8—
Coordination and Tradeoffs with Other Goals

It should work here because it's a rational idea, but it won't happen because the regions in this area are so parochial and selfish.
—Richard Sklar, (then) general manager of the San Francisco Public Utilities Commission, referring to a plan to centralize decision making for Bay Area public transit, Oakland Tribune, 10 September 1982

But my biggest problem would be the loss of local control. We at BART in Contra Costa County are already victims of a regional kind of government.
—BART Director Nello Bianco, referring to the same plan, Oakland Tribune, 10 September 1982

Any consideration of whether a formal system of central coordination is to be preferred to an informal system of mutual accommodation and adjustment must confront the type and extent of interdependence of the components to be coordinated. Other things

being equal, the more extensive the interdependence and the more it takes a multilateral form, the more appropriate would be a formal, centralized scheme of coordination. Systems described by less complex patterns of interdependence might be just as well coordinated by informal mechanisms. But other things are rarely equal.

In some cases formal solutions to problems of coordination may provide better results than informal solutions when those results are evaluated solely in terms of coordination. It is difficult if not impossible, however, to find an organizational system whose solitary reason for existence is to provide for coordination.

If what constituted an acceptable level of coordination could be ascertained, and it could be shown that additional increments of coordination necessitated lower attainment of some other important goals, we might tend to think twice about seeking that higher level. Rather than trying to achieve "optimal" coordination, we might choose to look for an organizational design that could provide an acceptable mix of goals. Such tradeoffs in goals are inevitable features of the process of organizational design. They may be set in legal terms or may have to do with the capacity of an organization to work well under one set of conditions and poorly under another. Frequently tradeoffs are no more than implicit. Sometimes they are quite explicit.

Simon describes the process as follows:

In the decision-making situations of real life. a course of action, to be acceptable, must satisfy a whole set of requirements, or constraints. Sometimes one of these requirements is singled out and referred to as the goal of the action. But the choice of one of the constraints, from many, is to a large extent arbitrary.^[1]

The problem therefore becomes one of deciding which organizational design permits the attainment of some satisfactory mix of the set of goals involved, especially where the achievement of any one goal may mean less of the others.

When we consider tradeoffs among conflicting goals, it is essential to probe deeper than simply the allocation of resources by an organization to one or another combinations of these values. The problem has to do with the design of organizational structures themselves. Superficial appearances of rationality are less important than achieving the given mix of values, whatever specific form it might take. For, as Niskanen notes, we

should judge social institutions pragmatically, not in terms of the perceived rationality of their structure and procedures, but in terms of their performance in creating those conditions [we] value.^[2]

If a determination is to be made of the overall effectiveness of one type of organizational system as compared with another, it is essential to state explicitly the mix of goals the system is supposed to achieve as well as the environmental constraints faced in the pursuit of those goals. What is the mix of goals that one system can provide as opposed to the mix provided by another? What other goals are important?

This chapter explores some characteristics of organizational designs that promote differentiation of services (flexibility), representation, and reliability.^[3] Obviously there are other important goals, but the discussion focuses on these three because of their immediate importance to the delivery of public services, and because they adequately illustrate the problems inherent in designing organizations to achieve conflicting goals. By characteristics, I refer both to the features of individual organizations, and to those of the organizational system as a whole, including such factors as organization size, extent of formal autonomy, extent of overlap and duplication, and number of organizations within the system.

Differences in Operating Environment

Although organizations dramatically improve on the ability of the individual to make rational decisions, they are themselves subject to limits on rationality. The more complex their operating environment, the more likely organizations will oversimplify their models of reality and be correspondingly less effective. The capacity of an organizational system to match its procedures to the environment it faces—differentiation and flexibility—is therefore an important consideration.

Given the perspective adopted in this study that organizations are instruments,

there is no such thing as a "good" organization in any absolute sense. Always it is relative; and an organization that is good in one context or criterion may be bad under another.^[4]

Building into an organizational design a capability for differentiating approaches as contexts vary and a capacity for flexibility in the face of changing conditions over time should improve the probability of effective performance and the "fit" of the "form" to its "context."^[5] Several factors make it difficult for organizations to provide differentiated services. I suggest here some approaches that help to mitigate the effects of these factors even if they do nothing to eliminate the factors themselves. Given that this is essentially the same approach the Founding Fathers took toward the problem of faction in the design of the U.S. Constitution, it seems that I am traveling in good company.

That individual decision makers are capable of achieving no more than bounded rationality is well established. Limitations of intellectual ability, time, and other scarce resources prevent the decision maker from comprehensively evaluating all or even most possible goals, or if goals are fixed, from evaluating all possible alternatives for achieving those goals. The decision maker is therefore constrained to make most decisions at the margin and to limit substantially the range of alternatives considered.^[6] It follows that after a certain level of complexity in the environment is reached, a given organization's model of reality will not become any more complex because of cognitive limitations on analysis. Even where conditions vary significantly in ways affecting the appropriateness of particular alternatives, the organization will tend to stick to the set of alternatives that it has already selected, in part because of the expense and difficulty of analysis. This can lead to forcing solutions onto areas where they are inappropriate and perhaps pathological. Asking a single organization to deliver services across several areas whose characteristics differ widely may be asking it to do too much. Rather than deliver services more effectively, it will be prone to deliver services poorly matched to the conditions of the areas it serves.

This in fact was one of the arguments made by the Antifederalists in their efforts to defeat the new Constitution proposed at the Philadelphia Convention:

Within a large territory the various regions would strive against one another; different climates, products, interests, manners, habits, laws, would lead to discord. How to legislate uniformly for a land so diverse? A law which suited one pan might oppress another.^[7]

The Antifederalists contended therefore that the answer was a series of small republics, each of which could deal effectively with what would amount to a homogeneous population living under conditions of little diversity. The activities of government could be best carried out at the local and state levels, not at the national level.

