these forces should be preserved. | |
Hierarchy in the Decentralized OrganizationAnother paradox about these successful organizations is that both use hierarchies to manage their decentralized structures. However, the hierarchies are not hierarchies of power, but rather aggregations of agents that actually serve reduce the amount of dependence and coordination between agents. The difference between these decentralized hierarchies and traditional hierarchies are based on the fact that they manage participation and complexity, rather than authority. These hierarchies are closely related to federalist ideas, as are the checks and balances that the hierarchies provide in eliminating undue influence and finding the most efficient consensus decisions. Indeed the hierarchical structure is a means of imposing order on chaos without destroying its benefits; the hierarchy itself is a large cooperative strategy, and is an extremely important feature of these organizations. |
|
| The Fractal Nature of Decentralized Hierarchies Because the goal of hierarchies in a decentralized organization is based on managing participation of many agents, they tend to be fractal(12) in nature. That is, the shape and format of each level of the hierarchy are similar, and each level of the hierarchy has a similar relationship to the level above it, as well as having a similar relationship to the one below it. Each also has similar rules governing its behavior and similar authority. There are explicit rules for interactions between organizations that help to coordinate activities (again, standardization serves as the method for managing the usability of information flowing between parts of the hierarchy). |
|
| For example, each level of the geographic hierarchy at Visa has similar powers, and their activities are largely identical. They are responsible for the budget and membership rules of their level of organization. Each has some authority over the group below, and grants some authority to the group above. Visa International in turn grants authority to the lowest level by submitting to the "rule" of a Board made up of member banks. The hierarchy on the Internet is also fractal as the Internet Access Providers look just like the Network Service Providers. In both cases the interactions are standardized. At Visa it is the Board process that forces standardization; on the Internet, the IETF actually standardizes the interaction between the layers, both technologically and in the standards setting process. | |
|
This fractal structure serves several purposes: (i) it reduces risks by insulating other subassemblies or agents from the failure of any other one, |
|
| In essence it replicates the ability for the benefits of interacting agents to exist at each level of the hierarchy, i.e., regional Visa organizations interact like agents with their peers, as do peer NSPs. It also has a tendency to replicate itself when adjacent organizations are formed. For example, the New England Area Research Network (NEARNet) was established in order to put together a network for the New England area to take advantage of all the Internet had to provide.(13) Not surprisingly, it was formed with many of the principles of the IETF itself, and included committees for standard setting, etc. | |
| Efficiency and Effectiveness The basis for consensus decisions at both Visa and within the Internet is whether or not the given solution will increase interoperability, and therefore the potential of the overall network, making everyone better off. However, consensus usually means that a solution satisfies the needs of the lowest common denominator and is suboptimal on some level for everyone. The hierarchical structure mitigates this effect by building in local consensus points, pushing consensus decisions down to the lowest level possible. As a problem arises, it works its way up through the hierarchy. If it is a problem that can be handled at the national level in Visa, then the consensus is reached, providing a much better solution since the lowest common denominator within a nation is probably "higher" and less suboptimal than that at an interregional level. |
|
| For example, even though magnetic stripe technology was chosen by Visa International, Visa France wanted to implement embedded chip technology because France had the infrastructure to support it. As long as it included a magnetic strip, France was able to add further functionality. By reaching consensus to use the embedded chip within its own country, Visa France achieved a local consensus that was more efficient in its local environment than was the global consensus. In other words, the participation of agents is gathered and coordinated at the appropriate level, diminishing some of the inefficient effects of cooperative behavior and increasing the overall effectiveness at the same time. | |
| Cycles The layers of hierarchy serve as "cycles" in Weick's framework. Since Visa is a conglomeration of many diverse parts subject to some common but often different rules of behavior, there is a high likelihood that the information coming into the governing structure will be highly equivocal. Weick predicts that there will be a small number of rules around processing this information, but a large number of process cycles. In fact, Visa has precisely such a structure. The By-Laws are explicit about how to process the information, which is done by a series of many board decision processes, or in Weick's words, cycles. Hock himself says the process "allowed organized information about problems to emerge."(14) |
|
