The assumptions of the structural frame are reflected in current approaches to organizational design. These suppositions reflect a belief in rationality and a faith that a suitable array of formal roles and responsibilities will minimize distracting personal static and maximize people's performance on the job. Where the human resource approach (to be discussed in Chapters Six through Eight) emphasizes dealing with issues by changing people (through training, rotation, promotion, or dismissal), the structural perspective argues for putting people in the right roles and relationships. Properly designed, these formal arrangements can accommodate both collective goals and individual differences.

Six assumptions undergird the structural frame:

The structural view has two main intellectual roots. The first is the work of industrial analysts bent on designing organizations for maximum efficiency. The most prominent of these, Frederick W. Taylor (1911), was the father of time-and-motion studies; he founded an approach that he labeled "scientific management." Taylor broke tasks into minute parts and retrained workers to get the most from each motion and every second spent at work. Other theorists who contributed to the scientific management approach (Fayol, [1919] 1949; Urwick, 1937; Gulick and Urwick, 1937) developed principles focused on specialization, span of control, authority, and delegation of responsibility.

A second ancestor of structural ideas is the German economist and sociologist Max Weber. Weber wrote around the beginning of the twentieth century. At the time, formal organization was a relatively new phenomenon. Patriarchy, rather than rationality, was still the primary organizing principle. Patriarchal organizations were dominated by a father figure, a ruler with almost unlimited authority and boundless power. He could reward, punish, promote, or fire on personal whim. Seeing an evolution of new models in late-nineteenth-century Europe, Weber described "monocratic bureaucracy" as an ideal form that maximized norms of rationality. His model outlined several major features:

After World War II, Weber's work was rediscovered, inspiring a substantial body of theory and research. Blau and Scott (1962), Perrow (1986), Thompson (1967), Lawrence and Lorsch (1967), and Hall (1963), among others, amplified the bureaucratic model. Their work examined relationships among the elements of structure, looked closely at why organizations choose one structure over another, and analyzed the effects of structure on morale, productivity, and effectiveness.

How does structure influence what happens in the workplace? Essentially, it is a blueprint for officially sanctioned expectations and exchanges among internal players (executives, managers, employees) and external constituencies (such as customers and clients). Like an animal's skeleton or a building's framework, structural form both enhances and constrains what an organization can accomplish. The alternative design possibilities are virtually infinite, limited only by human preferences and capacity.

We often assume that people prefer structures with more choices and latitude (Leavitt, 1978). But this is not always the case. A study by Moeller (1968), for example, explored the effects of structure on teacher morale in two school systems. One was structured loosely and encouraged wide participation in decision making. The other was tightly controlled, with centralized authority and a clear chain of command. Moeller found the opposite of what he expected: faculty morale was higher in the district with a tighter structure.

United Parcel Service, "Big Brown," provides a current example of Moeller's finding. In the company's early days, UPS delivery employees were "scampering messenger boys" (Niemann, 2007). Since then, computer technology has replaced employee discretion, and every step from pickup to delivery is highly routinized. Detailed instructions specify where and in what order packages are to be placed on delivery trucks. Drivers follow computer-generated routes (which minimize mileage and left turns, to save time and gas). Newly scheduled pickups are automatically inserted into the nearest driver's route plan. If a driver sees you as he trots to your door, you'll get a friendly greeting. Look carefully and you'll notice the truck keys on the ring finger of the left hand. Given such a tight leash, you might expect demoralized employees. But the technology makes the job easier and enables drivers to be more productive. As one remarked with a smile, "We're happy robots."

Do these examples prove that a tighter structure is better? Not necessarily. Adler and Borys (1996) argue that the type of structure is as important as the amount or rigidity. There are good rules and bad ones. Formal structure enhances morale if it helps us get our work done. It has a negative impact if it gets in our way, buries us in red tape, or makes it too easy for management to control us. Equating structure to rigid bureaucracy confuses "two very different kinds of machine—machines designed to de-skill work and those designed to leverage users' skills" (p. 69).

