Rethinking Public Services at Harvard College Library: A Case Study of Coordinated Decentralization

Caroline M. Kent

Harvard has not been immune to the enormous environmental changes that have taken place in the academic library world in recent years. Although its finances are more stable and much more generous than those of most other academic libraries, the Harvard College Library (HCL) has also had to contend with escalating materials and benefits costs; the necessity of networking and its accompanying costs; shrinking pools of available staff for new initiatives; a huge, ever-changing staff training crisis; and all of the problems accompanying a large organization’s need to change its methods and staff quickly.

Finding solutions to such problems requires that different parts and levels of any organization work together effectively. In an institution that prides itself on independent action, the achievement of consensus (or even respect for majority rule) can seem almost impossible to achieve. This chapter argues that HCL’s experience in restructuring its highly decentralized administrative model may be a microcosm of how academic libraries will face the problem of collaboration on a national scale.

Harvard’s Tradition of “Organized Anarchy”

Of all American academic libraries, Harvard’s system is probably the best qualified to represent a model of “decentralization.” Sidney Verba, who is both a well-known political scientist and the Carl H. Pforzheimer University Professor and director of the library, often remarks that the Harvard libraries’ organization resembles the Southern Confederacy of the last century. In our worst moments, however, we may more likely take after the German feudal states! Each of the faculties (e.g., medicine, law, or divinity) has its own library, and the directors of those libraries answer to the deans of their schools, not to the university library director. The university library director chairs the Harvard University Library Council, which comprises all of the directors of the libraries of the different faculties and looks for ways to coordinate the various libraries’ activities. Agreements are struck, “treaties” are signed, and collaboration and parallel activities engaged in—sometimes. When cooperation is not achieved, everyone smiles politely and tries again for collaboration at another time.

In all of Harvard, the Faculty of Arts and Sciences (FAS) presents the most distinct model of “organized anarchy.” FAS’s largest library unit, HCL, includes the following eleven major libraries: Widener (graduate social sciences and humanities), Lamont (undergraduate social sciences and humanities), Hilles (undergraduate social sciences and humanities), Cabot (undergraduate sciences, and graduate math and statistics), Kummul (geological sciences), Fine Arts, Music, Houghton (rare books and manuscripts), Tozzer (anthropology), Littauer (political science and economics), and Harvard-Yenching (Far Eastern languages and literatures).

HCL contains a major portion of the university’s libraries’ holdings—seven to eight million of the almost thirteen million volumes total. Despite its size, HCL does not control all the FAS libraries: Under the direct jurisdiction of individual FAS departments, museums, and research centers are approximately forty more libraries, ranging from a single, unstaffed room of books to the Ernst Mayr Library of the Museum of Comparative Zoology with a staff of eleven and a collection of more than 250,000 pieces. These forty-some FAS libraries have tacit representation on the University Library Council by the director of HCL, but their activities, budgets, and collections are independent. It would be logical to assume that because all FAS libraries nominally report to the director of HCL, they would act in an orderly fashion. Sometimes that is the case, other times not.

“Every Tub Stands on Its Own Bottom”

HCL’s decentralization is not accidental: It is a true reflection of the organization of Harvard University as a whole. Harvard’s cultural climate of independence has a way of asserting itself everywhere. Since early in the nineteenth century, one of the traditional and covering principles of Harvard has been the phrase “every tub stands on its own bottom.” This means that each administrative or academic “tub” (unit) must generate its own income and is responsible for paying its own expenses (the bottom line); or, as economic historian Seymour Harris defined it: “No college is responsible for the solvency of another, nor is the university.”1 (The use of this phrase is so ubiquitous that most Harvard administrators will recognize its shortened form of “ETOB” on memos.) This culture inevitably reproduces itself as it attracts like-minded librarians, administrators, and scholars, all of whom carry on the tradition. Bureaucratic-minded administrators sometimes find Harvard’s environment incredibly frustrating and are confounded by how decision making often slips away from them. Programmatic progress becomes dependent on the power of personalities, the quality of minds, and, when available, the goodwill of individuals.

