Principles of Selection for Electronic Resources.

ABSTRACT

THE AVAILABILITY IN ELECTRONIC FORMAT OF scholarly and scientific literatures and other forms of information relevant to the needs of library users has profoundly altered the challenges faced by collection managers. Although the traditional goals of achieving quality, relevance, and balance at a fair price still animate most collection-development efforts, judgments about these attributes of resources have become more ambiguous. The traditional standards have also been joined by new and highly important criteria which include the definition of the allowable user group and the purposes for which use will be permitted, multi-faceted concerns about the functionality of resources, and concerns about the availability of permanent archives. Drawing heavily on the ideas of the multi-library consortia, which have grown up partly in response to the advent of electronic resources, librarians have devised new criteria and means of assessing resources against them so that cost-effective acquisitions can be made in the new marketplace.

INTRODUCTION

It is a truism that academic libraries at the end of the twentieth century are caught between the demands of a traditional print-bound world of priceless resources built up over generations by their predecessors and those of "a new (electronic) world being born."

On the one hand, libraries struggle to find shelf space for burgeoning print collections, to slow the damage to highly acidic print collections being eroded by environmental pollutants--light, rough handling, and fluctuations of heat and humidity--and to maintain serial subscription lists which have been subjected to double-digit inflation nearly every year. On the other hand, libraries rush to meet the growing expectations of their highly computer-literate clientele.

Someplace along the way we seem to have leap-frogged from a recent past when the library tried to persuade skeptical users of the potential of electronic resources, to a present in which every user is an accomplished Web-surfer certain that every kind of information must be available and could be delivered to her library if only the staff understood. The days already seem distant when one heard about strategies to marshal grants, new money, or funds scratched together from small savings here and there to pay for electronic resources as an add-on. Every academic library of any size can now point to a significant conversion from print to e-resources, and hardly anyone bothers with the fiction that the latter has not come, to some degree, at the expense of the former.

Does this mean that everything you ever knew is wrong? Are the traditional standards of selection moot and, if they are not, how do they share space with the criteria of evaluation and selection specific to electronic media? What are the macro and micro considerations that bear on selection of electronic resources, and how do they relate to traditional criteria? These are the questions with which this essay will wrestle. Most of its examples will be taken from the world of academic librarianship, which has been most aggressive in its movement toward electronic access and in which the author dwells.

TRADITIONAL CRITERIA VERSUS NEW STANDARDS

The chief responsibility of a collection manager is to bring together a grounded understanding of her community and its information needs with a sophisticated and informed understanding of the publications universe. Decisions are taken on both a macro level (how large an approval plan should we have? with which publishers?) and a micro level (is this book appropriate? worth its price? and likely to be used?) to achieve balanced and affordable collections serving the main needs of the community at a variety of levels. The introductory statement of Virginia Tech's collection development policy (http://www.lib.vt.edu/info/colldev/ coll_dev_policies/GOALS.html) could probably represent the aspirations of most academic libraries:

   Collection development in the Virginia Tech University Libraries serves
   several purposes. Much the most significant of these is to satisfy the
   university's current needs for information resources in any format which
   will support its primary missions of teaching, research, and service. Our
   collection-building efforts reflect as nearly as possible the programmatic
   goals of Virginia Tech.

      Other goals shaping our collection development efforts are to build
   collections which will support in at least a basic way future university
   programs or areas of specialty; to furnish some basic support for the needs
   of the university's non-academic units; to provide some materials in nearly
   all areas of knowledge partly as a basis for users' self-education; and to
   serve as an information resource for other, primarily in-state, libraries
   with whom we enjoy partnerships.

   Within the context of these goals, the two stars which guide our efforts
   are the academic relevance and the quality of the materials we seek to add
   to our collections.

The purpose of the library will always be to support the needs of its community with the most relevant highest quality information resources, and no change in format will alter this. With electronic as with print resources, quality, level, and relevance must remain primary. Judgments about how well resources measure up to these standards will sometimes be easier in the electronic media but more often harder. Trial subscriptions, which can be shared with a broader audience than sample print issues, and which really have no analogue for books other than the cumbersome returns process within approval plans, make it easier to judge materials before committing to purchase.

Judgments are harder to reach in other respects. There is as yet no analogue to ISI's Journal Citation Reports which tell the potential purchaser how widely cited an electronic publication is. Fewer electronic publishers have the kinds of long-established reputations which librarians considering print materials routinely take into account. Content can also be hard to judge, because "how much" is hard to weigh in cyberspace, because what was here one month may be gone the next, and because subtle deletions such as the absence from scholarly journals of news snippets, errata, ads, or letters are not always easy to detect.