| Similarly, the decision making process on the Internet involves the RFC cycle so that the issues become clearer to all decision makers involved. The comments of Internet users serve to refine the information with their diverse and varied insights, as reflected in Kapor's comment about "fresh air" in the system. As with Visa, the IETF standard setting hierarchy does not aggregate authority, but in the case it does aggregate clarity and unequivocal information. | |
| Conversely, in both organizations, the structure and nature of the operational information that is passed between agents (i.e., packets of digital information) is highly unequivocal and thus there are precise rules about how and when it is passed, just as Weick predicts. This precision enables the interoperability that both organizations need to be successful. | |
| Influence The striking thing about Visa and the Internet are the way their structures reduce the ability of large stakeholders to unduly influence the system. Why isn't Citibank "running the show" at Visa? As it largest card issuer, one would think it would be in a position of power. Similarly, MCI and Sprint handle enormous amounts of Internet traffic, yet they have not had the power to enforce or introduce proprietary standards. This suggests that something about the structure of both these organizations limits influence. |
|
| Indeed, Visa's By-Laws are laid out such that there is fair representation. Voting rights are based on sales volume, but only to a certain point (see Chapter Five for specifics). With the Internet, the decision process is so based on meritocracy and there is so much participation that it would be difficult to force a proprietary standards that did not in fact best solve a particular problem. Thus the large players are kept in line by the power that is distributed to end-users. | |
| Hierarchies as Change Managers Both Visa and the Internet use tiers or hierarchies as a way to manage differing rates of change. That is as things become more complex, hierarchies are often used as a way to isolate and manage parts of the system that have different dynamics. In computing, we moved to a two-tiered client server model partially because the technology of the user interface is changing at different rate from the data storage technology; by separating the two functions into tiers, the functions gain the ability to change at different rates. Similarly, the core network traffic on the Internet is changing more slowly that the user interface or access is, making a tiered structure efficient. |
|
| Visa also uses the hierarchy to manage different regional concerns. For example, the banking industry is changing much more quickly in the U.S. than in Europe, and the regional structure allows that change to occur with much less impact than if Europe and the U.S. had to both adopt new procedures based on the regulatory change in the US. | |
Evolution and LearningKnowledge Transfer and LearningBoth Visa and the Internet have forms of knowledge transfer and stored learning that define adaptive systems. In Visa, the Board structure facilitates the spread of important competitive knowledge, as does the market system in which Visa products are placed. For example, a new Visa product is widely advertised, and all other agents learn of it through normal market mechanisms like trade journals and industry reports. Information about coordination activities is spread by the board process. |
|
| In the Internet, the Internet itself is the method of spreading knowledge. The whole RFC process is one that spreads information and learning, and once a standard is defined, its specifications are published on the Internet as a record for all those who need access to it. | |
| Another requirement of an adaptive system is that learning must be preserved. When Visa somehow adapts to better fit its environment, that may mean that the structure of the organization changes. For example, a new region is being created due to the changes in Eastern Europe. In a sense, the changes to the By-Laws serve to preserve learning the way the specification and publishing of standards does on the Internet. | |
| Changing the By-Laws at Visa serves another important function. One of the major tenets of Peter Senge's views on organizational learning is that structure influences behavior. He says, "when placed in the same system, people, however different, tend to produce similar results."(15) Therefore Visa's ability to change its structure literally serves to adapt and change behavior, enabling the organization to evolve along with its environment. One question that surfaces at this point is that organizing along geographic lines might not be the best format in the future given that geographic boundaries are becoming seemingly less important. It may well be that the banks would rather be organized on a different dimension. The By-Laws specifically state powers in terms of geography, but the By-Laws also have rules about how the By-Laws can be changed. It is theoretically possible the Visa could even evolve into an organization based on a completely different hierarchy if it were willing to change the By-Laws to reflect that new dimension along which the hierarchy would be organized. | |
| Since these learning mechanisms allow adaptation to occur, they also allow both the Internet and Visa to evolve along an unpredictable evolutionary path. | |
Federalism and Decentralized OrganizationsComplexity theory provides a powerful lens through which to view these two organizations, but there is a political lens which is just as appropriate a framework by which these two organizations can be understood. Federalism, which provides for autonomy and cooperation, is an equally powerful lens through which to examines these organizations. In fact Charles Handy, a London Business School professor who believes that federalism must be an important part of organizational design in the future, talks of subsidiarity, a directly analogous concept to both subassemblies and agents. Federalism, in so far as it uses hierarchy to manage participation, also seems to be a cooperative strategy adopted by "subsidiaries," "subassemblies," or "agents" collectively to manage competitive and cooperative behavior. |