The Bay Area transit system is marked by wide variation in the conditions facing the several operators. These conditions include technical issues to be resolved in providing service and special problems that confront each operator. Constraints deriving from formal political arrangements and clientele groups are dealt with later in this chapter, in the section on representation.

The size and shape of a jurisdiction, in concert with the existence of geographic barriers, affects the routes that can be established. These factors also influence development patterns, which in turn affect travel patterns. Among the operators, travel patterns vary a great deal in terms of off-peak travel, local versus trunkline service, number of major destinations, and length of trips. Differences between bus and rail are also significant. The latter remains fixed, the former can be modified as conditions indicate.

AC Transit operates in territory that is large in size and long and narrow in shape. There is one principal metropolitan area within its jurisdiction, Oakland, and it provides commuter service to San Francisco. On average, the population is of medium density in District One and low density in District Two. The only other operator with which AC is highly interdependent is BART. AC runs only buses, encompasses two counties, and is governed by an elected board.

Samtrans also has a geographically long and narrow jurisdiction, but has a smaller overall population, with density running from medium to low. In addition to local service, Samtrans provides trunkline service to San Francisco and the Daly City BART station, which are both at the northern end of its jurisdiction. AC and Samtrans both encompass hilly terrain. Samtrans is interdependent with BART, Muni, and Santa Clara and runs only buses. It is coterminous with San Mateo County and is governed by the county supervisors sitting as a transit board.

Santa Clara has a larger, generally flat territory. It also has the lowest population density in the Bay Area. County residents depend

heavily on the automobile for transportation: only about 5 percent of the trips in Santa Clara County are made on public transit. There is a weak focus on San Jose as a destination. Santa Clara has only Samtrans (and BART at one station) to consider in problems of interdependence. It is the most functionally independent of the six operators. Santa Clara operates only buses, run as part of the county transportation agency (which is also responsible for county roads and airports).^[8]

Golden Gate has the longest, narrowest territory of the six. It is bounded by hills that serve as geographic barriers. Beyond U.S. 101, it has few highways to serve. Almost all its commuters go to San Francisco, with little drop-off before that destination. There is far less indigenous employment in Marin and Sonoma counties than in San Mateo or Alameda counties. The major streets run north-south with few east-west lines of communication. Relative to the trunk-lines, local service demands are not very great. Neither is there much of a reverse commute. Golden Gate runs both ferry and bus service and operates the Golden Gate Bridge and its approaches. It is governed by a composite board of local Bay Area government officials, and its principal revenue is derived from tolls on the Golden Gate Bridge.

Muni contrasts with the other bus operators more than they differ from each other. It operates in a compact, almost square, geographic area, with a higher population density than any other agency. Travel destinations are widely dispersed. Furthermore, there is considerable off-peak travel by residents. Muni operates buses, trolleys, light-rail trains, and cable cars. Unlike other operators, it carries many tourists on all of its modes, not just on the famous cable cars. Muni's jurisdiction is also a focal point for other operators—BART, AC, Golden Gate, and Samtrans. Muni lies entirely within the city and county of San Francisco and is operated as a city agency (which means it must compete with other city agencies for funds), responsible to the San Francisco Public Utilities Commission and Board of Supervisors.

BART is different. It employs a heavy-rail mode and was "superimposed" on existing agencies. By the very nature of its design, it is entirely a trunkline service. It depends greatly on other operators for feeder service to its trunklines. Superimposition means that BART competes more with others than any of the other operators. It also

has the largest jurisdiction of any operator, encompassing three counties. It is run by an elected board.

Consider for a moment the ramifications of these differences. High-density, compact areas are best served by grid systems of bus routes, especially where there are a number of general destination areas. Conversely, low-density areas characterized by elongated shapes are better served by looping bus routes. The former depend on close headways with frequent transfers, the latter on fewer transfers and correspondingly longer headways.^[9] Travel patterns typified by long commutes along narrow corridors (such as Sonoma County to San Francisco) preclude buses from making more than one trip per peak period, necessitating split shifts for the drivers. On shorter routes straight shifts are possible. Technical requirements for equipment differ between long-haul and short-haul runs, differences ranging from seating configurations to transmission gearing and air conditioning. Maintenance requirements are also different.

Population density affects the kind and quality of service that can be offered, particularly frequency of buses. Routing and scheduling are also affected by the distribution of destinations. If destinations are dispersed, it matters whether they are along a corridor or radially distributed. The need for cooperation from other operators—high levels of interdependence—to operate one's own system effectively alters the need to consider outside factors. Organizations competing for passengers face a similar problem, as does the operator that is a focal point for the passengers of other operators. In this respect, Santa Clara can operate much more freely of other agencies than can BART or Muni. The institutional environments of the operators affect their funding and independence. AC can operate with greater independence than Muni, for example. Similarly, the number of local jurisdictions that the operator encompasses affects the complexity of its institutional environment. Complexity of this type, for example, is more characteristic of BART than of Samtrans.

General differences in operating environments cannot be analyzed apart from the major problems peculiar to individual agencies,^[10] because "each operator faces uniquely different operating problems. These individual operating problems and the management regimes which have emerged to cope with them also tend to limit the areas of mutually beneficial coordination and cooperation."^[11] These unique problems combine with different operating conditions to defy atten-

tion by a single management regime either simultaneously or in series. No transit agency in the Bay Area has its own problems sufficiently under control to permit it to acquire the additional burdens that would result from any formal consolidations.

At Muni, improving service standards, including operator competence and courtesy, was essential. Muni was plagued by a series of accidents. Its physical plant was badly rundown. The cable cars were in the process of a multimillion dollar restoration. Maintenance quality at Muni was also below par. And Muni had to devote time to constructing and debugging its new Metro light-rail system. Because Muni is a department within the city and county of San Francisco, it must compete for budget attention with other departments; it has no dedicated source of subsidy.