At the same time, the decentralization and independence of the Harvard libraries have resulted in great richness for our users. If a department or center felt ill served by the existing organizations, or if faculty and researchers simply wanted their resources very close to hand, they could start a library and hire a librarian! The department then had to support that facility and staff financially but, traditionally, no effort was made to prevent the creation of such collections. This, of course, made for certain duplications in the collections. But Harvard’s wealth rendered that a relatively slight concern except for the most expensive of titles. The result for the experienced user was easy availability of their most desired, specialized titles. This tradition continues with Widener Library still getting requests from groups of scholars for new subcollections (intellectual “boutiques”) in buildings where they will be most readily available to scholars and students. So, the Sanskrit scholars and the paleographers might want their materials in restricted-access rooms, or the Celtic scholars might want their periodicals shelved in their seminar room.

Such “organized anarchy” has worked to the advantage of the user except in one salient area—library public services for the collections. Here’s a true-to-life example of what a Harvard library user can experience: A senior undergraduate concentrator in history is in the midst of writing an honors thesis. The topic is interdisciplinary, so the student must use the collections of five of the Harvard libraries frequently. She must remember from which of the five she checked out the 150 books in her room because she must make sure to renew all of them on different schedules. Some libraries allow the student to call book renewals in, but Widener (which holds the major history collections) requires her to haul her 100 or so books back every month to renew them. She has to carry a different photocopy card for each library she uses, but in her case that’s only five, so she does not feel too bad (she knows people who carry ten). She must remember, that if she needs reference help in the evening, Widener reference is not open after 7:00, although she can usually find someone in the undergraduate libraries to help her—provided her question is not related to Widener’s collection.

HCL has traditionally had abnormally low public services staffing. In 1989, when I became head of research services in Widener—then a collection of 3.4 million volumes—there were five full-time reference librarians (a number more fit for a four-year liberal arts college library with fewer than a million volumes). Circulation staffs fared somewhat better (although Widener still has no evening or weekend supervisor on duty; students are assigned this role). And anyone who used Harvard interlibrary loan (ILL) units will attest to how slow they had once been—the result of low staffing levels and little administrative support.

Also, the various service units often had very little to do with each other. Following Harvard’s tradition of independence, every circulation department had its own fine structure, loan periods, and so forth. Reference desks existed in some libraries, but not in others. Some reference departments had active research instruction programs, whereas others would not even engage in building tours. As for ILL, some libraries did it and some did not—like circulation, every library was free to develop its own charge and loan policies.

In a library where collections were valued above all other considerations, confusion for users was regarded as an acceptable situation. Even public services personnel, when faced with the threat of a lower materials budget, would often defend (albeit resentfully) the lack of expenditures in service areas. In many U.S. academic libraries, reference librarians enjoy a certain level of institutional prestige and respect; at Harvard, they traditionally have been the least-recognized group of librarians. Staff grumbled about their lack of standing and support but simultaneously enjoyed enormous freedom of action in their independent “tubs.” But all of the tubs, universitywide, began to sink with the advent of shared automated systems and networking.

Shooting Fish in a Barrel, or Sinking the Tubs

A presumption of total organizational and fiscal independence and decentralization can only work at a university when there are no development and supportive costs that the community must share. Networking and networked resources, which require collaborative efforts and payments, made Harvard’s administrative model vulnerable to failure.

Ironically, the general problem that Harvard’s principle of ETOB represents—how to distribute both costs and income to the individual units that incur those costs and generate that income—had inspired in the 1980s a theory of financial organization in higher education called responsibility center budgeting (RCB). Now, in the 1990s, Harvard had to discover how ETOB (or what had been adapted as RCB at some other academic institutions) fits into the networked environment.