|
| Federalism is based on shared authority between a central governing body and its constituent parts which give up some of their sovereignty in exchange for the benefits of cooperative behavior; the constituent parts agree to abide by certain principles and the role of the governing structure is simply to mediate in matters of dispute and to act as a steward of the common good. In Dee Hock's words, the role of the federal government is to manage the conditions of the system, but not necessarily the behavior. It is not hard to see how federalism relates directly to both Visa and the Internet. | |
| Purpose and Principles Federalism does not concern itself with procedures as much as it does with meta-rules, or rules about how to make rules. The U.S. Constitution is a prime example of this. It states the principles by which citizens agree to live together and gives information about how rules or laws will be made, but it does not in general specify the laws themselves. That is left to the legislators, while the fit between the laws and the Constitution is judged by the Supreme Court. The purpose of the government is to provide conditions under which states can cooperate and through which the interactions between states can be equitably and fairly defined. It tries to do so with the knowledge that a federal government cannot possibly predict all the issues that will arise, so it has procedures that help define and solve problems. |
|
| Interestingly Thomas Jefferson, one of the most famous Federalists, was also a firm believer in innovation. The father of the U.S. patent laws, Jefferson believed firmly that individuals or agents), should be able to innovate and retain the fruits of their innovations regardless of the fact that an innovation is a public good, and may be subject to the stewardship of the government. Both the Internet and Visa also allow their respective agents to retain whatever benefits they can capture as a result of their innovations. | |
| As in the U.S. Federal system, checks and balances are built into Visa in terms of the voting rights of members and the specification of who has the right to make decisions. Additionally the members themselves sit on a board at the top of the organization, performing somewhat the same role as the Supreme Court. In the Internet, the checks and balances thus far have been the high visibility of the issues being uncovered, but there is concern over how long that can and will continue to be sufficient. | |
| Shared Values One of the key aspects of a well planned federalist system is that values need to be well aligned. Indeed, most difficulties arise when different agents have different objectives. However, business might well be an easier environment in which to implement federalism because in general it is easier to define shared values around a certain goal---that is, shared values are less subject to emotional or moral inspection in the business world. While Federalism in government seems sensible, it is rarely easy to implement because of differing political beliefs. However, in business there is usually an overriding goal to which people have subscribed by virtue of choosing to work for the company trying to implement a Federalist structure. That is, if you accept employment or membership at Visa, you have implicitly accepted the stated purpose of "enabling the exchange of electronic value." Similarly, the community on the Internet has thus far been homogenous and well aligned in their values. |
|
| With respect to values, federalism does not try to define how those values are to be manifested in behavior. Dee Hock proffered the example that one could never write a manual on how not to steal, yet we can all agree that stealing is wrong. He also offered the insight that it is the simple rules that are easy to write down, while the intangible, unmeasurable things that often mean the success or failure of a company are not. Why try to codify these things? Hock feels that such effort would be better spent on providing a platform where good ideas can take shape and be executed without being impeded by rules and procedures. | |
| Self-Policing Behavior The other benefit of decentralized authority is that it causes self-policing behavior. When people have power and authority, they are unlikely to give it up. In general U.S. citizens will react violently when they feel their power is being usurped by the government. Witness the fight over abortion rights, and the violent behavior of those who feel their rights are being violated. |
|
| In Visa, member banks acted as policemen when early in Visa history, in order to facilitate the rapid acceptance of Visa cards by retailers, Dee Hock made a special agreement with J.C. Penney to accept Visa cards. Hock allowed J.C. Penney to connect directly into the Visa system without going through a bank. The member banks violently reacted to Hock's move because they knew that one of their rights had been violated: Visa had gone into competition with banks in a sense, cutting them out of the transaction fees from Penney's. The result was an explicit By-Law making such future agreements "illegal." | |
| Similarly, Internet users who share the value that invasive advertising has no place on the Internet, have literally kicked those who violate this tenet off the Net. Violators are either inundated with hate mail, or complaints are sent to their network provider until the provider is forced to discontinue service. | |