On the other hand, BART has a relatively new physical plant. Its resources have been devoted to debugging its highly complex rail system in order to improve service reliability and safety. It has also expanded its operating hours and direct service to San Francisco. Each change has required adjustments in train schedules and maintenance operations. BART management has also had two major labor strikes and a host of all-consuming technical and political problems created by the 1979 tube fire with which to contend. Their after-effects are still being felt. BART has also had to spend resources to rebuild public confidence in the wake of its failure to deliver on the many promises made during its design and construction phase. It has had difficulties with Contra Costa County, which pays for rail service, but receives very little, and Daly City, which receives rail service but lies outside the BART district. And some effort is being made toward expanding service in bart's southern area.

Golden Gate has two thorny political problems to confront. Its widely anticipated ferry service has been plagued by problems of reliability. The gas-turbine-powered ferries are both expensive and difficult to maintain.^[12] Because they have never approached the passenger load that was forecast for them, they require large subsidies. To make matters worse, the subsidies pay for service provided to relatively affluent Marin County commuters. Tied into the issue of the ferries is the levy of tolls for the Golden Gate Bridge. Once the bridge debt was paid off, toll revenues were diverted to subsidies for the buses and ferries. The ferries consume a disproportionate amount of these subsidies.

Santa Clara has had to devise an entire bus system to replace the dying private operators it supplanted. One of its major problems is to provide adequate service for transit captives in a county that has low population density and heavy reliance on automobile travel. Santa Clara is probably the antithesis of Muni in this respect. It has also had to remedy a bad management decision, made early in its history, to purchase propane-powered buses that proved unreliable.

AC, as one of the oldest operators, has its routes and clientele well ordered and deals with a stable population in most of its service area. Its primary efforts are directed to fine-tuning and day-to-day operations. For some years its maintenance program has been considered to be among the best in the country. However, in the wake of Proposition 13, AC had to devote attention to managing cutbacks, including both layoffs and service cuts. Both are highly charged political issues. It is also facing declining patronage on its transbay runs (and some other runs in the East Bay) that compete with BART service. Its declining passenger load has become its single largest problem.

In general, Samtrans has had no special problems analogous to those described for the other operators.

The interaction of operating conditions with the problems peculiar to each operator heightens the differences between agencies and the approaches they must employ to be effective. Nowhere is this more evident than in the problem of fare evasion. Muni is faced with a loss of fare revenues estimated to be 11 percent of fares collected. On the other hand, BART lost only about $1 million during a two-year period in which it collected some $90 million in fares, a little more than 1 percent. Muni uses cash fares, passes, transfers, and magnetically encoded farecards. BART uses only magnetically encoded farecards. Differences in techniques are a function of different transit modes. Each system is therefore open to unique kinds of fare evasion and fraud. Furthermore, Muni must rely on its operators alone for enforcement of its fare policies, whereas BART has its own police force, including both uniformed and undercover officers. The high usage of Muni's services precludes a self-service fare system combined with roving inspectors, such as San Diego Transit uses on its "Tijuana Trolley." Different passenger loads, types of vehicles, and fare devices make for different solutions to fare evasion. No one system can work for all.

It is simply unreasonable to expect any one set of managers to

contend with so many different operating conditions and so many different urgent problems. By themselves they are complex and difficult; combined they are no less than overwhelming. The results of a consolidated transit agency for the Bay Area would be standardization, oversimplification, and de facto decentralization to an extent that would defeat the intent of such a consolidation. Even lesser plans for consolidation of pairs of operators would increase the complexity of problems to impossible levels. Where it might make sense to consolidate operators on the basis of similar operating conditions, such as Samtrans and Golden Gate, there are great differences among clientele and urgent problems. Where there is geographic contiguity (if not overlap), such as between AC and BART, differences in technology and operating conditions would make consolidation equally complex.

Consistency, Goal Displacement, and Institutionalization

So far only cognitive limitations on the ability of organizations to contend individually and effectively with variations in operating conditions have been discussed. There are also important social and psychological processes that tend to suppress differences in organizational approaches and simultaneously to produce drives for homogeneity: pursuit of consistency as an intrinsically valuable goal, displacement of goals, and institutionalization.

For some years boarding charges for bus service at Washington Metro differed among the three local jurisdictions. However, there was persistent and strong pressure on the District of Columbia to bring its fares in line with the fares in Virginia and Maryland, both by those two jurisdictions and by Metro's management. The District was pressured to raise its fares to give the three jurisdictions a consistent fare policy. The District of Columbia, however, resisted these pressures because it felt an obligation to subsidize transit services for its extensive poor population. In January 1981 the District finally came into line with the other jurisdictions, not because of a change in philosophy, but because financial constraints compelled it to do so. At last the fares were consistent.

Now, why should fares be consistent? In this case, consistency for its own sake appears to be the answer. The drive for consistency be-

came more powerful because it occurred within the context of a single operating entity. The problem is that "the pursuit of consistency from one coordinating decision to another in central coordination often degenerates into the pursuit of some kind of superficial uniformity."^[13] At Metro it was argued that different fare structures were too confusing for patrons, but it seems more likely that "limited competence and weak motivation both tempt[ed] coordination to substitute a routine yea- and nay-saying for a more creative approach to consistency and coordination."^[14]

At the Metro one finds a much stronger (and more effective) drive to produce consistent fare policies than in the Bay Area transit system. This occurs even though differences in philosophy and wealth among Metro's members indicate that a differentiated fare structure would probably serve their constituencies better. In the Bay Area, MTC may push for consistency of fares, but it can only indirectly influence the adoption of such a policy by the operators. Because of its multiple independent agencies, the Bay Area exhibits less pressure for consistency in the first place, and greater ability to resist such pressure in the second, than does Washington's unified organizational structure.