Structure, then, need not be machinelike or inflexible. Structures in stable environments are often hierarchical and rules-oriented. But recent years have witnessed remarkable inventiveness in designing structures emphasizing flexibility, participation, and quality. A prime example is BMW, the luxury automaker whose success formula relies on a combination of stellar quality and rapid innovation. "Just about everyone working for the Bavarian automaker—from the factory floor to the design studios to the marketing department—is encouraged to speak out. Ideas bubble up freely, and there is never a penalty for proposing a new way of doing things, no matter how outlandish. Detroit's rigid and bloated bureaucracies are slow to respond to competitive threats and market trends, while BMW's management structure is flat, flexible, entrepreneurial—and fast. That explains why, at the very moment GM and Ford appear to be in free fall, BMW is more robust than ever. The company has become the industry benchmark for high-performance premium cars, customized production, and savvy brand management" (Edmondson, 2006, p. 72).

Dramatic changes in technology and the business environment have rendered old structures obsolete at an unprecedented rate, spawning a new interest in organizational design (Nadler, Gerstein, and Shaw, 1992; Bryan and Joyce, 2007). Pressures of globalization, competition, technology, customer expectations, and workforce dynamics have prompted organizations worldwide to rethink and redesign structural prototypes. A swarm of items compete for managers' attention—money, markets, people, and technological competencies, to name a few. But a significant amount of time and attention must be devoted to social architecture—designing structure that allows people to do their best:

Two issues are central to structural design: how to allocate work (differentiation) and how to coordinate diverse efforts once responsibilities have been parceled out (integration). Even in a group as small and intimate as a family, it's important to settle issues concerning who does what, when the "what" gets done, and how individual efforts mesh to ensure harmony. Every family will find an arrangement of roles and synchronization that works—or suffer the fallout.

Division of labor—or allocating tasks—is the keystone of structure. Every living system creates specialized roles to get important work done. Consider an ant colony: "Small workers ... spend most of their time in the nest feeding the larval broods; intermediate-sized workers constitute most of the population, going out on raids as well as doing other jobs. The largest workers ... have a huge head and large powerful jaws. These individuals are ... soldiers; they carry no food but constantly run along the flanks of the raiding and emigration columns" (Topoff, 1972, p. 72).

Like ants, humans long ago discovered the virtues of specialization. A job (or position) channels behavior by prescribing what someone is to do—or not do—to accomplish a task. Prescriptions take the form of job descriptions, procedures, routines, protocols, or rules (Mintzberg, 1979). On one hand, these formal constraints can be burdensome, leading to apathy, absenteeism, and resistance (Argyris, 1957, 1964). On the other hand, they help to ensure predictability, uniformity, and reliability. If manufacturing standards, airline maintenance, hotel housekeeping, or prison sentences were left solely to individual discretion, problems of quality and equity would abound.

Once an organization spells out positions or roles, managers face a second set of key decisions: how to group people into working units. They have several basic options (Mintzberg, 1979):

Creating roles and units yields the benefits of specialization but creates problems of coordination and control—how to ensure that diverse efforts mesh. Units tend to focus on their separate priorities and strike out on their own, as New York's police and fire departments did on 9/11. The result is suboptimization, an emphasis on achieving unit goals rather than focusing on the overall mission. Efforts become fragmented, and performance suffers.

This problem plagued Tom Ridge, who was named by President George W. Bush as the director of homeland security in the aftermath of the terrorist attacks. Ridge struggled to pull together previously autonomous agencies. But he was more salesman and preacher than boss—he lacked the authority to compel compliance. Ridge's slow progress led President Bush to create a cabinet-level Department of Homeland Security. The goal was to cluster independent security agencies under one central authority. Unfortunately, the new structure created its own problems. Folding the Federal Emergency Management Agency into the mix reduced FEMA's autonomy and shifted its priorities toward more focus on security and less on disaster management. The same agency that had responded nimbly to hurricanes and earthquakes in the 1990s was slow and ponderous in the aftermath of Hurricane Katrina and lacked authority and budget to move without a formal okay from the new secretary of homeland security (Cooper and Block, 2006).

Successful organizations employ a variety of methods to coordinate individual and group efforts and to link local initiatives with corporation-wide goals. They do this in two primary ways: vertically, through the formal chain of command, and laterally, through meetings, committees, coordinating roles, or network structures. We next look at each of these strategies in detail.

With vertical coordination, higher levels coordinate and control the work of subordinates through authority, rules and policies, and planning and control systems.