RCB links immediate program management with meaningful authority over resources as a way to shift incentives for cost-effective performance down the organizational hierarchy. But when RCB is implemented at an institution, it is important to exert enough central control so that unwarranted duplication and misuse of fiscal control do not occur. If this method of budgeting is not directed carefully, “it can result in dissension, discord, distortion, dissonance, disaster and doom.” 2

“Doom” is a strong word, but being forced to share can certainly lead to “discord and dissonance” at Harvard. Responsibility center budgeting was not “implemented” at Harvard. Instead, it grew out of the institution’s history and unwittingly inspired a general model adapted at other academic institutions. Certainly, there was little in Harvard’s past to facilitate the kind of central support necessary for effective development of a computer network. Poorer units, which traditionally had great difficulty in supporting themselves, were forced to choose between either not automating or begging for central support, even though they knew that support would be translated into political as well as economic debt.

Some units risked sinking into debt to the institution for the sake of being networked, whereas other units remained unnetworked until very recently. Moreover, there was no effective central networking group; instead, each unit undertook the effort on its own. Academic computing (the Office of Information Technology), which had maintained the old mainframe systems and had tried to implement centralized networking, became an easy target for administrative takeover. As Harvard’s networking situation became increasingly confusing, and as the role of the network in the university’s enterprise grew in importance, “who owns the fiber” emerged as a major institutional issue: who installs, who supports and, most important, who pays.

This universitywide struggle to be networked presented fundamental problems for the libraries. Technical support was clearly needed for computing, but of the scores of libraries on campus—some with virtually no organizational infrastructure—which ones had the “right” to own the fiber or, conversely, an obligation to pay for it? Even more significant was how the fight for fiber affected the delivery of services. Implementing campus networks was extremely difficult when your fiber was “owned” by multiple administrative units. If you needed fiber pulled across a road to a library, it could take years to figure out who was financially responsible to do that. One departmental library remained unnetworked for two years: The fiber came to the wall of the library, but the fiber was “owned” by another academic department that had no interest in sharing it with the library.

Public services personnel were pulled into this struggle whenever they sought to find the right help and support arrangements for the users of the libraries, whose questions ranged from “How do I get access to HOLLIS at home?” and “Why don’t you have the MLA on a network?” to “Why isn’t my building networked?” Reference librarians found themselves lending their e-mail accounts to frustrated users, hooking up modems for them, and giving them advice on software purchases.

The university libraries, despite their history of decentralization, were making serious headway in learning to cooperate for the implementation of HOLLIS (the online library system). Starting in the early 1980s, the libraries had begun to develop distributed funding arrangements for its support. In that framework, very disparate libraries came to various agreements on HOLLIS’s development. As time passed, the learning curve of cooperation got trickier when other databases were loaded either in the HOLLIS environment or made accessible through HOLLIS Plus, the system’s networked gateway. Groups of Harvard libraries had to agree on a funding model based on use: HCL might pay 40 percent, the law school 20 percent, the Kennedy School of Government 10 percent, and so forth. Although there were times when the libraries were cranky with each other over the time it took to develop distributed funding models, the effort was succeeding. No longer was each library “tub” standing on its own bottom. The libraries were developing their cooperative framework into a model that the rest of the university would have been well served to pursue.

“What to Do, What to Do” or, “How to Avoid Hand-Wringing”

Why did this essay start with the traditions and cultures of Harvard University as they surround its libraries? The answer lies in a basic idea, one that applies to any library embedded in a university: When instigating change, respecting an institution’s culture results in forward movement; not respecting it, in stalemate or decay. Restructuring an organization, therefore, depends on understanding what your institutional culture and values are, and figuring out where you intend to go and how you can get there without violating revered norms. In other words, any library wanting to determine its own future must engage in some sort of formal planning with an eye on local traditions. Planning is important in any institution. For Harvard in the age of the network, it is critical that planning occur, and that the deliberation process be inclusive and widespread. All players must be engaged and invested in the results in order for the institution even to hope for cooperation. That is not a reflection on staff members’ willingness to cooperate but, rather, on certain tensions between operating units which represent the independent desires and demands of their respective faculties and funding agencies.