Much of the bibliographic apparatus that associated articles with journals and journals with publishers of known reputation has disappeared or become indistinct. It is much harder to evaluate an electronic resource that pools millions of articles into a homogeneous mass most frequently searched by subject than to evaluate a journal containing articles of relatively uniform and known quality.

Some of our traditional standards take on new meaning in the electronic arena. Currency can come to mean, not the latest week or month as with a magazine or loose leaf, but whether a resource is updated every fifteen minutes in the case of news or stock quotes or daily in the case of journal collections. The degree to which resources may be shared also has new meaning. It is wonderful that the same information can be used by several people at once, but it is also possible that some potential users are disenfranchised by technological incompatibilities or that restrictions on interlibrary lending may be more severe than with print.

Although the basic principles of collection development apply and should predominate in the selection of resources in electronic format, they can no longer have the stage to themselves. Instead of taking the original meaning of the expression "the lion's share" (which meant everything, or the portion one would reasonably expect to surrender to a lion), traditional criteria have shrunk to having the modern meaning of the phrase. This is simply because if some criteria grow in a decision-making process, others must shrink. In the print world, relevance, quality, and level in proportion to cost could be everything because realistically no one used such terrible paper as to make much marginal difference, and the best publishers used the best paper anyway. But in the electronic arena, it is perfectly possible to find oneself considering a resource that is programmatically relevant and intellectually impeccable but which one rejects because it is not accessible to Mac users or to off-campus users, or because it is down two days a week or insufferably slow at 3:00 P.M.

In writing this essay, I have profited greatly from the "Principles for CSU Acquisition of Electronic Information Resources" promulgated by the California State University (http://www.co.calstate.edu/irt/seir/ EIR.prin.html). This document is so comprehensive in its enumeration of the relevant considerations, and proceeds through them so logically, that it would seem foolish to try to reinvent it. Instead, I have chosen to structure the essay in the sequence that the principles follow and to end each section of the essay with a reproduction of the matching part of the principles.

Readers are referred with equal enthusiasm to the invaluable "Statement of Current Perspectives and Preferred Practices for the Selection and Purchase of Electronic Information" (http://www.library.yale.edu/ consortia/statement.html) issued by the ICOLC. This gives a cogent discussion of the criteria that should govern decisions about the acquisition of electronic information resources.

Since this concludes the introduction and remarks about the relationship of selection for electronic resources to the larger collection-building enterprise, sections I and II of the California Principles follow.

PRINCIPLES FOR CSU ACQUISITION OF ELECTRONIC INFORMATION RESOURCES

I. Introduction

The following are provided as a guide to the libraries of the California State University in developing and reviewing proposals and in negotiating contracts with providers of information in electronic formats. Many of the issues in the emerging electronic age will require discussion, experimentation, and collaboration on an ongoing basis.

II. Collection Development

A. Collection development policies and criteria should be applied consistently across formats including electronic resources.

B. Principal considerations include:
Establishing a rationale for the acquisition of each resource
Meeting faculty and student information needs
Providing access to electronic resources and integrating them
into library programs

C. Balance should be sought among:
Disciplines
Instructional and research resources
Different needs of each campus

D. Priority should be given to those electronic resources which
offer:
Integrity of the database
Economies of scale
Benefit to the greatest number of users
Timely availability
Extensive content
Increased functionality
Enhanced access to remote users
Improved resource sharing
Ease of archiving and replacing

E. The CSU libraries should be able to specify the content and the format for databases to be acquired. For example, the CSU should be able to supply the vendor of a full text periodicals database with a list of titles for inclusion. Selection decisions should not be compromised by provider-defined linkages between print and electronic products of the same version.

F. An electronic resource should have sufficient content to evaluate its usefulness and to justify its selection.

G. Acquisitions should be compatible with the goals of the CSU Library Strategic Plan and the Unified Information Access System (UIAS) project.

PRICING

Millions of words have been written about the crisis in serials pricing, the necessity for the academy to recapture control of what it produces, and the hope that electronic publications will allow a fresh opportunity for this recapturing, while introducing cost savings for both commercial and nonprofit publishers alike. For all the justified resentment and worry about pricing, the problem in the print world has not been that pricing is difficult to understand, but that distasteful facts are understood all too well.

It is much harder for collection managers to assess the prices they are likely to pay for electronic resources than for print. With print, you knew what something cost. You bought it or you didn't. If there was more demand than could be met by one copy, or by having a subscription only in one branch, you bought a second copy.

With electronic resources, there is variation in exactly what one might be buying and variation in the basis for computing charges. The resource itself is much more likely than in the print world to come in a modular manner, with optional extras such as the substantial fees one pays for a fuller backset of historical stock quotes on Dow Jones.