Authority

The most basic and ubiquitous way to harmonize the efforts of individuals, units, or divisions is to designate a boss with formal authority. Authorities—executives, managers, and supervisors—are officially charged with keeping action aligned with goals and objectives. They accomplish this by making decisions, resolving conflicts, solving problems, evaluating performance and output, and distributing rewards and sanctions. A chain of command is a hierarchy of managerial and supervisory strata, each with legitimate power to shape and direct the behavior of those at lower levels. It works best when authority is both endorsed by subordinates and authorized by superiors (Dornbusch and Scott, 1975). On the USS Kennedy, for example, the chain of command was crystal clear and universally accepted. It is one reason that the ultimate goal—bombs on target—was consistently attained.

Rules and Policies

Rules, policies, standards, and standard operating procedures limit individual discretion and help ensure that behavior is predictable and consistent. Rules and policies govern conditions of work and specify standard ways of completing tasks, handling personnel issues, and relating to customers and other key players in the outer environment. This helps ensure that similar situations are handled in comparable ways. It reduces "particularism" (Perrow, 1986)—responding to specific issues on the basis of personal whims or political pressures unrelated to organizational goals. Two citizens' complaints about a tax bill are supposed to be treated similarly, even if one citizen is a prominent politician and the other a shoe clerk. Once a situation is defined as one where a rule applies, the course of action is clear, straightforward, and, in an ideal world, almost automatic.

A standard is a benchmark to ensure that goods and services maintain a specified level of quality. Measurement against the standard makes it possible to identify and fix problems. During the 1970s and 1980s, American manufacturing standards lagged, while Japanese manufacturers were scrupulous in ensuring that high standards were widely known and universally accepted. In one case, an American company ordered ball bearings from a Japanese plant. The Americans insisted on what they saw as an unusually high standard—only twenty defective parts per thousand. When the order arrived, it included a separate bag of twenty defective bearings, and a note: "We were not sure why you wanted these, but here they are." Pressure for world-class quality spawned growing interest in "Six Sigma," a statistical standard of near perfection (Pyzdek, 2003). Although Six Sigma has raised quality standards in many companies around the world, its laser focus on measurable aspects of work processes and outcomes has sometimes hampered creativity in innovative companies such as 3M (Hindo, 2007, pp. 8–12). Safer, more measurable options may crowd out the elusive breakthroughs a firm needs.

Standard operating procedures (SOPs) reduce variance in routine tasks that have little margin for error. Commercial airline pilots typically fly with a different crew every month. Cockpit actions are tightly intertwined, the need for coordination is high, and mistakes can kill. SOPs consequently govern much of the work of flying a plane. Pilots are trained extensively in the procedures and seldom violate them. But a significant percentage of aviation accidents occur in the rare case that someone does. More than one airplane has crashed on takeoff because the crew missed a required checklist item.

SOPs can fall short, however, in the face of "black swans" (Taleb, 2007)—freak surprises that the SOPs were never designed to handle. In the 9/11 terrorist attacks, pilots followed standard procedures for dealing with hijackers: cooperate with their demands and try to get the plane on the ground quickly. These SOPs were based on a long history of hijackers who wanted to make a statement, not wreak destruction on a suicide mission. Passengers on United Airlines flight 93, who had learned via cell phones that the hijackers were using aircraft as bombs rather than bully pulpits, abandoned this approach. They lost their lives fighting to regain control of the plane, but theirs was the only one of four hijacked jets that failed to devastate a high-profile building.

Planning and Control Systems

Reliance on planning and control systems—forecasting and measuring—has mushroomed since the dawn of the computer era. Retailers, for example, need to know what's selling and what isn't. Point-of-sale terminals now yield that information instantly. Data flow freely up and down the hierarchy, greatly enhancing management's ability to oversee performance and respond in real time.

Mintzberg (1979) distinguishes two major approaches to control and planning: performance control and action planning. Performance control imposes concrete outcome objectives (for example, "increase sales by 10 percent this year") without specifying how the results are to be achieved. Performance control measures and motivates individual efforts, particularly when targets are reasonably clear and calculable. Locke and Latham (2002) make the case that clear and challenging goals are a powerful incentive to high performance. Performance control is less successful when goals are ambiguous, hard to measure, or of dubious relevance. A notorious example was the use of enemy body counts by the U.S. Army to measure combat effectiveness in Vietnam; field commanders became obsessed with "getting the numbers up." The numbers painted a picture of progress, even as the war was being lost.