| Power to the End-user It is interesting that by leaving power, control and authority at the periphery of the Internet and Visa, the end-user has acquired the most power. For example, the purpose of the Internet is to exchange information, and the end-user has the ability to access vast amount of information and construct it however she wants. She can put together random bits of information to make a decision about what country to visit, or she can use it to formulate her own political views. She can also use it to broadcast those views to million of people at almost no cost, no matter how esoteric her ideas or how small and widely dispersed her audience might be. The point is that she decides what her "product" is by virtue of being a consumer and a producer and purchasing access to the Internet. Similarly, the proliferation of options in the credit card business means that cardholders can chose whichever card product gives them the options they want, whether award programs, low interest rates, or high credit limits. They may complain about getting so many card offers in the mail, but they nonetheless have great power in the market place. |
|
Other Common Elements of these Decentralized StructuresTechnology Mirrors Decentralized StructureAs previously mentioned, in both cases the technology underlying these organizations mirrors the decentralized nature of the governance structure. This is an important feature of these organizations. In Visa, the information systems are decentralized as a result of the regions' right to chose a supplier for these services. In the Internet the systems are decentralized so that they are invulnerable to sabotage or national disaster. This is another example of how structure determines behavior, but a full explanation of this phenomenon is beyond the scope of this thesis. See also the research of Kim Clark at the Hard Business School. I will however touch upon the use of information technology in coordinating large organizations in Chapter Eight. |
|
| Importance of Individuals In both organizations, strong individuals have been germane to the development of governing structures. Visa would not have happened has it not been for Dee Hock and his incredible ability to influence the key people involved in the process. He was involved in every decision in the beginning, and one could construe Visa as a dictatorship in those days rather than a decentralized organization. Nonetheless, his legacy is one of distributed control and authority. |
|
| On the Internet, the key technologists who built the network have remained committed to its openness and ubiquity. Their selflessness has contributed to its growth and interoperability and the importance of their contributions cannot be overestimated. Without these key people, it is hard to know whether these organizations would have evolved in quite the same way. | |
| Meritocracy The importance of the meritocratic nature of both organizations cannot be overlooked. The Internet gives people the ability to express themselves on any issue they want. Through that recorded and widely dispersed expression, people gain influence and respect. Those that are especially well respected will most likely eventually serve in some way on one of the governing entities of the Internet. This system works well because when it comes to reaching consensus, the people who reach the top are likely to be those that have good judgment, shared values, and the influence to build consensus, making the consensus process as efficient as possible. |
|
| In Visa, there might be more self-interest in play from each board member because, after all, he or she is representing the interests of a multi-billion dollar institution. However, the election process of the board also ensures that the people who reach the top are those that are most respected and who have shown their ability to act for the common good, and the decisions made by these people are likely to be the decisions that are in the best interest of the group as a whole. | |
| Motivations In both cases, the motivation for participating in the formation of these groups seems to have a great deal to do with the overall vision of each organization. People at Visa mentioned how excited they were to be working on an idea that they knew would change the world. They also mentioned that in the beginning that they took pay cuts to come to Visa because they believed in Dee Hock's vision and because it was all so new. Similarly, those involved with the development of the Internet over the past few years are excited by its potential, and leery of those who would make it into "just another radio or television." The sense that the purpose of both organizations was something that would have major impact on the world seems to be an important part of the process, whereas money seems to have been largely absent as a motivation in either organization. |
|
| Measuring Success Another interesting characteristic that both organizations share is that they lack well defined measurements of success. Visa measures the volume of money going through its system as a percentage of the total "personal consumption expenditures" and the number of transactions. However, because banks have the power and responsibility to sign up cardholders, the staff at Visa International really has no direct control over how much volume Visa does. What then is the criteria for Visa International or a Visa staff member having done a good job? Hock purposefully implemented an explicitly subjective review process at Visa that did not rely on any quantitative measurement; he sanctioned what he called a "godfather bonus" which a manager could give out to an employee at anytime for any reason. Hock said his goal was to make employees financially independent so that they would feel comfortable disagreeing with management any time they felt there was sufficient reason. He was not interested in the quantifiable aspects of their performance, but rather the quality of their work. |
