These events and the subsequent discussion stemmed from an earnest look at a question reminiscent of Freud, “What do users want?” The following assertions have been made on behalf of users.

• Users want simplicity and immediacy. Users are accustomed to a single search box on an entry screen, without an obligation to categorize search keywords.
• Users want the search interface to be intuitive; this does not mean they want the response to their searches to be simple-minded. On the contrary, as Kautzman and Ryan have noted, “An intuitive interface is not by definition ‘dumbed down’ or anti-scholarly.”⁷
• Users benefit from prevention of dead-end searches, provision of enhancements like tables of contents and cover art, and navigation assistance that groups results and enables jumps to “more like this.”
• Users want one system to search, to cover a wide information universe. They frequently do not know about the separate silos that exist and what is in each; even when they do, having to search them one at a time is unwelcome.
• They expect immediate access to full-text online resources wherever possible. For other materials, they want a full range of fulfillment options, not restricted to those services the library controls.
• They appreciate the opportunities to annotate, review, and “tag” resources, as well as share them with others. Users like results sorted by relevance, a key component of which is how others like themselves have used the resources.
• Instead of having to come to the library, physically or virtually, they would appreciate the above services being delivered to where they are—be it an institutional portal, course-management system, commercial search engine, and other mechanism.

Libraries determined to reform their operations and modernize themselves in response to these demands and expectations have a couple of choices of how to regard metadata. One view sees library metadata as overkill in light of what users of commercial search engines can be satisfied with. Any search functionality or data element not enjoying significant usage is a prime candidate for elimination. The granularity of library metadata probably is responsible for presenting too many searching choices on online catalog screens. With the arrival of e-resources in library workloads, the current methods of resource description appear less and less scalable. The availability of full-text online tempts the conclusion that summary-level metadata is unnecessary. The large expense going into a behind-the-scenes cataloging operation represents an opportunity cost for libraries, in terms of what they could be doing for more directly beneficial user services. If users are abandoning the catalog for Amazon and Google, it is an indictment of the outdated practices and products of cataloging activity.

Another view sees metadata as the library’s unique “value add” in this information age. In Web 2.0 terms, metadata gives those who use it a competitive advantage in the information marketplace. The richness of the metadata creates opportunities for providing a wide variety of library services. Bibliographic records support more than searching catalog inventory.⁸ In a Web environment, they can provide a rich set of possibilities for browsing relationships that a title has with related material. Reasonably consistent data can support precise report writing against a file of bibliographic records, along parameters not offered in live catalog searching.

A solution to poor-quality experiences using existing online catalogs is separating the front-end interface from the valuable back-end data store. An even greater means of justifying and recovering the cost of creating metadata is capitalizing on the expanding markets for it. In the future, online catalogs will be just one of the places that library metadata goes.

At the June 2006 Technical Services Administrators of Large Research Libraries Discussion Group meeting in New Orleans, an update on the efforts to make library data available to commercial search engines elicited the observation that receipt of the initial metadata increased the appetite for more of it. At an August 2006 Digital Library Federation workshop on implementation of the Open Archives Initiative, presenters confirmed the trend is moving away from seeking lowest common denominator, unqualified Dublin Core toward seeking whatever richer schema a data provider happens to have.⁹ Another insight from the same workshop is that metadata’s role does not end once a described resource has been discovered and delivered. Metadata also supports the use of the resource, enabling a user to obtain a citation of it, as well as to find resources related to it. For nontextual objects, the metadata explicates what the user has found.

The benefits of library classification systems seem to be steadily growing over time. Classification can provide a unified browse of library resources on similar subjects—those on open shelves, in use, in remote storage, or in digital format. High-level subject browse categories can be created for subsets of the collection such as e-resources, by mapping from existing classification data. When a set of subject browse categories needs to change, particularly being made more or less granular, this can be achieved by remapping from the classification, versus starting from scratch. Powerful analysis of collections and assessments for accreditation are supported by classification available in cataloging data.

While this future for library metadata looks very bright, there are some changes we will need to make. If we wish to sustain the level of metadata we are accustomed to now, we need to be willing to change a long-standing cataloging model. Libraries have historically approached new-title cataloging as starting with the existing bibliographic utility record and custom editing a copy of it for the local file. Revisions to existing cataloging records often are made directly in the local file. Some of the original cataloging energy involved was channeled back to the bibliographic utility’s record, through the Program for Cooperative Cataloging, the OCLC Enhance program, and others.

At a regional or national level, we need to be willing to start working in a communal file, where the work on bibliographic records need only be done once. Imagine the savings and efficiency that could be gained if serial title changes, authority work to maintain headings, and other upgrades to data could be done once and automatically shared by others. Think about the prospects of having additional resources for placing more material under better bibliographic control, gearing more of our cataloging toward more original and unique work, and being able to serve a wider variety of research needs. We have to be willing to cooperate in using each other’s records (largely) as-is, to agree on standards, to trust each other.

Libraries should not always feel that they have to be the ones to provide the metadata. If there is similarity in the uses made of bibliographic descriptions by publishers, vendors, and libraries for their inventories, and if there is significant overlap among needed data elements, then it makes sense to pursue a single metadata creation effort whose results we all can use. A good example of this is the recent arrangement between North American libraries and Casalini Libri to share a national-level bibliographic record.

The model of relying on national library provision of high-quality cataloging records may be shifting to a wiki. We will need to rely on ourselves, together with strategic partnerships with others, for sustaining a rich underlying metadata that can guide users to many bibliographic services. Users––and fulfillment of their wide range of needs––depend on good metadata for finding, identifying, selecting, and obtaining resources.

Call this the vision of the One Good Record. The least scalable aspects of metadata are the local variation and the duplication of effort.

The more a library goes it alone in metadata creation policies and efforts, the sooner it will run up against the limits of what it can sustain and the sooner it will feel the need to retrench the content of the record.