Action planning specifies methods and time frames for decisions and actions, as in "increase this month's sales by using a companywide sales pitch" (Mintzberg, 1979, pp. 153–154). Action planning works best when it is easier to assess how a job is done than to measure its product. This is often true of service jobs. McDonald's has very clear specifications for how counter employees are to greet customers (for example, with a smile and a cheerful welcome). United Parcel Service has a detailed policy manual that specifies how a package should be delivered. The objective is customer satisfaction, but it is easier to monitor employees' behavior than customers' reactions. An inevitable risk in action planning is that the link between action and outcome may fail. When that happens, employees may get bad results by doing just what they're supposed to do.

People's behavior is often remarkably untouched by commands, rules, and systems. Lateral techniques—formal and informal meetings, task forces, coordinating roles, matrix structures, and networks—pop up to fill the void. Lateral forms are typically less formal and more flexible than authority-bound systems and rules. They are often simpler and quicker as well.

Meetings

Formal gatherings and informal exchanges are the cornerstone of lateral coordination. All organizations have regular meetings. Boards confer to make policy. Executive committees gather to make strategic decisions. In some government agencies, review committees (sometimes known as "murder boards") convene to examine proposals from lower levels. Formal meetings provide a lion's share of lateral harmonization in relatively simple, stable organizations—for example, a railroad with a predictable market, a manufacturer with a stable product, or a life insurance company selling standard policies.

But informal contacts and exchanges are vital to take up slack and glue things together in fast-paced, turbulent environments. Pixar, the animation studio whose series of hits has included Toy Story, Finding Nemo, The Incredibles, and Cars, relies on a constant stream of informal connections among managers and engineers in its three major groups. Technologists develop graphic tools, artists create stories and pictures, and production experts knit the pieces together in the final film. "What makes it all work is [Pixar's] insistence that these groups constantly talk to each other. So a producer of a scene can deal with the animator without having to navigate through higher-ups" (Schlender, 2004, p. 212).

Task Forces

As organizations become more complex, the demand for lateral communication mushrooms. Additional face-to-face coordination devices are needed. Task forces assemble when new problems or opportunities require collaboration of diverse specialties or functions. High-technology firms rely heavily on project teams or task forces to synchronize the development of new products or services.

Coordinating Roles

Coordinating roles or units use persuasion and negotiation to help others dovetail their efforts. These are essentially boundary-spanners, individuals or groups with diplomatic status who are artful in dealing across specialized turfs. For example, a product manager in a consumer goods company, responsible for the performance of a laundry detergent or low-fat snack, has what is primarily a coordinating role, spending much of the day pulling together functions essential to the product's success: research, manufacturing, marketing, and sales.

Matrix Structures

Beginning in the 1960s, many organizations in unwieldy environments developed matrix structures. By the mid-1990s, Asea Brown Boveri (ABB), the electrical engineering giant, had grown to encompass some thirteen hundred separate companies and more than two hundred thousand employees worldwide. To hold this complex collection together, ABB developed a matrix structure crisscrossing approximately a hundred countries with about sixty-five business sectors (Rappaport, 1992). Each subsidiary reported to both a country manager (Sweden, Germany, and so on) and a sector manager (power transformers, transportation, and the like). The design carried the inevitable risk of confusion, tension, and conflict between sector and country managers. ABB tried to create structural cohesion at the top with a small executive coordinating committee (thirteen individuals from eight countries), an elite cadre of some five hundred global managers, and a policy of communicating in English, even though it was a second language for most employees.

The structure worked through the 1990s, and ABB became one of Europe's most admired companies. But the inherent tensions eventually took a toll, and after a business downturn in 2000, ABB began to generate more bad news than good (Reed and Sains, 2002). Nonetheless, variations on ABB's structure—a matrix with business or product lines on one axis and countries or regions on the other—are common in global corporations.

Networks

Networks have always been around, more so in some places than others. Cochran (2000) describes how both Western and Japanese firms doing business in China in the nineteenth and twentieth centuries had to adapt their hierarchical structures to accommodate powerful social networks of merchants and workers deeply embedded in Chinese culture. One British firm tried for years, with little success, to limit the control of "Number Ones" (who headed local networks based on kinship and birthplace) over the hiring and wages of its workforce. The proliferation of information technology beginning in the 1980s led to an explosive growth of computer networks—everything from small local grids to the global Internet. These powerful new lateral communication devices often supplanted vertical strategies and spurred the development of network structures within and between organizations (Steward, 1994). Powell, Koput, and Smith-Doerr (1996) describe the mushrooming of "interorganizational networks" in fast-moving fields like biotechnology, where knowledge is so complex and widely dispersed that no organization can go it alone. They give an example of research on Alzheimer's disease that was carried out by thirty-four scientists from three corporations, a university, a government laboratory, and a private research institute.