|
| Interestingly, one of the first things that Ms. Rossi of Wells Fargo Bank mentioned as a frustration in having worked with Visa was that "they never measure anything!"(16) Scott Loftesness also mentioned how ethereal measurements of success are at Visa. Clearly not everyone is comfortable with the lack of measurement, and clearly not every function in Visa International is unmeasurable. Yet because its success is defined by interoperability and brand recognition, which are both difficult to measure, some degree of subjective judgment seems inherent in evaluating Visa. | |
| What is success on the Internet? Right now an individual's success is driven by he judgment he exhibits or the quality of the solution he devises. In many cases he gets no compensation for solutions, only the satisfaction of knowing he has somehow made the Internet better, and that his solution was better than others proposed. The success of the Internet overall is more based on interoperability, and quality of access, both of which are difficult to measure. However, when one has contributed to either, it is apparent. For example, Marc Andreesen, the developer of Mosaic, received no compensation for developing that tool. Nonetheless, the impact of his "solution" on the Internet's success is vast, and its graphical nature has made the Internet dramatically less arcane, improving the quality of access. His contribution will gain him the role of a steward in the future---when he has an opinion on a new standard, others will listen to him by virtue of past achievements. | |
| Perhaps the most telling thing is that both Visa and Internet are tremendously valuable entities, yet neither one has any value on the traditional market. Wall Street would not know how to value either if it tried. Both are simply a collection of hardware and consensus mechanisms which add far more than the sum of their parts, yet cannot be valued using any traditional measures. In distributing responsibility and power, Visa and the Internet have also distributed investment; one of the reasons Visa is hard to value is because the capital and labor that make it a success are distributed at the member bank level, and are not under the ownership or employ of Visa International, while the major investment in the Internet is in the end-users' hardware.(17) | |
| Accountability A very interesting aspect of both systems is the way they deal with accountability and responsibility, both legally and otherwise. By leaving the interface with the end-users solely at the bank level, Visa isolates a great deal of the customer service accountability there as well. For example, if a cardholder detects an error on her statement (which is produced by the issuing bank, not Visa), the cardholder expects the bank to straighten out the situation, and in fact the bank has the responsibility to do so. If the error is made within the Visa system, Visa is liable to the bank, but the cardholder is unaware of any restitution on Visa's part. There is an arbitration process within the By-Laws, but there is also explicit lack of indemnification on Visa International's part and an explicit delegation of legal responsibility. |
|
| One of the big issues on the Internet today is that there is a similar delegation of responsibility to the point that no one is in control at all. If Wells Fargo Bank initiates home banking on the Internet for example, and there is a serious error made in transition, Wells Fargo will be liable for the error in the customer's mind and probably in the customer's account. However, Wells Fargo has no control over transmission over the Internet, nor does it even have the ability to trace where and how the error may have occurred because the Internet is designed to make routing transparent and dynamic. Until the Internet became the object of commercial attention, accountability was never an issue. Now that there is more at stake, reliability and accountability are key risk factors for those wanting to use the Internet to exchange value as well as information. How issues of accountability will effect the Internet remains to be seen, but since there is no mechanism to build trust, and ensure reliability, this lack of accountability will be a major disincentive to further commercial development, and seems to be creating pressures to centralize accountability and hence control of the Internet. | |
| Level of Analysis Finally, the fractal nature of the design of these organizations seems to obscure the issue of the level of analysis. As stated in the beginning of this chapter, it is hard to know just how to define which level in each organization should appropriately be labeled an agent. After all, it is the individual users on the Internet that come up with diverse software solutions, not the network service providers. |
|
| Similarly, one hesitates to omit the end-users of Visa's system as agents as well since their actions impact Visa's behavior. The opposite problem is also in effect. Visa is an organization whose product is coordinating many agents. However, it is itself an agent when seen in the context of the whole credit card industry. It competes with Mastercard and American Express, yet the phone devices that authorize charges at the merchant's site service all cards, indicating that Visa coordinates with its competitors at some level, just as it enables member banks to do. | |