There were actually three different planning efforts that went into redefining public services at HCL. First, an informal effort was begun with the appointment of a new associate college librarian for public services in 1988. Second, a formal, librarywide strategic planning process took place at the request of the newly appointed director of HCL in 1991. And third, a formal planning effort specifically for public services became part of a broader move toward a “gateway” library in 1993–1994.

Beginnings of Change in Public Services

In 1988, Lawrence Dowler was appointed associate college librarian for public services (of HCL). Recognizing the increasing importance of service in the upcoming decade, he immediately sought ways to reorganize public services in Widener and throughout HCL. The changes made within Widener were structural in nature: Seven former divisions were reorganized into three large departments: Research Services, Access Services, and Government Documents and Non-Book Formats.3 Within each of those units, internal reorganizations also took place.

Although having direct managerial control over the public services units of Widener, the associate college librarians (of HCL) have an ambiguous relationship with the outlying units. Associate college librarians oversee their respective areas throughout HCL’s libraries, but individual departments in the libraries answer to the heads of those same libraries, who in turn answer to the director of HCL. Inevitably, the associate college librarians are left with an arsenal of “c” words: communicate, collaborate, cooperate, coordinate.

Such managerial ambiguity led to the formation of a new coordinating committee in the spring of 1988, the Public Services Issues Committee.4 Although the committee did not include every HCL library, it was fairly representative. After an early agenda-setting meeting, the following concerns of the various public services units emerged: poor communication; lack of information; problems with particular service areas (reserves, ILL, security); conflicting policies; and problems with special-format materials, particularly with regard to a consistent technical approach for the delivery of technically dissimilar bibliographic databases.

By 1991—on the eve of formal, systemwide strategic planning—the committee had evolved to a point where greater numbers of people needed to be engaged. Consequently, it expanded into three new structures: the Public Services Advisory Committee, which was the services administrative advisory group to the associate college librarian for public services; the Council for Reference and Research Instruction; and the Access Services Council. This concurrent internal reworking of public services units with the changes taking place in HCL’s central administration was fortuitous. Public services staff since 1988 had been rethinking their roles within the library and with their users. As no other single library group, they were ready for, and committed to, broad change.

Strategic Planning at the Harvard College Library

In 1990, shortly after his appointment as director of HCL, Richard De Gennaro engaged the library in a formal, inclusive strategic planning process.5 It included task forces on services, collections, intellectual access, space planning, and staff and organizational development. From their deliberations came Commitment to Renewal: A Strategic Plan for the Harvard College Library, in 1993.

Because the reorganization of public services was already well under way when this formal planning effort was announced, HCL’s public services staffs were particularly enthusiastic. They did carry, however, a heavy history of being underfunded and understaffed (at the time of strategic planning, there were only twenty professional staff members spread over the eleven HCL public services units). This history left those staff members both willing to cooperate and collaborate with each other but unsure of their relationship to higher administration. But the following strategic goal was set: “Help students and faculty to achieve maximum benefit from library services and resources by making use of the library more convenient and efficient.” Such a statement would be self-evident at many universities; at Harvard, it was radical.

Much to the shock of public services staff, the administrators listened to them. Over the next several years, the number of professional positions associated with public services units rose from twenty systemwide to thirty. New initiatives, often arising from the public services councils, received fair consideration. In addition to those new positions, support was received for modernizing book retrievals from remote storage, for establishing a document delivery system, for building an electronic classroom, and for creating several coordinator positions below the associate college librarian level.