Many large global corporations have evolved into interorganizational networks (Ghoshal and Bartlett, 1990). Horizontal linkages supplement and sometimes supplant vertical coordination. Such a firm is multicentric: initiatives and strategy emerge from many places, taking shape through a variety of partnerships and joint ventures.

In designing a structure that works, managers have a set of options for dividing up the work and coordinating multiple efforts. Structure needs to be designed with an eye toward desired ends, the nature of the environment, the talents of the workforce, and the available resources (such as time, budget, and other contingencies). The options are summarized in Exhibit 3.1.

Vertical or Lateral?

As noted, vertical coordination rests on top-down command and control. It is efficient but not always effective, and it depends on employees' willingness to follow directives from above. More decentralized and interactive lateral forms of coordination are often needed to keep top-down control from stifling initiative and creativity. Lateral coordination is often more effective but costlier than its vertical counterparts. A meeting, for example, provides an opportunity for face-to-face dialogue and decision making, but risks squandering time and energy. Personal and political agendas often undermine the meeting's purpose. A task force fosters creativity and integration around pressing problems but may divert attention from ongoing operating issues. The effectiveness of coordinators who span boundaries depends on their credibility and skills in handling others. Coordinators are also prone to schedule meetings that take still more time from actual work (Hannaway and Sproull, 1979). Matrix structures provide lateral linkage and integration but are notorious for creating conflict and confusion. Networks are inherently difficult to manage.

Basic Structural Options.

Organizations have to use both vertical and horizontal procedures for coordination. The optimal blend of the two depends on the unique challenges in a given situation. Vertical coordination is generally superior if an environment is stable, tasks are well understood and predictable, and uniformity is essential. Lateral communications work best when a complex task is performed in a turbulent, fast-changing environment. Every organization must find a design that works for its circumstances. Consider the contrasting structures of two highly successful organizations: McDonald's and Harvard University.

Structural Imperatives

Why do McDonald's and Harvard have such radically different structures? Is one more effective than the other? Or has each evolved to fit its unique circumstances? In fact, there is no such thing as an ideal structure. Every organization needs to respond to a universal set of internal and external parameters (outlined in Exhibit 3.2). These parameters include the organization's size, age, core process, environment, strategy and goals, information technology, and workforce characteristics. All these characteristics combine to dictate the optimal social architecture.

Size and Age Size and age affect structural shape and character. Problems crop up if growth (or downsizing) is not matched with fine-tuning of roles and relationships. A small, entrepreneurial organization typically has very simple, informal architecture. Growth spawns formality and complexity (Greiner, 1972; Quinn and Cameron, 1983). If carried too far, this leads to the suffocating bureaucratic rigidity often seen in large, mature enterprises.

Structural Imperatives.

In the beginning, McDonald's was not the tightly controlled company it is today. It began as a single hamburger stand in San Bernardino, California, owned and managed by the McDonald brothers. They virtually invented the concept of fast food and their stand was phenomenally successful. The two tried to expand by selling franchise rights, with little success. They were making more than enough money, disliked travel, and had no heirs. If they were richer, said one brother, "we'd be leaving it to a church or something, and we didn't go to church" (Love, 1986, p. 23).

The concept took off when Ray Kroc arrived on the scene. He had achieved modest success selling milk shake machines to restaurants. When many of his customers began to ask for the McDonald's milk shake mixer, he decided to visit the brothers. Seeing the original stand, Kroc realized the potential: "Unlike the homebound McDonalds, Kroc had traveled extensively, and he could envision hundreds of large and small markets where a McDonald's could be located. He understood the existing food services businesses, and understood how a McDonald's unit could be a formidable competitor" (Love, 1986, pp. 39–40). Kroc persuaded the McDonald brothers to let him take over the franchising effort. The rest is history.

Core Process Structure is ideally built around an organization's basic method of transforming raw materials into finished products. Every organization has a core technology that includes at least three elements: raw materials, activities that turn inputs into outputs, and underlying beliefs about the links among inputs, activities, and outcomes (Dornbusch and Scott, 1975).

Core technologies vary in clarity, predictability, and effectiveness. Assembling a Big Mac is relatively routine and programmed. The task is clear, most potential problems are known in advance, and the probability of success is high. Its relatively simple core technology allows McDonald's to rely mostly on vertical coordination.