Costs may vary by the allowable number of simultaneous users, generally based on a sliding scale so that each successive increase in the number of users costs less. Costs may also vary according to how the user community is defined, with access to remote campuses or to students not studying on any campus generating surcharges. Frequently the size of the user community, measured in FTE students or the sum of students, faculty, and staff, is used as the basis for pricing. Sometimes only the relevant subset of the population, such as the number of students and faculty in engineering, is used.

For some resources, costs are based simply on measured use, much as a long distance phone bill is calculated. We might argue that all other bases for costing are at bottom a surrogate for anticipated usage, which serves as a measure of the value of the resource to the user community, and that it would be easier and more equitable if all library pricing were based on observed usage. The problem with this approach is that library collection managers must be able to estimate and control their costs in advance and can tolerate unpredictable costs only in fairly confinable subsections of their budgets.

Even when costs are based on criteria that are predictable in the short run, collection managers find it difficult to estimate the real magnitude of the commitments they have made. In the print world, resources are actually purchased and owned. Electronic media, however, are often leased. Collection managers' midnight worries include the fear that publishers have priced electronic resources at artificial lows in order to encourage migration that will raise the pain of cancellation and thus enable large future increases. Paradoxically, the same collection managers regret the continued tendency of many publishers and aggregators to offer electronic publications at moderate cost only to those libraries which retain their print subscriptions. It is unlikely that this pricing practice, which forces libraries to decide whether to discard print materials that duplicate electronic content or to devote labor, binding, and storage costs to them, will continue in a mature electronic marketplace.

A final issue complicating pricing concerns is the increasing tendency of library collection managers to act from more than a local perspective. The shift to electronic media is probably the single most significant cause of the dramatic growth in the number and power of library consortia. Whereas interlibrary cooperation in building print collections required cumbersome and problematic efforts to build complementary collections, it is now relatively easy and increasingly frequent for libraries to join together in purchasing access to electronic resources that are then functionally identical from library to library. Because publishers are not printing and mailing multiple copies and need generate only a single invoice to serve widely scattered customers, significant cost savings are possible.

Besides offering sometimes remarkable reductions in prices, consortial buying also affects the basis for pricing. For example, VIVA (the Virtual Library of Virginia) has refused, as a matter of policy, to acquire materials that would not be made available to all users. Yet the inclusion of an enormous community college population would generate untenable costs for many resources whose pricing has been based on FTE. Many VIVA vendors have resolved this dilemma by including the community colleges for free or nearly so. Anyone experienced in library consortia can tell that both the libraries and the vendors are sorting through difficult and ambiguous times, slowly working out standards while "making it up as we go along."

Most consortial purchases are made by state consortia. Most of these link all state-supported academic libraries, but several others incorporate private academics or include public and school libraries. Larger consortia also play an important role. In a remarkable precedent, SOLINET, a regional added-value broker of OCLC services, succeeded in negotiating prices for access to Lexis-Nexis's Academic Universe, giving libraries anywhere in the country entry into an enormous buying pool enjoying large volume discounts. The library community has even developed the International Coalition of Library Consortia (ICOLC), a consortium of consortia, whose listserv and meetings have the important effect of shortening the learning curve for individual consortia while accelerating the development of uniform and rational practices. The ICOLC home page (http://www.library.yale.edu/consortia/) gives useful information about the history and purposes of this influential body.

As publishers have merged, there has been a countervailing tendency for libraries to act collectively. In addition to the move toward formal consortia, there has been a growth of enlightened long-term interest based in part on moral aversion to past pricing practices. Many collection managers are consciously attempting to bias their selections toward the publications of societal nonprofit publishers. Indeed, the Association of Research Libraries' SPARC (Scholarly Publishing & Academic Resources Coalition), an initiative expressly undertaken to foster competition in the scientific publishing marketplace, has set out to found and nurture reasonably priced nonprofit publishing ventures set up against commercial rifles seen as overpriced (http://arl.cni.org/sparc/index.html).

III. Costs and Pricing

A. Information providers should offer a choice of pricing models from which the CSU libraries may select. These models could be based on various criteria including simultaneous use, user population, number of locations, library budget, and so on.

B. Pricing models which are based upon campus or system FTE often do not recognize that the content offered may have interest to only a limited segment of the total user population. Where a data base may not be of broad or general interest, pricing should be based on more appropriate criteria such as size of the actual user community or projections of use based on estimates or actual recorded usage.

C. A choice between unlimited use and pay-per-search pricing should be offered where possible. This would allow the CSU libraries a period in which to consider the value of each cost method while recording actual usage.