Displacement of goals has to do with both Veblen's notion of "trained incapacity" and Dewey's concept of "occupational psychosis." "In general, one adopts measures in keeping with one's past training, and under new conditions which are not recognized as significantly different, the very soundness of this training may lead to the adoption of the wrong procedures."^[15] Processes inherent in bureaucratic forms of organization tend to reinforce this propensity:

1. An effective bureaucracy demands reliability of response and strict devotion to regulations.

2. Such devotion to regulations leads to their transformation into absolutes; they are no longer conceived as relative to a set of purposes.

3. This interferes with ready adaptation under special conditions not clearly envisaged by those who drew up the general rules.

4. Thus, the very elements which conduce toward efficiency in general practice produce inefficiencies in specific circumstances. . . . These rules in time become symbolic in cast, rather than strictly utilitarian.^[16]

Although Merton aimed his discussion of goal displacement at the problems of changing conditions facing one organization over a period of time and of anomaly in the face of generally applicable rules, it is not difficult to see the role that displacement of goals plays in a situation characterized by great diversity. The same problem of trained incapacity arises as does the absolutism of rules. Like a push for superficial consistency, displacement of goals weakens the ability of an organization to deliver different kinds of services where it is appropriate to do so. If several organizations are employed, each matched to a specific set of conditions, the effects of goal displacement in this sense will be minimized even though the base causes remain untouched.

If one consequence of goal displacement is the conversion of procedures and rules from utilitarian to symbolic in character, institutionalization refers to sociopsychological processes that move ultimately toward the same end:

To institutionalize is to infuse with value beyond the technical requirements of the task at hand. The prizing of social machinery beyond its technical role is largely a reflection of the unique way in which it fulfills personal or group needs. . . . The test of infusion of value is expendability . If an organization is merely an instrument, it will be readily altered or cast aside where a more efficient tool becomes available. . . . The transformation of expendable technical organizations into institutions is marked by a concern for self-maintenance. . . . Various elements in the association have a stake in its continued existence.^[17]

Thus Muni personnel infuse the use of a grid system for bus routing with intrinsic value; they personally identify with that particular technique. Their criticism of agencies, such as Samtrans, that do not use such a system neglects the importance of differences in operating conditions for determining the appropriate techniques.^[18] The same process may also be seen in the controversy over whether to use a light-rail system or buses for public transit. Attachment to light-rail by its proponents exceeds its instrumental value for providing effective public transit.

As techniques or operating paradigms undergo a transformation from extrinsic (or instrumental) to intrinsic valuation, it becomes increasingly difficult to adapt them to different demands and different conditions. How can the effects of this process be mitigated? One

approach is to minimize the variation in operating conditions faced by any one organization, so that institutionalization would have the least possible effect. Instead of consolidating organizations, organizational designers might elect to create new ones in order to match them as closely as possible to homogeneous sets of operating conditions. Attachment to particular techniques or operating paradigms will be less likely to cause serious disturbance under this arrangement. Muni's grid system is appropriate to its operating conditions, so institutionalization of that technique is not important, as long as relevant conditions do not change significantly. Even though Muni and Samtrans are interdependent, we might be slow to consolidate them, as they face such different operating conditions that differentiation of services is an important goal.

However, understanding the processes of goal displacement and institutionalization only helps explain motivations for individual or group behavior. It does not explain how they are translated into organizational behavior. Organizational goals and procedures are established by individuals who collectively have sufficient control of organizational resources to commit them in certain directions and withhold them from others.^[19] This collectivity within the organization is referred to as the "dominant coalition." How is the dominant coalition linked to goal displacement and institutionalization? "So long as the organization presents favorable spheres of action to individuals in highly discretionary jobs, we have strong motivation for them to avoid decisions which would end those spheres of action."^[20]

Techniques or operating paradigms with which those who comprise the dominant coalition identify will be employed by the organization. Inasmuch as the fate of the members of the dominant coalition depends on those particular techniques or operating paradigms, the more will they resist their modification or deletion. The stronger the dominant coalition within the organization the more likely it is that operating paradigms that are no longer instrumentally useful (or would be misapplied) will be retained. Retention of particular techniques serves personal or group ends; the dominant coalition provides the mechanism for their achievement. As the survival of the horse cavalry in the U.S. Army up to World War II (even after German armor and mechanized infantry had decimated the renowned Polish cavalry) attests, institutionalization and the dominant coalition

combine to make even obviously obsolete operating paradigms highly resistant to change.^[21]

Furthermore, by setting the goals of the organization, the dominant coalition focuses the organization's attention (which is limited by intellectual abilities and organizational resources) on particular sets of problems. It cannot deal with all problems and therefore is likely to resist the intrusion of new problems. It therefore tends to deemphasize those that do not fit its own ideas of what the organization is about, thus making differentiation more difficult.

For example, at Washington Metro the dominant coalition was composed of rail design and construction personnel at a time when private bus companies were collapsing around them amid public clamor for a public bus agency. Although its full energies were focused on the design and construction of the rail system, Metro was forced to absorb bus operations because it was the only public transit agency available.

Metro's dominant coalition regarded the buses as an intrusion and a distraction from its real goal—building a railroad. Rail was the preferred mode of transit in any case. There was real rail-bus conflict. Had Metro acquired bus operations under different conditions or at a later date, conflict likely would have been less severe, but the bus people would still have had to fight their way into the decision-making processes of the organization. As it was, bus operations were relegated to stepchild status and received little attention from Metro management. Buses are still not run as effectively as they are at AC Transit, which was established at the same time as BART, permitting it to focus all of its attention on bus operations while BART specialized in rail design and construction.