In contrast, Harvard's two core processes—research and teaching—are far more complex and less predictable. Teaching objectives are knotty and amorphous. Unlike hamburger buns, students are active agents. Which teaching strategies best yield desired results is more a matter of faith than of fact. Even if students could be molded predictably, mystery surrounds the knowledge and skills they will need to succeed in life. This uncertain technology, greatly dependent on the skills and knowledge of highly educated professionals, is a key source of Harvard's loosely coordinated structure.

Core technologies often evolve, and significant technical innovation calls for corresponding structural alterations (Barley, 1990). In recent decades, struggles to integrate new technologies have become a fateful reality for many firms (Henderson and Clark, 1990). Existing arrangements often get in the way. Companies are tempted to mold innovative technologies to fit their existing operations. A change from film to digital photography, slide rules to calculators, or "snail mail" to e-mail gives an advantage to start-ups less committed to the old ways. Christensen (1997) found this in his study of the disk drive industry from 1975 to 1994, for example. Innovation in established firms was often blocked not by technical challenges but by marketers who argued, "Our customers don't want it."

Organizations try to insulate internal operations from outside pressures, but changing environments are a potent force. Organizations depend on the environment to provide raw materials and consume products and services. Stable, mature businesses—such as railroads, furniture manufacturers, and elementary schools—deal with slow-changing and predictable external pressures. As a result, they rely on simpler forms of organizing. Organizations with rapidly changing technologies or markets— such as high-technology electronics firms—confront a much higher degree of uncertainty. New products may be obsolete in six months or less. Uncertainty and turbulence press for new roles and more elaborate, flexible approaches to vertical and lateral coordination.

Some organizations are more susceptible to outside influences than others. Public schools, for example, are highly vulnerable to external pressures because they have so little capacity to claim the resources they need or to shape the results they are supposed to produce. In contrast, an institution like Harvard is insulated from such intrusions by its size, elite status, and large endowment. The university can therefore afford to offer low teaching loads, generous salaries, and substantial autonomy to its faculty. A Harvard diploma is taken as sufficient evidence that instruction is having its desired effect.

Strategy and Goals Strategic decisions are future oriented, concerned with long-term direction (Chandler, 1962; Mintzberg, 1994). Across sectors, a major task of organizational leadership is "the determination of long-range goals and objectives of an enterprise, and the adoption of courses of action and allocation of resources necessary for carrying out these goals" (Chandler, 1962, p. 13).

A variety of goals are embedded in strategy. In business firms, goals such as profitability, growth, and market share are relatively specific and easy to measure. Goals of educational or human services organizations are typically much more diffuse: "producing educated men and women" or "improving individual well-being." This is another reason Harvard adopts a more decentralized, loosely integrated system of roles and relationships.

Historically, McDonald's had fewer, more easily quantifiable, and less controversial goals than those of Harvard. This aligned well with the McDonald's centralized, top-down structure. But that structure has become more complex as the company's size and global reach have fostered levels of decentralization that allowed outlets in India to offer vegetarian cuisine and those in France to run ads attacking Americans and American beef (Tagliabue, 1999; Stires, 2002; Arndt, 2007).

To complicate matters still further, stated goals are not the only ones an organization pursues. Westerlund and Sjostrand (1979) suggest various others:

Understanding linkages among goals, structure, and strategy requires a look beyond formal statements of purpose. Schools, for example, are often criticized if structure does not coincide with the official goal of scholastic achievement. But schools have other, less visible goals. One is character development, often espoused with little follow-through. Another is the taboo goal of certification and selection, as schools channel students into tracks and sort them into careers. Still a third goal is custody and control—keeping kids off the streets and out from underfoot. Finally, schools often herald honorific goals such as excellence. Strategy and goals shape structure, but the process is often complex and subtle (Dornbusch and Scott, 1975).

Information Technology New technologies continue to revolutionize the amount of information available and the speed at which it travels. Once accessible exclusively to top-level or middle managers, information is now easy to get and widely shared. E-mail has made communication immediate and far reaching. With the press of a key, anyone can reach another person—or an entire network. All this makes it possible to move decisions closer to the action.