D. The unit cost for electronic content should be less than the unit cost for its print equivalent. In a period of transition, where both print and electronic formats of like content are offered, the print should be heavily discounted. The CSU libraries should not be required to purchase both the print and the electronic versions.

E. Unit cost should decline as the volume of use increases. Information providers should specify volume thresholds or breakpoints at which the overall unit cost will decrease.

F. The cost of providing access services and the basic cost of the content should be separate. The CSU libraries should be free to negotiate a license for electronic content which is separate from the access service. Should the same content be offered through multiple access providers, the CSU libraries should be free to choose which access service best meets its needs. The CSU libraries should be able to change access providers, if necessary, without having to renegotiate a license agreement for content.

G. The pricing of information and access services to the CSU libraries should reflect its contributions in helping to create a marketplace through training and exposure of products to new and future customers.

LICENSING

Having decided that a resource is desirable and cost-effective for its mission, collection managers approach the challenge of acquiring a license defining key terms of the use to be allowed. Whereas formerly a library might negotiate a few contracts a year to establish relationships with its serials vendors, book jobbers, and bindery, it is now commonplace to negotiate and sign a license for each separate electronic resource. Ann Okerson (1997) has usefully described the advent of wholesale licensing and the concerns it raises.

In larger institutions, the move from the decision to collect to negotiations and the making of final arrangements signals a transfer of responsibility from the chief collection development officer to the head of library acquisitions. In smaller institutions, these roles are often combined in one person. Regardless, it is not uncommon that legal counsel outside the library and technical staff within the library or its parent institution will be involved before any agreement is signed. Each of these actors brings essential knowledge to the table, and successful implementation of an electronic resource often requires close work among them with multiple iterations back and forth.

The basic purpose of a site license is to establish four things:

1. What is the full extent of the resource being acquired or accessed?

2. Who can use it?

3. For what purposes may it be used?

4. What specific human or computer resources will be applied to effectuate and monitor the terms to which the parties have agreed?

In defining the resource that is being acquired, several key dimensions must be specified precisely. In many cases, the library will be acquiring access to a resource through a vendor who does not publish the resource. Generally, for an ongoing subscription, this is not problematic. But concerns arise if the library purchases a backset, giving it permanent ownership rights to the publisher's material. License negotiators must, from there forward, assure that third-party providers of access to the material will honor the library's ownership rights, charging at most a nominal fee for accessing the backset, and that the publisher recognizes that, having been paid once, it cannot expect direct or indirect payment for anything other than the contents of the current subscription.

The ICOLC Statement speaks to the concern about archival access to a resource made available by a triangular relationship among library, publisher, and added-value provider: "The provider should grant to the consortium and its member libraries a perpetual license when the consortium purchases the content. That perpetual license must be transferable should the consortium or library wish to change providers, agents or vendors, or to switch from obtaining information from the provider's Web site to local or regional mounting" (emphasis in the original).

Often the resource being acquired is a collection of publications being offered by an aggregator who brings together the collections of disparate publishers under a dynamic canopy of site licenses. It is essential that both parties understand whether and to what degree the exact title mix of such resources may change over time. Both the Academic Universe offered by Lexis-Nexis and IAC's Expanded Academic Index have been somewhat more dynamic and unpredictable in their title listings than librarians had expected.

The ICOLC's "Statement" clearly summarizes the need for clear mutual understanding of both prices and content when it states: "All terms and conditions should be negotiated and clearly stated in the contract. Hidden charges, after-the-fact retroactive charges, changes in content, or any changes in commitment are not acceptable without re-negotiation."

Chronology is another important dimension to be defined. Many electronic resources are available for only the past few years. The library and vendor must mutually understand whether, over time, the file will grow or will cover a "rolling" time period. In the case of JSTOR (http:// www.jstor.org/support/), a Mellon-sponsored project which republishes in electronic form the text of core journals in key disciplines, the time limitations work in the opposite direction from most rolling periods of access in that all but the last few years of text are available.

Nothing is more critical or has led to more stress both from the point of view of philosophical conflict between library and vendor rights and from the point of view of technological implementation than the definition of the user community. For print resources, most libraries define the user community as anyone who can come to the resource, though they may have a narrower definition of the community of potential borrowers. For electronic resources, the problem is to ensure that authorized users have access in a way that protects the vendor's product by ensuring that no one else has access.

The typical university will seek to guarantee use to all current students, faculty, and staff regardless of physical location. While vendors and publishers do not usually dispute the rights of members of the on-campus university community, ambiguities arise with satellite campuses, for which there may be a surcharge, and more seriously in the case of students seeking remote access for distance education classes. The recent competition among universities as well as non-traditional entrants into the field for preeminence in distance education has only served to make more urgent the necessity that these definitional issues be clearly resolved in the case of online resources.

