Archives & Museum Informatics

Conferences

Publications

Seminars

Consulting

Research

Search archimuse.com

Share

last updated:
October 7, 2014 2:55 PM

Economic, Social, and Technical Models for Digital Libraries of Primary Resources:
the example of the Art Museum Image Consortium (AMICO)

published in New Review of Information Networking, #4, 1998, pp 71-91.

David Bearman and Jennifer Trant1
Archives & Museum Informatics/Art Museum Image Consortium

Introduction

The potential of electronic technology to transform scholarship and teaching has remained largely a promise for almost a decade. This paper explores a recent effort to overcome many of the barriers to integrated digital humanities resources, and presents a new model for collaborative, not-for-profit electronic publishing. The authors have been involved in discussions of the requirements of a social, economic and technical system for distributed access to the vast archive of human culture for many years.

The model reported here has its origins in a series of conferences and workshops, experiments and projects, that date from the 1992 Irvine Conference on Technology, Scholarship and the Humanities2. The Irvine meeting raised many issues and highlighted opportunities for the humanities to exploit information networks as a vehicle for furthering scholarship. A follow-on meeting hosted by the Getty Art History Information Program (aHIP)in early 1994 pushed these issues further, particularly in reference to digital imaging3. The forty participants who occupied critical positions in the electronic publishing, distribution and consumption chain explored barriers to the transformation of scholarship in the digital age and recommended means to overcome them. Consensus formed around three critical issues:

To address these identified barriers, AHIP established its Imaging Initiative4, that sponsored educational activities and standards research5 and launched the experimental Museum Educational Site Licensing Project (MESL) for which Trant served as the first Project Director. MESL demonstrated the possibility of overcoming these barriers, but its experimental, short-term nature limited its utility as a model for a full-scale implementation.6

MESL made a contribution to our understanding of higher educational use of digital museum documentation, but left a number of critical issues unresolved. During the course of the MESL Project, some technological delivery barriers serendipitously disappeared with the rapid adoption of the World Wide Web. Even working with a limited data set (of approximately 10,000 images at the end on two years), however, revealed the many, varied and relative meanings of "critical mass" to different communities of users. At the same time, the web exposed the fragility of the intellectual property system. But MESL museum participants and university educators reached an agreement on principles of managing licensed museum intellectual property, brokered by Mary Berghaus Levering, Deputy Register of National Copyright Programs, U.S. Copyright Office.7 Based on these results, three members of the MESL management committee took the next step, and set out to organize a self supporting entity that would enable the educational use of museum multimedia documentation.8 In March 1997, at the Museums and the Web Conference in Los Angeles, they hosted the first of four planning meetings that led to formation of the an independent, not-for-profit membership organization: the Art Museum Image Consortium, Inc. (AMICO).9

AMICO is making a self-conscious effort to overcome the barriers identified at the 1994 conference, by articulating, deploying and testing new economic and social models.10 We believe this rethinking of traditional roles and relationships has relevance for other large-scale, digital libraries of primary materials.

Background to the formation of AMICO

The philosophical basis of AMICO has its roots in both of a number of commercial efforts to acquire museum intellectual property (of which the most notable were by Bill Gates' firms) and the problems identified by the MESL Project.

In 1989, Bill Gates capitalized a firm called Interactive Home Systems (IHS), and set out to acquire digital rights to a wide range of digital images. IHS purchased images that were traditionally held by photo houses, and also made an aggressive effort to acquire perpetual, exclusive, rights to digital images of museum holdings. The museum community reacted immediately and negatively. The concept of selling rights on an exclusive, worldwide, basis even for a limited period of time is contrary to museum policies and practices, and perpetual grants of rights are unheard of. Large amounts of money managed to attract a very small number of institutions, many of which were in a precarious financial position. By 1994, Interactive Home Systems had reorganized as Continuum Productions Inc. The mission of the new company was the same, although the strategies had changed slightly; they backed off from demanding exclusive rights, but still sought rights in perpetuity. Continuum also failed to make significant acquisitions of museum content, and the firm underwent another name change, to Corbis, before seeming to pull away from active efforts to obtain museum rights in the fall of 1997.

Museum directors were approached many times by Gates' firms. Initially they were offered prospects of huge profit and up-front "signing bonuses;" as time went on the directors realized that such deals were unlikely to make their museums rich and that the legal and policy implications of assigning such broad rights were far too great. Motivations for engaging in digital, and web-based activities were changing. Museums had limited success in commercializing cultural intellectual property world-wide; but increased educational access and use were seen as mission-critical results of digital projects. Some museum directors, such as Maxwell L. Anderson, liaison for information technology for the Association of Art Museum Directors (AAMD), strongly felt the need for a museum-owned alternative, that could fulfil the function of making digital images of museum collections available, while at the same time reflect museums' values and priorities.

The Museum Educational Site Licensing Project (MESL), was formed as an experimental proof-of-concept for such an entity. Along with Bearman and Trant, Anderson saw MESL as the initial step towards a full-scale implementation.11 But in the spring of 1995, senior management at AHIP became reticent.12 Instead of planning for a follow-on MESL moved into an extended phase of reporting and evaluation.13 Anderson, Bearman and Trant felt strongly that it was important to capitalize on the momentum generated by MESL; a separate initiative would need to be organized if museums were to succeed in effectively managing their digital intellectual property.

Planning for an independent organization began under the auspices of the Association for Art Museum Directors (AAMD). In March of 1997, representatives of almost forty museums met to discuss whether a consortium would be useful and, if so, how they would plan it. This group endorsed further discussions and announced their collaboration to non-profit organizations.14 Between March and July 1997 three additional planning meetings were held; a web site was set up, and draft documents were circulated and discussed online by those museums continuing to be represented in the process. By August, twenty-five museums had drafted the principles for a new consortium, defined the terms for licenses to its collective digital library, and identified the services that such a consortium could provide to its members. Participants in the planning process were invited to join the AMICO, and it was formally established as a program of the Association of Art Museum Directors Educational Foundation in October 1997. Twenty-three of the twenty-five organizations involved in planning AMICO joined as founding members.