| Finally, one can look at the underlying technology of both organizations as a metaphor for decentralization and coordination. I do not believe that it is a coincidence that both organizations have this ambiguity with respect to what level of analysis is appropriate, but rather it is a feature common to truly decentralized organizations. | |
Why did Visa and the Internet succeed where so many have not?Many organizations have tried to become more decentralized with varying results. What was it about Visa and the Internet that gave them a greater chance of success? |
|
| I believe that the issue of accountability comes into play most prominently here. In both cases, organizations were evolving for which there was no legal precedence. This meant that the organizations were free to evolve without the encumbrances of regulation or product liability law which make "clean sheet design" so difficult in many other industries. In fact, one Visa employee talked about how exciting it was to be at Visa because they were breaking new jurisprudence ground everyday.(18) | |
| The other key factor is that, as previously mentioned, the information technology underlying the organizations has been a powerful influence. Both organizations were involved in developing decentralized logical networks built to enable many-to-many communication. With that type of goal at the heart of the organization, it may well be easier to remain decentralized---again, structure influences behavior. | |
| There were other factors that made these two organizations prevail against our tendency toward centralized command and control. To say that it was simply one or two factors that led to their success would be antithetical to what complexity theory espouses. However, both these features had a significant impact on their successful decentralization. | |
Visa and Internet: Not an IdentityWhile there are startling similarities between these two organizations, there are some salient differences as well. Visa has a mechanism in place that manages the issues around interoperability as well as the competitive forces at work within Visa. The Internet has only a body that manages interoperability, the IETF. Up until now, the Internet has been entirely emergent: it was a "pull" system in that there was no one actively pushing solutions via the Internet for economic gain, but instead solutions were "pulled" into existence when problems arose. There was little money to be made on the Internet, and thus the only activities that needed to be managed were cooperative in nature. |
|
| Whereas the Internet was a pull system, Visa has always been a "push" system. At Visa there are forces that cause competitors to push new solutions on to the market in order to gain more customers, usually from other competing member institutions. Thus Visa always has had the need for a governing structure that monitored competition. After comparing these two organizations, I believe that the Internet soon will have to develop its own central structure to coordinate the competitive behavior of network service providers and other levels of agents and to coordinate the investment necessary to avoid the tragedy of the Internet's commons: bandwidth. Already there has been fragmentation on this dimension, as just this month the National Science Foundation has declared that the Internet has become so crowded and slow that the NSF will pay for a special high speed TCP/IP network dedicated to the supercomputers at all NSF research locations. This special high speed line will guarantee a certain speed of service, but it is also an example of the forces pushing the Internet toward proprietary solutions. | |
| If a central group does not develop, the government will come in to regulate it just as it has with the broadcast media. The Internet has the choice---develop a coordinating body or have the government do it for them. The analysis of Visa seems to suggest that if the major players were thinking strategically, they would enlist the help of visionaries, and do it themselves before the government has its chance. | |
| A coordinated group of network service providers may not be enough. That group may well be a subgroup in a larger organization that coordinates between other stakeholders such as major content providers, telephony, and retailers, or the logical layers of the Internet's hierarchy. How that body might be designed is addressed in Chapter Eight below. | |
| < Ch. 5 Ch. 7 > | |
|
Footnotes (1) Dee Hock, September 27, 1995. (2) Russel Ruthen, "Adapting to Complexity," Scientific American, January 1993, 132. (3) Ibid. (4) Karl E. Weick, The Social Psychology of Organizing, Reading: Addison-Wesley Publishing Company, 1979: 110. (5) Weick, 89. (6) Weick, 114. (7) From a discussion with Charles Osborn, April 7, 1995. (8) D. C. Lynch and M. T. Rose (Eds.), Internet System Handbook, Reading, MA: Addison-Wesley Publishing Company, Inc., 1993. (9) John Rugo, Personal interview, March 21, 1995. (10) Hock, March 23, 1995. (11) Tom Flanagan, Sloan MOT Candidate, April 21, 1995. (12) The term fractal refers to things that are similar in shape and structure when aggregated and disaggregated. For example, a graph of the performance of a single stock over the course of a day has a similar appearance and structure to the graph of all stocks over the last 50 years. Similarly, a cell has organelles that serve the same function as the organs of which the cell might be part in a more complex animal. (13) John Rugo, March 21, 1995. (14) Dee W. Hock, "Out of Control and Into Order," seminar, MIT, September 27, 1995. (15) Peter Senge, The Fifth Discipline, Doubleday, New York, New York, 1990: 42. (16) Deborah Rossi, Personal interview, March 23, 1995. (17) Fred Luconi, May 5, 1995. (18) David Wagman, March 22, 1995. |
|