The definition of the user community for the typical university is fuzzy at two peripheral areas besides distance users. Many universities are associated with research centers and parks whose corporate entities are profit-seeking businesses. The degree to which access should be offered along the finely shaded gradient of these relationships is ambiguous. So far the headaches involved in defining these rights precisely seem to have led both parties to look the other way, but the potential exists for serious problems in this area. Most university libraries also welcome walk-in traffic, reasoning that the goodwill associated with helping visiting scholars or members of the local community justifies the small marginal costs of helping a non-primary clientele. Vendors and publishers sometimes fear a loss of potential revenue from these users and may seek to negotiate terms denying their access.

How a resource may be used is another important matter for negotiation. Some publishers of electronic journals have sought to bar all interlibrary lending, reasoning that the forwarding of electronic text is so easy that it invites abuse. Some publishers, such as Academic Press, have since relaxed their positions on this somewhat by such compromises as allowing the lending of printed copies downloaded from the electronic version. It is generally understood and accepted that the same principles of fair use that apply for print media pertain in the electronic world. Of course, this only shifts the locus of ambiguity since "fair use" controversies have yet to be resolved for print.

Vendors have sometimes sought to restrict use to certain purposes, such as Lexis-Nexis's former restriction to use for instruction over research. Usage is also occasionally restricted to certain user classes. Access to the Dialog Corporation's Classroom Instruction Program, for example, is restricted to students, and only to those students who have undergone a brief training program. The vendor's goal is presumably to avoid lost revenue by restricting usage to those less likely to be able to pay, and to promote future revenue by introducing the next generation of users to their resource. Restrictions of this type inevitably burden the library and make subscription somewhat less desirable.

There may be instances where the nature of use and the identity of the user interact. For example, a library might insist on the rights of its walk-in users and yet recognize that it is simply not fair to a publisher to allow attorneys from the community to spend hours researching case law on an electronic resource acquired with university funds to serve university missions. The library might agree to restrict such specific kinds of usage while generally allowing other walk-in traffic.

However a user community is defined, the definition must be susceptible to some sort of operational definition that can actually be implemented and monitored in an efficient manner. For this reason, discussions about site licensing involve a simultaneous consideration of the theoretical question of rights and the practical aspects of authentication. The increasingly dominant trend is for access to online electronic media to be governed by internet protocol (IP) address. Access is allowed to all users whose IP addresses indicate institutional affiliation with a subscribing institution.

Nearly every user community will have users whose right to the resource is unarguable, but whose access through the Internet comes through an Internet provider not affiliated with the parent institution. In such cases, "proxy servers" are used. The user is asked to come in through a university computer where some reasonably failsafe means of security, usually based on a password, is used to prove identity. The campus computer then routes the inquiry, which now bears the virtual postmark of the university, an address within the defined IP range, on to the provider's server.

Some service providers still insist on the use of passwords, both for affiliated and for walk-in users (or out of a desire to deny usage to walk-ins). Academic libraries' patience with this more onerous approach is waning, though libraries are less loath to use passwords when they can be made part of signon scripts. When this is possible, the computer rather than the individual "remembers" the password, and authentication is in every practical sense by machine, much as with IP authentication.

Technical issues also cloud the definition of "simultaneous user." Where a library is acquiring a resource whose use will be controlled by simultaneous users, it is important to know exactly what kinds of transactions count. If a user has followed a hot link from the main resource to a cited journal article maintained on another resource, or is reading an article from the main resource that has been cached by her computer, does she count against the total? An understanding of this issue is needed to inform the decision about how many user authorizations are required.

Libraries' demands of a good site license go somewhat beyond the simple issue of access. There are often conflicts with vendors about liability in the event of a catastrophe or about whether the laws of the vendor's or the institution's state would apply in the event of litigation. Libraries sometimes seek to limit their obligations to monitor and prevent abuse, negotiating to have only reasonable prudence expected.

One of the most significant ancillary expectations is for good statistics. Since actual usage is not generally the basis for billing, libraries cannot automatically expect usage statistics with their invoices. Yet statistics are still needed for a variety of reasons. Statistical reports on the amount of use of each database, or the usage of each title within a collection, are invaluable in helping the library to make the most cost-effective commitments with its scarce dollars. Statistics on response time can be used to diagnose technical problems on either end of the exchange. Where the library has purchased access for a given number of simultaneous users, statistics are needed to fine-tune the purchase. Especially useful in this regard are reports on the number of turnaways (more users wanting access than are permitted). In the absence of data, libraries often buy access for more users than are required and need statistics to show that satisfactory access could be had with fewer ports. The importance of statistics is so critical that the International Coalition of Library Consortia (ICOLC) has issued its own "Guidelines for Statistical Measures of Usage of Web-based Indexed, Abstracted, and Full Text Resources" (http:// www.library.yale.edu/consortia/webstats.html):

IV. Licensing

A. The license should include permanent rights to use the information that has been paid for in the event that a licensed database is subsequently canceled or removed. Sliding year access, backfiles, and access to complete runs should be specified.