In the 2003 invasion of Iraq, for example, U.S. and British forces had an obvious advantage in military hardware. They also had a powerful structural advantage because their superior information technology let them develop a much more flexible and decentralized command structure. Commanders in the field could change their plans immediately in response to new developments. Iraqi forces, meanwhile, had a much slower, more vertical structure that relied on decisions from the top. A major reason that Iraqi resistance was lighter than expected in early weeks was that commanders had no idea what to do when they were cut off from their chain of command (Broder and Schmitt, 2003).

Later, however, the structure and technology so effective against Iraq's military had much more difficulty with the rising resistance movement, which evolved a loosely connected structure of entrepreneurial local units that could adapt quickly to U.S. tactics. New technologies like the Internet and cell phones enabled the resistance to structure itself as a network, or "complex adaptive system" (Waldrop, 1992, p. 145)—a set of loosely connected units, each pursuing its own agenda in response to local conditions. The absence of strong central control in such networks can be a virtue because local units can adapt very fast to new developments and because the loss of any one outpost does little damage to the whole.

By increasing the flow of information, improved technology reduces uncertainty. Galbraith (1973) defines uncertainty as the difference between the information an organization has and the information it needs. One way that organizations reduce this gap is by increasing their ability to process information (with information systems, for example). A second alternative is to reduce the need for information by creating self-contained units (which can work on their own using information at hand) or by adding slack resources (extra copying machines or support staff, for example, so that people don't have to fight over access).

Information technology has made flatter structures both possible and inevitable because "the information-based organization needs far fewer levels of management than the traditional command-and-control model" (Drucker, 1989, p. 20). As technology spread in the late twentieth century, organizations around the world made deep cuts in both management levels and support staff, a process that still continues.

Nature of the Workforce Human resource requirements have changed dramatically in recent decades. Many lower-level jobs now require higher levels of skill. A better-educated workforce expects and often demands more discretion in daily work routines. Increasing specialization has professionalized many functions. Professionals typically know more than their supervisors about technical aspects of their work. They expect autonomy and prefer reporting to professional colleagues. Trying to tell a Harvard professor what to teach is an exercise in futility. In contrast, giving too much discretion to a youthful, low-skilled McDonald's worker could become a disaster for both employee and customers.

Dramatically different structural forms are emerging as a result of changes in workforce demographics. Deal and Kennedy (1982) predicted early on the emergence of the atomized or network organization, made up of small, autonomous, often geographically dispersed work groups tied together by information systems and organizational symbols. Drucker makes a similar observation in noting that businesses increasingly "move work to where the people are, rather than people to where the work is" (1989, p. 20).

Challenges of Global Organization

In sum, numerous forces affecting structural design create a knotty mix of challenges and tensions. It is not simply a matter of deciding whether we should be centralized like McDonald's or decentralized like Harvard. Many organizations find that they have to do both and somehow accommodate the competing structural tensions.

Two electronics giants, Matsushita in Japan and Philips in the Netherlands, have competed with one another around the globe for half a century. Over time, Matsushita developed a strong headquarters, while Phillips was more decentralized, with strong units in different countries. The pressures of global competition pushed both to become more alike (Bartlett, 2006). Philips struggled to gain the efficiencies that come from selling the same products around the world. Meanwhile, as Matsushita gradually discovered, "No company can operate effectively on a global scale by centralizing all key decisions and then farming them out for implementation. It doesn't work.... No matter how good they are, no matter how well supported analytically, the decision-makers at the center are too far removed from individual markets and the needs of local customers" (Ohmae, 1990, p. 87).

The structural frame looks beyond individuals to examine the social architecture of work. Though sometimes equated with red tape, mindless memos, and rigid bureaucrats, the approach is much broader and more subtle. It encompasses the freewheeling, loosely structured entrepreneurial task force as well as the more tightly controlled railway company or postal department. If structure is overlooked, an organization often misdirects energy and resources. It may, for example, waste time and money on massive training programs in a vain effort to solve problems that have much more to do with social architecture than people's skills or attitudes. It may fire managers and bring in new ones, who then fall victim to the same structural flaws that doomed their predecessors.

At the heart of organizational design are the twin issues of differentiation and integration. Organizations divide work by creating a variety of specialized roles, functions, and units. They must then use both vertical and horizontal procedures to lash the many elements together. There is no one best way to organize. The right structure depends on prevailing circumstances and considers an organization's goals, strategies, technology, people, and environment. Understanding the complexity and variety of design possibilities can help create formal prototypes that work for, rather than against, both people and collective purposes.