Representation

Public transit is subject to more political conflict at the local level than virtually any service, excluding police services and education. The issue is characterized by intense feelings, diversity of opinion, and substantial subsidy of capital outlays and operating costs by local taxpayers. Political differences and problems of representation are inescapable facts of transit organization, and any design that em-

phasizes efficiency and coordination to the exclusion of these other goals is doomed to failure, as an Oakland city councilman noted:

Academicians seem to forget at times that we cannot begin with a clean slate. There are all sorts of problems of equity, conflicts of organizational character and methods, and diverse actors and interests involved which would mitigate against any merger or consolidation proposal. These problems and differences must be worked out.^[22]

Given differences among various interests in the Bay Area transit community, representativeness is another important goal.

Public organizations "draw their support, in the coinage of taxes and political legitimacy, from the coffers of the general public,"^[23] yet these organizations do not (for the most part) operate under market conditions. So consumers have only indirect influence on their activities. Moreover, because such entities operate on the basis of subsidies, those bearing the cost of subsidies and those benefiting from them may be different. Representation of these different groups is therefore especially important in the provision of public transit and other similar public services.

Because the daily operation of a modern government is impossible by direct democracy, and some delegation of authority is therefore required, a key problem becomes linking the views of those governed with the behavior of their representatives. Although several factors affect the quality of representation in any organizational system—for example, characteristics of groups, the position of groups within the social structure, and the quality of the groups' organizations^[24] —in the context of a discussion of organizational design, access to decision-making points is a more important consideration:

Power of any kind cannot be reached by a political interest group, or its leaders, without access to one or more key points of decision in the government. . . . The key decision points may be explicitly established by the formal legal framework or they may lie in the gaps and interstices of the formal structure, protected by custom or semi-obscurity.^[25]

Group characteristics are outside the scope of this study. As my concern is with the impact of organizational design on representation, I focus on the problems of access and accountability: What features of an organizational system are particularly effective for pro-

moting access and accountability? What changes can be wrought in the configuration of the organizational system to improve existing access and accountability?

Another argument made by the Antifederalists in support of small republics was their capacity for ensuring "a strict responsibility of the government to the people."^[26] The problem of a republic is to make the representatives responsible to their constituents. Toward that end, short terms of office, frequent rotation, and numerous representatives contribute, but are insufficient to ensure it: "Effective and thoroughgoing responsibility is to be found only in a likeness between the representative body and the citizens at large."^[27] According to Melancton Smith, representatives "should be a true picture of the people; possess the knowledge of their circumstances and their wants, and be disposed to seek their true interests."^[28] It was important to minimize the diversity of interest contained within a particular jurisdiction for it was possible for governmental actors to know clearly the interests of their constituents only under those conditions. This point continues to be relevant. Smaller jurisdictions not only promote greater access to governmental actors but the greater ability of those actors to comprehend the interests of their constituents as well.

A defining feature of the federal and state governmental systems in the United States is that they contain a multiplicity of points of access. This is no less true for local government and public transit agencies as they exist in the San Francisco Bay Area and in the Washington, D.C., metropolitan area, where one finds not only multiple independent agencies, but significant overlap among those agencies as well. In systems characterized by mutual adjustment processes "almost any value that even relatively small numbers of citizens moderately or strongly wish to see weighed into the policy-making process will be weighed in at some value significantly above zero,"^[29] because processes of mutual adjustment (characterized by multiple independent organizations) permit multiple points of access.

Clearly, a principal advantage of multiple points of access to decision making is the increased probability that one will find someone congenial to one's point of view who also happens to be in a position to make decisions. This is one of the underlying features of Lindblom's disjointed incremental decision-making system. "Multiplicity copes with the inevitability of omission and other errors in

complex problem solving."^[30] Multiple access points are of course simply a form of organizational redundancy, which is most effective when domains of the access points overlap. However, multiplicity can be effective, even when there is no overlap, through the informal application of pressure from one organization to another. As Lindblom notes: "multiple decision makers . . . will compellingly call to each others' attention aspects of the problem they themselves cannot (or will not) analyze."^[31]

Although, over time, the fact of multiple independent organizations may tend to promote parochial interests at the expense of more broadly defined "public" interests,^[32] it is not the case that where differences in interest exist consolidating those organizations will decrease or eliminate them (or, for that matter, the conflict resulting from them).^[33]

Formally, AC is supposed to provide transit service for a specific geographic area. But within the larger Bay Area transit system, it represents the views of its patrons as opposed to the patrons of BART. The Oakland Tribune reported the comments of ac's (then) general manager, a long-time participant in Bay Area transit politics:

"If we become an adjunct of BART or part of a consolidated system, we might be put on a back burner," Nisbet said. "We worry that the needs of people who depend on AC might be overshadowed by a glamorous rail system that serves a different clientele. White, middleclass people going to San Francisco might benefit, but it would be at the expense of service we provide to lower income people who may ride a bus for only a few blocks or who may have no need to ride BART."^[34]

Or take MTC. While formally playing the role of a Management and Planning Organization for the nine Bay Area counties transit systems, MTC also acts informally as a lobbyist on behalf of the Bay Area in the state and federal capitals, and within the Bay Area represents regional interests.^[35]

Although some in the public administration are prone to dismiss conflict between organizations as an artifact of their independence, conflict often reflects basic underlying differences among the clientele they serve. However,

even single organizations will often not pursue logically consistent policies because there is no way to aggregate the preferences of a few

individuals—say members of a city council, school board, or legislature—so that the resulting outcomes of decision-making processes are logically consistent with each other.^[36]

In such cases merger does not eliminate conflict; it often elevates it beyond its original level by predicating successful coordination on agreement across a wider range of values, and by creating a zero-sum game. It reduces the potential for successfully decomposing complex sets of problems, and makes their solution more difficult. Certainly conflict among local jurisdictions in Washington Metro has not been eliminated by the fact of a single operating organization.