The Gateway Concept

Early in 1992, following a recommendation in the original strategic plan, HCL engaged in a planning effort that was devoted to public services. In response to the needs for renovating the Lamont Library and for addressing, more broadly, the effects of technology and networked resources on public services units in HCL, conceptual development of a “gateway” library began:

Gateway is a metaphor for access to knowledge and evokes the image of crossing a threshold and entering a dramatically expanding world of information and learning; the library, as gateway, is the means by which students and faculty will locate and use this information. The gateway we envision is the constellation of services, the organization required for providing these services, and the spaces dedicated to student learning.6

The working groups in this public services planning process included more than a hundred staff members throughout HCL. They were organized into several task forces and working groups. Some task forces had broad charges; other working groups, created to address particular problems, had very specific charges. All of them produced a large number of recommendations, such as an ongoing program of user needs assessment, an improved document delivery facility, more widespread technological support, and a larger service staff.7

Concurrent with the gateway planning process was a universitywide effort to create a statement of organizational or community values. Articulating such a statement can be an important institutional step if it raises tacit assumptions to a conscious level where they can be discussed and validated. This was undertaken by a group known as the Steering Committee on Staff and Organizational Development, in 1993–1994. Although the effort to get any community engaged in defining its values can result in cynicism, that committee—by thoroughly involving a substantial majority of the community—managed to produce a statement reflective of HCL’s explicit and implicit values. Seemingly obvious phrases, such as “serve our community” and “provide access to a broad array of information sources,” all rather novel for Harvard College, were essential to validate the public services staffs’ interest in developing plans for the future.8

The gateway planning, which was wholly focused on library services to users, and the statement of community values were, together, the final validation of public services in HCL. Reference, research, access, and interlibrary loan staff could finally feel that those issues they had always regarded as being so important were, indeed, widely regarded as central to HCL’s mission.

The Future

Is there something in Harvard’s struggles that has some utility or application to other libraries? It is undisputed that Harvard has the largest number of noncentralized libraries among U.S. universities. And no library administrator reading this has any intention of developing such an extremely decentralized administrative model! Most libraries have regarded service as central to its mission for decades; Harvard is a Johnny-come-lately in that respect. Can our peculiar history, therefore, have relevance for any other library system?

I would argue that it does, for an important reason: With the increasing prevalence of networking, the greater ease with which we distribute remote resources, and the pressures in many educational systems to engage in remote teaching, HCL’s emerging model for planning and collaborating among disparate players may be a microcosm of how academic libraries will have to deal with each other on a national scale. Collaboration, team building, elimination of unnecessary bureaucracy, respect for institutional culture—these are all key processes for restructuring academic organizations.

In saying all that, however, I do not mean to imply that HCL has solved such problems to anyone’s satisfaction. Still, we have clearly demonstrated that continual attention to our current position relative to where we intend to be (that is, strategic planning) is helping us cope with the lag of organizational development behind technological change. Do not make the mistake of feeling too much self-satisfaction over our struggles; you, or the next generation of library thinkers, may well face essentially the same struggles of how to get along in the networked environment.


  1. Seymour E. Harris, Economics of Harvard (New York: McGraw-Hill, 1970), 226.
  2. Edward L. Whalen, Responsibility Center Budgeting: An Approach to Decentralized Management for Institutions of Higher Education (Bloomington: Indiana Univ. Pr., 1991), 156.
  3. In 1995, a fourth department, Interlibrary Loan and Document Delivery, was created.
  4. Committee participation at Harvard is very important (although often frustrating). The solution to many problems can only come by putting all the right players in a room, closing the door, and hoping that the conclusions and agreements arrived at will actually stick after the individuals leave the room.
  5. The strategic planning process has been well described by Susan Lee, “Organizational Change in the Harvard College Library,” Journal of Academic Librarianship 19 (Sept. 1993): 225–30.
  6. Lawrence Dowler, Gateways to Knowledge: A New Direction for the Harvard College Library (Cambridge: Harvard College Library, 1993), 4.
  7. Harvard College Library, Gateway Planning Committee, Final Report (Cambridge: Harvard College Library, 1994), 3–4.
  8. Harvard College Library, Values Statement (Cambridge: Harvard College Library, 1995).