B. Information providers should employ a standard agreement that describes the rights of libraries and their authorized users in easy-to-understand and explicit language. The terms should reflect realistic expectations concerning CSU's ability to monitor use and discover abuse. Agreements should contain consistent business and legal provisions; however, nothing should prohibit CSU's rights under the laws of California.

C. Authorized users are current students, faculty, staff, administrators, and all other on-site users of the campus or University. Authorization and authentication of users is a shared responsibility of both the providers and the CSU.

D. Licenses should permit fair use of all information for educational, instructional, and research purposes by authorized users, including viewing, downloading, and printing.

E. Licenses should not limit CSU's rights to enhance or reformat data if content and integrity are preserved in order to make the data more visible or convenient for CSU users within "fair use."

F. CSU use data should be available to CSU as part of contractual provisions. Confidentiality of individual users and their searches must be fully protected. Use data generated by CSU may be made available to the information provider.

G. The CSU libraries should have the right to renegotiate contract terms where concessions have been provided to others on the basis of most favored customer status.

H. Contract start dates should be synchronized with the beginning of the fiscal year.

FUNCTIONALITY

It is functionality above all that has taken over a portion of the overall basis for making selection decisions about electronic resources. There is obviously no point in having the most potentially useful intellectual resource if technological impediments make it impossible to get from here to there, or if the reproduction of scientific photographs and charts is so fuzzy that information content is badly compromised.

As bad as functional issues and problems can be at present, they were worse in the early 1990s. Each system seemed to have its own search engine. Few were really intuitive. Keyword searching was not standard. Special command formats such as "au=" or ".au" for author were common. The introduction of the World Wide Web and standard Web browsers, together with technological developments reducing storage and processing costs to the point where brute force rather than software elegance can bring results, have brought the welcome disappearance of derived title keys and other hard to remember search idiosyncrasies.

The near ubiquity of Web interfaces has not meant that functionality can be taken for granted or that differences in functionality might not be sufficient to sway a selection decision. In fact, because Web technologies, extensive networking, and the growth of full text are so successful, many libraries have found that their patrons are staying away in droves. Across the country, reference activity is down and use from offices or homes is way up. The implication is that libraries may never have the reference opportunity to show the user how to exploit a resource--so it had better be fairly self-evident.

Functionality should be assessed from both the library's and the patron's points of view. Both library and user require that a variety of platforms be supported, and ideally that platform issues be immaterial. Both benefit if the search software relies on a standard browser. At the other end of the continuum is nonstandard software that must be physically installed on user machines.

The library and its patrons also share an interest in documentation, though not necessarily the same documentation. From the library's point of view, a certain degree of systems documentation may be needed. At my own library, each online resource has a "designated expert" who is expected to keep current with any vendor documentation. Often e-mail or listservs supplement traditional print documentation in keeping library staff aware of changes in a resource.

From the patron's point of view, good documentation may mean user manuals or flip charts or other kinds of "cheat sheets." Increasingly, of course, good documentation simply means online help whose availability is obvious. Both global and context-sensitive help should be available. Although libraries typically expect to prepare a great deal of user documentation on their own, inexpensive and attractive vendor documentation can play an important role in instructional efforts.

Of course, the issue of documentation is handled best when a system is so intuitive that no help is needed. Most libraries are willing to assume that their users know how to use a standard Web interface, and therefore streamline training efforts for resources with standard presentations. Sometimes, however, the complexity and variety of the data being presented make this impossible. In these cases, the provision of both a simplified and an expert search interface is ideal. The Web of Science is a good example of a resource which can be searched in complex ways but which gives the naive user with a simple request a clean and intuitive way to search.

Even within the comfortable domain of standard Web interfaces, there are important attributes that separate one resource from another. Elegant and intuitive design is certainly important. The ability to send citations or full text to oneself as e-mail or to order documents not covered within the library or consortium's license are important features which may or may not be present.

Libraries typically seek to train users on the electronic resources that will be of the most interest to their work. Even where the interface is fairly simple, it is important to convey to users the scope of each resource. A class in educational psychology, for example, should be made aware of the differences in coverage between ERIC and PsychINFO and should be taught that each uses a different controlled vocabulary. Training needs raise an important issue that the library should negotiate in its site license for any resource whose use is governed by the number of simultaneous users. Whereas a small number of simultaneous users may be adequate 95 percent of the time, several times that many users may need access for the increasingly popular "hands on" component of training a large class.