When interests are represented by separate organizations, it is relatively simple to ascertain the agendas of those organizations. Conflicts between organizations tend to be out in the open and are therefore more easily comprehended than when they occur within single organizations where different interests may be represented by informal factions. Conflict within organizations may also result in internal sabotage. Therefore, in situations where very real differences in interest exist it may be better to retain multiple independent organizations in order to contain and manage conflict more effectively.

For example, at Golden Gate there are conflicts at the board level that frequently revolve around differences in interest between the northern counties (Sonoma, Marin, Mendocino) and San Francisco. Where San Francisco, in order to alleviate traffic congestion and parking problems, seeks to stem the tide of automobiles into the city by charging higher tolls on the Golden Gate Bridge, the northern counties (especially Marin and Sonoma) seek to maintain lower tolls to facilitate that access for their constituents. This conflict overshadows and colors other issues at Golden Gate.

Informal organization facilitates representation in two ways. It plays an indirect role by coordinating multiple independent organizations, thereby making possible the coexistence of effective coordination and multiple points of access. It also directly affects representation by providing access to decision making through informal channels that complement formal channels. This role is performed in two ways: when individuals or groups develop informal ties to decision makers, and when other organizations (such as city governments) make informal connections with transit organizations and are thereby better able to represent their own clienteles.

Sometimes elected governing boards of organizations are less im-

portant than their general managers and staffs as points of access to the decision-making process. Administrators frequently control their boards to such an extent that they become little more than rubber stamps for staff policies.^[37] Certainly this was the case during one general manager's tenure at AC, and for one at Metro as well. In fact, it is a common drive for general managers to seek to control their governing boards. To the extent they are successful, they eliminate a great deal of uncertainty from their immediate environment. However, simultaneously, they reduce the board's representative function and increase the importance of public access to the staff.

Control of governing boards appears to be enhanced by several factors, including nonpartisan elections, lay board members, and highly complex technical issues. Established informal norms against serious questioning of management also play an important role. In the case of AC transit, board members who tried to play activist roles were sanctioned informally by other members. The ability of the board to perform its representative function is also limited by the general manager's ability to monopolize expertise and information. Recognizing that monopoly, people from outside the organization, say from MTC, who have difficulty working on programs with the general manager and staff will go directly to board members with information on these programs, apprising the general manager of their action, but simultaneously weakening his hold on the board.^[38] The problem of monopoly of information is especially important, because, unlike legislative bodies such as the U.S. House of Representatives, these boards have no full-time professional staffs to provide them with independent sources of information and advice.

Issues of representation were frequently raised in the early organizational planning and development of both BART and AC, issues that remain at the heart of the current organizational arrrangements. "Important people in Contra Costa County opposed AC district formation because they feared domination by downtown Oakland."^[39] The proposed AC district was generally well received in Contra Costa County, but business interests were more ambivalent:

The chairman of the County Board of Supervisors, H. L. Cummings, opposed the ACTD [Alameda-Contra Costa Transit District]. This development-oriented activist thought the East Bay district benefited only downtown Oakland; he also objected to the structure of representation on the board of directors, which, Cummings felt, favored Ala-

meda County. . . . The East Bay district plan was geared to taking over the Key System which did not serve central Contra Costa. His county's prospects for service in an East Bay district were very uncertain.^[40]

Business groups favored BART and opposed the formation of AC fearing that it would be dominated by Alameda County. "Moreover, these opponents feared the unlimited taxing power granted to the district."^[41] In fact, in 1959, after the formation of AC, Contra Costa seceded from AC, but in 1960 some of the cities that had received Key System service rejoined the district. Cummings "stood for BARTC's (BART Commission's] plan because it offered the promise of service for Contra Costa."^[42] Perhaps ironically, as implemented, BART provides rail service only to Richmond, which was already well served by the Key System and then AC. And it was not until 1980 that Richmond began receiving any direct service to San Francisco. But, at the time, "BARTC's plan offered a more attractive political and transit vehicle for Contra Costa's development aspirations."^[43]

Officials in San Francisco were concerned that enabling legislation for BART was so structured that the city would not receive services commensurate with its financial contributions.^[44] Peninsula speakers insisted that peninsula communities should not be forced to accept a regional rapid transit system without their explicit consent.^[45] "A councilman from the peninsula city of San Carlos told the hearings that his people questioned the benefits regional transit would bring to them," given the economic and demographic changes occurring in San Mateo County. They were no longer simply a bedroom community for San Francisco.^[46] Now, of course, the residents of San Mateo County have no voice on the BART board because the county withdrew from BART in 1962. And although they pay no taxes to subsidize bart's operations, they benefit from the presence of the BART Daly City station.

The disjuncture between those receiving services and those paying for them is evident in a report released in 1957 by the Commonwealth Club of San Francisco. It questioned the construction of a rail line from San Francisco to Marin that would make up some 16 percent of the overall construction costs, yet would serve only an estimated 4 percent of the population, and furthermore would make up only about 3 percent of the assessed valuation of the BART district.^[47]

Conversely, the southern section of BART's East Bay line was added to the original plan in order to make Alameda County's share of the construction proportional to its financial contribution.^[48] A similar circumstance has obtained in the scheduling and implementation of construction for the Washington Metro Rail system. In these examples, the independence of the local jurisdictions permitted them to exercise a kind of veto over plans that went against their interests, whereas in the case of Contra Costa County the fact of overlapping transit organizations gave the county interests a choice of plans to support.