Vendors have handled this need in a variety of ways. Where trust is high, passwords are distributed to allow the library a temporary increase in the number of users. Sometimes it is necessary for the library to call ahead to schedule such use, a cumbersome requirement at best. Sometimes a highly truncated training database is made available. The library entering an agreement for a new resource should specify its needs in this domain.

Response time is of course another critical element of functionality. To the user, it means time well, and efficiently, spent. Besides being the user's advocate, the library is internally affected by response time since poor response can generate the need for more machines or for an increase in the number of simultaneous users allowed under a site license. The key determinants of response time are the power and efficiency of both the database server and the user's machine, the nature of the resource (graphics-happy pages are slow), and telecommunications. Many libraries notice poor response time in mid-afternoon when the internet is busy. The source of any problem could be anywhere, including "the last mile" on the campus telecommunications network. Where slow response turns out to be the vendor's fault, the library faces a dilemma that may affect its selection decision.

Obviously the worst response time is infinite response time encountered when a system is down. Downtime problems have been severe for some vendors, so much so that on occasion it has been necessary to replicate a resource on a mirror site to which traffic can be sent when the primary site is unavailable.

The availability of full text introduces some new elements to the functionality a library may expect from its vendor. The vendor should be able to provide links to full-text publications to which the library has rights. Within a network, this may mean that, based on IP recognition, users at some libraries may be able to traverse hot links that are not enabled for users at another institution. The degree to which the database provider bears responsibility for communicating with libraries and fellow providers so as to recognize rights and enable links is at present a matter of some ambiguity. Regardless of the locus of this responsibility, however, it is reasonable for the library to expect that there should be no false links. Patrons should not be led to attempt to follow article links or to request modules within an information resource where the institution has no rights and the requested use will ultimately be denied.

As bandwidth capacity increases and computer capacity grows, we can expect to see powerful new features which will expand both the intellectual content available and the functionality issues raised by its presentation. We can expect to see the inclusion of source data susceptible to downloading and re-analysis, the provision of more multimedia content such as music, multi-dimensional rotations, and zoomable photographs.

The development of e-journal capabilities is a fast-moving front. Linda Stewart (1996) has usefully cataloged the concerns of chemists for such basic features of today's electronic journals as portability, comfort, convenience, permanence, and openness to serendipity. Steven Bachrach et al. (1998), taking matters further, have cataloged the unique advantages we can look for in future chemical journals. And a registry has already been started for electronic journals containing embedded multimedia (http://www.public.iastate.edu/~CYBERSTACKS/M-Bed.htm).

An entire rethinking of scientific and scholarly publication, spurred when the possibilities of the new technologies really sink in, will raise more profound issues, including the possible disaggregation of articles from their journals or components from books, publication before review, or the inclusion of corrections and commentary in articles that could conceivably never have the final canonical form they now have when printed. The electronic book itself will raise a new set of possibilities. It will also introduce such questions as whether it is better to acquire text in physical format or through downloads, whether to do so via anticipatory collection development or on demand, how many simultaneous users should be acquired or how long "checkout" periods should be, and how secondary uses can be monitored and policed. All these issues are less far away than we may think and will affect the criteria libraries use in choosing electronic resources.

V. Functionality

A. Documentation should be clear, concise, and comprehensive. Instructions and examples should be provided for both the search engine and database-specific features. Documentation should be provided in both online and print format. Online help should be context-sensitive.

B. The system capacity and network infrastructure of an information provider should be technologically up-to-date and provide for optimum response time.

C. System-based error messages should provide clear indication of the nature of the error.

D. Information resources should be platform-independent. Vendors should provide information in industry standard display and output formats.

E. Vendors should provide options that meet Americans with Disabilities Act compliance.

F. Vendors should provide sufficient notice to keep the CSU libraries informed of format, content, and platform-based changes.

G. Vendors should provide training or instructional passwords if a limited number of access ports are available.

H. The CSU libraries should be able to make reasonable requests for customization at the system-wide or campus level, including, but not limited to, adding local periodical and serial holdings from individual libraries.

I. Interfaces and search engines should be easy to use for first-time users, yet provide advanced searching capabilities making full use of searching features.

ARCHIVING

It is unsurprising that traditional parlance has linked "libraries and archives." In a print world, the library that bought an item automatically became its archive, provided that it did not discard or lose it. Now, however, it is increasingly common for libraries to lease resources. Even when information resources are nominally owned, the transience of the technology and the frequent lack of physical custody and control make the question of long-term access problematic.