The existence of multiple independent organizations in the same system not only provides more points of access to decision making by interested parties but also means that those organizations may themselves act as representatives of affected interests. The transit operators provide a focus for the organization of interests. The relations of the MCTD and Golden Gate are instructive on this point. MCTD was a paper organization that generated tax revenues and was eligible for MTC-administered funds. MCTD contracted with Golden Gate for local service in Marin County. An annual battle occurred between MCTD and Golden Gate for Transit Development Act funds allocated for use in Marin County. If MCTD got the money, it went into local service. If Golden Gate got the money, it went into commuter service. Without the existence of both organizations with overlapping interests, it seems probable that one or the other sets of interests would have been dealt out of the game. Each championed the interests of the groups it perceived to be its clientele. But MCTD went out of existence in 1984.

This contrasts sharply with the situation at Metro, at least within the District of Columbia, with a single planning entity: "Blacks in southeast Washington, for example, probably considered the rearrangement of their transit system unreasonable—and so it was, from their perspective and for their interests. Yet it made good sense for long distance commuters."^[49] The existence of MCTD in overlap with Golden Gate helped to ensure that local clientele were not ignored in favor of long-haul patrons: "A monopoly bureau is likely to tailor programs for a specific interest group, while overlooking other groups' interests." In terms of representation,

Whatever the cause of selective orientation, the political costs of monopoly bureaus in heterogenous task environments can be substantial.

To those not in the chosen clientele group, bureaucratic behavior may appear arbitrary and capricious. . . . Multiple bureaus, using personnel with different expertise, with different equipment, and with diverse programs will satisfy a broader range of persons.^[50]

Overlap of organizational jurisdictions promotes representation of diverse interests in two ways. As in the case of the BART-AC overlap, patrons are offered a choice, and through competition the quality of service is maintained or improved. As in the case of MCTD and Golden Gate, overlap resulted in the provision of services to competing interests. In the first instance one set of interests is better served by organizational service competition, whereas in the second divergent interests are served better by competition at the planning stage.

Organization size also enters into issues of representation and problems of access. The larger the organization, the more difficult it is to reach those individuals with sufficient discretion to make a decision. There is a basic tension between limiting the span of control within an organization and the number of levels through which a matter must pass before it can be acted upon.^[51] It stands to reason, therefore, that if the span of control is held roughly constant while the size of the organization increases, the number of layers in the organization hierarchy will have to increase. Simultaneously, in a single large organization, the number of individuals responsible for particular areas will be fewer than in several smaller organizations. After all, a principal reason for consolidation is to eliminate redundant administrative positions, thereby reducing the number of accessible decision points while increasing the number of layers.

Local control of government functions is decreased by larger organizations: "I have learned that the larger the government entity, the more difficult it is to communicate your needs and solve your problems. Integration of transit operators would make them more remote than they already are."^[52] From the standpoint of representation it is more useful to have six organizations with six boards of directors and six staffs to serve the Bay Area than a single consolidated entity or a consolidation of any subset of those six.

Ostrom, Tiebout, and Warren have referred to large organizations designed to deal with metropolitanwide problems as "gargantua." They contend that

gargantua, with its single dominant center of decision-making, is apt to become a victim of the complexity of its own hierarchical or bu-

reaucratic structure. Its complex channels of communication make its administration unresponsive to many of the more localized public interests in the community.^[53]

The existence of multiple independent organizations or multiple independent, overlapping organizations makes possible more points of access to the decision-making process. It increases the probability that through sheer numbers different interests will find someone congenial to their points of view. It increases the probability that these differing points of view will be forcefully represented and makes more probable that minority sectors will have some say over activities that affect their lives. However, multiorganization is no guarantee of responsiveness. The likelihood of response is simply improved. "The multiorganizational system [is] not behaviorally more responsive to clientele, but it provide[s] a richer set of options and more resistance to reducing that set."^[54]

Reliability

In the public administration an emphasis on efficiency is often closely allied with a focus on coordination. Although concern for efficiency is largely directed toward private organizations, public organizations have never escaped this orientation, especially in the post-Proposition 13 era. I understand that any given decision can be described as efficient

if it achieved the greatest possible results with given opportunity costs, or if it achieved a given level of results at the lowest possible opportunity cost.^[55]

By no means is concern for economic efficiency misplaced, but a singular worry about efficiency to the exclusion of other goals is at best problematic. In part, economic efficiency has gained a certain preeminence because of all possible values used to evaluate organizational performance, it is probably the most easily understood and measured.^[56]

Concern for economic efficiency is often closely related to movements to consolidate apparently messy organizational systems that possess varying degrees of competition and duplication and overlap.

However, autonomous organizations in the same organizational system, especially when they are characterized by overlap, provide benefits that would be absent from a system without those characteristics. It is not even clear that such a system is any less efficient than a consolidated arrangement.^[57] Obviously it is important to look beyond economic efficiency.

Landau has argued persuasively that the pursuit of reliability by organizations may well prove more beneficial in the long run than concern for efficiency. Instead of managing for success, Landau asserts that we are better off managing to protect against failure.^[58] Although failure may come from many sources, concern for reliability recognizes the essential fallibility of human organization, whatever its source. Where the provision of services is essential to the public welfare, concern for reliability is not a luxury but a necessity. Inexpensive operation of public transit or any other public services is of little utility if it is prone to unmitigated failures.^[59]

It is no less an error to assume that reliability can be obtained through the perfection of the parts of a system:

In public administration the standard policy for improving the performance characteristics of an agency has rested upon the classical axiom that the reliability and efficiency of an operating system, man or machine, is dependent on the reliability and efficiency of each of its parts, including linkages.^[60]

The theory of redundancy sets aside the "doctrine that ties the reliability of a system to the perfectibility of its parts and thereby approaches the pragmatics of a system in action more realistically."^[61] With the development of certain redundancies, it is possible to construct a highly reliable system from basically unreliable parts. The probability of system failure decreases exponentially with simple increases of redundancy. Redundancy becomes increasingly important as the interdependence of system components increases, because the failure of any one component will send shocks rippling through the system. The 1983 strike at Chrysler's Ohio door subassembly plant shut down six other assembly plants at an estimated loss of $75 million, all because there was no reserve inventory of doors, a tactic intentionally pursued by Chrysler management in an effort to improve efficiency. The organization had voluntarily stripped itself of its own redundancy.

In the detection of errors, overlapping competing organizations generate far more information than in situations characterized by monopoly organization, a fact duly taken into account in the unorthodox administrative style of Franklin Roosevelt. Once an error does occur, insofar as the organizations are independent but duplicate each other, one can provide emergency backup for the other.

In public transit, duplication of services by two agencies does not necessarily mean that they are ill-coordinated nor does it mean that waste is involved. It makes sense, however, to specify the failures redundancy is intended to protect against. In the Bay Area transit system there is the potential for both technical and organizational failures. Technical problems have to do with failures of equipment, including train breakdowns and tunnel emergencies for BART and Muni; problems with the Golden Gate and Oakland Bay bridges that would prevent traffic from crossing them, along with closures of other major arterials such as Highway 101; and weather or mechanical problems that would prevent Golden Gate's ferries from operating. Organizational failures include shutdowns due to labor slowdowns and strikes.

BART and AC transbay services are both organizationally and physically independent. BART's trains cross the bay in a tube resting on the bottom, while AC buses use the regular traffic lanes of the Bay Bridge. If both used the bridge, and the bridge was obstructed, the fact of two different modes with overlapping service would be useless. However, both AC and BART have been able to backup each other on numerous occasions.^[62] BART and Muni are able to provide backup for each other in a similar manner in San Francisco, although the fact of shared stations diminishes their physical independence. If the Golden Gate Bridge is forced to close, its ferries can transport a portion of the transit load, and vice versa.

Organizational failures have resulted from labor slowdowns and strikes. The Bay Area is characterized by multiple operators with different unions. Washington Metro, a unified operating authority, has a single union for both rail and bus modes. A comparison of the effectiveness of each system during labor strikes indicates the value of redundancy.

In 1979 and 1980, transit strikes in Los Angeles and New York crippled public transit. Despite strikes at BART in 1979 and AC in 1974 and 1977, for the most part people dependent on public transit

were still able to get around. The organizations are separate. No union local represents workers at more than one operator. In fact, workers at some operators are represented by more than one union. The Bay Area has several different unions, not simply locals of the same union. The contracts negotiated have different provisions and different renewal dates. Even if one union strikes a given operator, other unions at that operator have not been likely to go on strike, let alone unions at the other operators. For example, at Golden Gate bus drivers and ferry personnel are represented by different unions. When ferry workers struck in 1979, the buses continued to roll, providing necessary backup services.

Although Metro may have a capacity similar to the Bay Area for providing backup to protect against technical failures, it does not have the ability to deal with organizational failures caused by labor problems. Apparently the development of a single union occurred with the blessing of the Metro management, perhaps because of the greater simplicity of dealing with only one labor union. "Of course when a strike does occur, we can confidently predict that it will be more disruptive than those in the Bay Area because of the combined effect of nonredundant service and unified union."^[63] And so has been the case.

As we have seen, informal organization compensates for the weaknesses of an organizational system characterized by multiple, formally independent organizations, by providing a coordination mechanism to deal with the system's interdependence. Informal organization also compensates for the problems created by multiple overlapping organizations through personal trust and reciprocal relationships, thus helping to stabilize a redundant system. Other factors also tend to improve the stability of redundant systems:

a. Legal complications that make merger difficult and expensive—for example, when jurisdictions overlap only partially.

b. Strong support of individual organization clientele.

c. Sound financial condition of the organizations involved.^[64]

Bendor also concluded:

Redundancy is more stable, and therefore more practical if overlapping bureaus do not have a powerful superior authority close at

hand. . . . Though a few higher-ups may promote redundancy, I believe superiors more often reorganize duplication out of existence than they promote it. For this reason redundancy is probably more feasible among special districts than among regular departments, because districts are less frequently embedded in hierarchies.^[65]

Redundancy presents managers with a more demanding arrangement to manage. Conflicts and confusions over channels and lines of authority occur frequently. Also justifying a redundant system to a penurious public is increasingly difficult. Therefore, whatever assists a system of independent organizations to persist will also help to maintain the stability of redundancy, thereby promoting system reliability. By providing coordination mechanisms, informal organization indirectly helps to maintain redundancy by maintaining the conditions that facilitate it. Although a centralized system may support some redundancies, a decentralized system is far more likely to provide the conditions necessary for long-term support. One cannot imagine a single organization having the excess capacities of AC and BART for transbay service for very long. In fact, the elimination of that redundancy has been a primary argument made in favor of consolidation of the two organizations.

Conclusion

The argument made in this chapter differs significantly from those made in previous chapters. Earlier chapters examined the capacity of flat informal systems to provide effective coordination under certain conditions of interdependence. This chapter's concern turned to goals other than coordination. I suggested that where a mix of goals is sought, it is not unusual to achieve higher levels of some values at the expense of lesser amounts of other values. If we assume that under some conditions formal centralized systems provide better coordination than an informal system, is that additional increment of coordination achieved only at the cost of other goals, such as differentiation, representation, or reliability? An informal system of coordination, by permitting multiple independent organizations, under the conditions I have identified—complex operating environments, diverse interests, and potential technical and organizational fail-

ures—tends to promote these three goals more effectively than a formal centralized system, while simultaneously providing at least a satisfactory level of coordination. Ultimately, however, the question remains as to which tradeoffs among goals one wishes to make in any particular organizational design. As Simon has noted, such tradeoffs are a function of individual motives that lead people to select some goals rather than others as premises for their decisions.^[66]