Despite the rapid movement to the Internet, there are still electronic products physically sent to and owned by libraries. Most of these are on CD-ROM. Those that represent a static resource, such as The Pennsylvania Gazette, a Revolutionary era newspaper, are typically considered to be owned. With subscriptions, ownership and the right to retain CDs in the event of cancellation is sometimes included in the contract and sometimes expressly denied. Even when the physical medium is unambiguously owned, the library may be concerned that technological changes will leave it without compatible hardware and may therefore want to reserve the right to transfer the information content to future media. (It may not be realistic to assume that such conversion would always be very important to a library upon the real occasion, given the decline in the value of most noncurrent information resources.)

When a resource is available on the Internet, the library has no physical possession and clearly has grounds for concern if an archive is seen as important. Even the vendor's assurance of perpetual rights cannot assuage concerns for continued access should the vendor cease to exist. Vendors have on occasion made arrangements for third parties to take possession in escrow of their resources should they go out of business.

A more common alternative is for a third party to host the data owned by a publisher. For example, with its Electronic Collections Online (ECO) project, OCLC has moved the text of electronic journals actually owned by a variety of publishers to its own computers, where access to the archive for those subscription years actually "owned" by individual institutions before they canceled their subscriptions is guaranteed in perpetuity. Since OCLC's long-term survival and good faith are not seen as at risk, this arrangement is an adequate assurance of perpetual rights for most libraries.

VI. Archiving

A. The CSU libraries have a legitimate interest in maintaining collection integrity through archives of the electronic resources they have licensed or otherwise acquired. The CSU libraries have a mission to ensure easy access to archival electronic material.

B. The CSU libraries should have permanent rights to use information that has been paid for in the event that a licensed database is subsequently canceled or removed.

C. Responsibility for providing archival access should be clearly defined in all agreements and licenses.

D. The CSU libraries should have the right to make or obtain electronic or printed copies for archiving and for use in perpetuity.

E. In the event that the information provider is unable to maintain archival access, the CSU libraries retain the right to maintain archival access on their own servers and to specify appropriate formats or methods for the transfer and storage of archival information.

Electronic Access to Resources Committee 1/30/97

CONCLUSION

New information technologies will never make irrelevant the traditional goals and values of collection development, but they have introduced important new elements to decisions about selection and retention. What really is owned (or leased), who may benefit and for what purposes, whether convenient access really can be guaranteed, and how long the resource will last are all in play in a manner that was never true before. The peripheral vision of collection managers is tested by technological, legal, and consortial issues which must be viewed with some acuity so that what is seen with the main focus of vision is not jeopardized. Although the specific challenges to collection managers will change, there is no reason to believe that they will be simplified.

REFERENCES

Bachrach, S.; Burleigh, D.; & Krassivine, A. (1998). Designing the next-generation chemistry journal: The Internetjournal of chemistry. Issues in Science and Technology Librarianship, 17(Winter), 1-7.

California State University. (1999). Principles for CSU acquisition of electronic information resources. Retrieved July 23,1999 from the World Wide Web: http://www.co.calstate.edu/irt/seir/EIR.prin.html.

International Coalition of Library Consortia. (1999). Guidelines for statistical measures of usage of Web-based indexed, abstracted, and full text resources. Retrieved July 23, 1999 from the World Wide Web: http://www.library.yale.edu/consortia/webstats.html.

International Coalition of Library Consortia. (1999). Statement of current perspectives and preferred practices for the selection and purchase of electronic information. Retrieved July 23, 1999 from the World Wide Web: http://www.library.yale.edu/consortia/ statement.html.

Okerson, A. (1997). The transition to electronic content licensing: The institutional context in 1997 (Scholarly Communication and Technology Conference of the Andrew W. Mellon Foundation, Emory University, April 24-25, 1997). Retrieved July 23, 1999 from the World Wide Web: http://www.library.yale.edu/~okerson/mellon.html.

Stewart, L. (1996). User acceptance of electronic journals: Interviews with chemists at Cornell University. College & Research Libraries, 57(4), 339-349.

Virginia Tech University Libraries. (1999). Collection development policies: Goals statement. Retrieved July 23, 1999 from the World Wide Web: http://www.lib.vt.edu/info/colldev/coll_dev_policies/GOALS.html.

Paul Metz, Virginia Tech University Libraries, P.O. Box 90001, Blacksburg, VA 24062-9001

PAUL METZ is Director, Collection Management and College-Based Services for the Virginia Tech University Libraries. He has been active in university governance, the Virtual Library of Virginia (VIVA), and library research and publishing.