At the first meeting of the Board, in January 1998, the members voted to establish AMICO as an independent, not-for-profit organization, thus ensuring that institutions not eligible to participate in AAMD (by virtue of not being art museums or not being in North America) could join AMICO. Broad representation was recognized as valuable; since the strength of the consortium lies in its collective library, no barriers were placed on its potential growth. AMICO was incorporated in the spring of 1998, and now operates as a US-Based, membership governed, charitable, not-for-profit organization.

The AMICO Model

As constituted, AMICO represents a number of social, economic and technical departures from the manner in which digital libraries have commonly been constructed. AMICO is a self-consciously innovative institution, striving to develop new models for the creation of digital representations of primary materials for educational use. Throughout the planning process AMICO members consciously tried to separate and articulate the roles in the creation, collection, distribution and use of digital library resources. New roles have been defined for most in order to create what the members of AMICO hope will be a self-sufficient system, capable of expanding to incorporate the art, and art holding institutions from around the world.

The architects of the AMICO model believe that it is in no way limited to the creation of digital libraries of art. It can also serve as a basis for a new economy of scholarly publishing in the digital age. The model requires the rethinking of social relations between participants in the process: content holding institutions, content aggregators and value added publishers, content distributors and service providers, content subscribers, and content users. It is more a social contract than a commercial publishing model - the content holders take on added costs and responsibilities in return for the subscribers promises to exercise respect for their intellectual property. Participation is motivated by shared goals and objectives.

The basic framework for the AMICO model is:

These relationships between parties are outlined in figure 1. The social, economic and technical foundations of the AMICO Model are outlined in the sections that follow. Each is considered from the perspective of some or all of the players in the content creation, distribution and use cycle.

relationships between parties
CLICK FOR LARGER IMAGE (71kb)

Social Roles in the AMICO Model

Content Holding Institutions

AMICO enables non-profit institutions holding bodies of primary materials to make them available for education use without having to bear the burden of creating and maintaining distribution systems. Content holders:

Figure 2 diagrams the types of content contributed to the AMICO Library by members.

types of content

These significant responsibilities are assumed voluntarily by AMICO member museums, because their missions center around education, and the provision of access to their collections. Participating in AMICO is seen as a way of furthering their own educational objectives. Each year, content holders agree to provide additional new materials under a non-exclusive license to the. Individual institutions and groups of institutions may need to seek outside funding to enable them to fulfill their obligations to provide such documentation and rights to the collective, but the collective per se is not underwriting these expenses.

It is hoped that the members, working together, can develop frameworks for acquiring rights for educational uses from living artists and artists estates holding copyright in contemporary works. If this turns out to be possible, it will be a major benefit to the participants as acquisition of such rights on a work by work, use by use, basis is time consuming and often can be costly.15

The Consortium as Content Aggregator and Value Added Publisher

Content holding institutions form a consortium to make their digital representation available because the contribution they can make to a digital library is more valuable in the context of others. Together members can also define value-added processing that can be done collaboratively much more efficiently that individually. Because content holders own and govern the consortium, they can set the terms under which their materials are provided to subscribers and they can determine the nature of benefits provided by the consortium to its members. In order to encourage as many other institutions as possible to join, membership is open to any non-profit collecting institution willing to adhere to membership terms.

Since these terms include obligations to contribute data and personnel time, the members of the consortium need to make sure that the benefits of participation are real and are made obvious to staff throughout their own institutions. The principal benefits are:

The consortium also develops specific licenses with and for different types of educational subscribers and administers the licensed distribution of the Library. It agrees to protect the intellectual property licensed to it by members from misuse. In this way, members are enabled to publish their content without the cost of being their own publishers, as well as benefiting from the value-added indexing created consortially.

Content Distributors and Service Providers

The AMICO model is quite unusual in that the consortium does not create software to deliver content directly to subscribers. AMICO has decided that existing distribution services are best prepared to deliver the AMICO Library to their constituencies with attention to the special needs of those clients. Since AMICO is a non-profit, its library is available to multiple non-profit and governmental data distributors. AMICO believes that competition between these distributors can be desirable to provide potential subscribers with a choice of different value-added services and to encourage the development of specialized software to exploit the Library. The value of the AMICO Library is also enhanced through its integration into the context of a full range of distributor's products and services. Distributors perform several services for AMICO that would otherwise fall to the consortium,16 including

In return, AMICO allows the distributors to charge a fee over and above the AMICO license fee, to recover these costs. Distributors may also cross-license software they develop to other distributors. In principle, this could provide savvy distributors with a secondary source of income from their AMICO distribution.

Content Subscribers

The only subscribers to the AMICO Library are educational institutions.17 Subscribers may define their own user communities, within broad limits defined in the appropriate AMICO Library Agreement. Uses are restricted to those with educational purposes; neither commercial use, nor redistribution of any part of the Library is permitted. In return, educational institutions are expected to have intellectual property policies and procedures in place, and to educate their clients about proper use of licensed resources. Subscribers are not responsible for what their users do, per se, but they are responsible for enforcing their own rules if they learn of violations by their users. Hence, AMICO and the subscriber have a pact to try to develop responsible use of licensed resources.

Like distributors, subscribers may develop specialized tools to manage the AMICO Library and are free to cross-license such tools to distributors or to other subscribers. Mechanisms for subscribers to contribute content to the library in lieu of some or all their license fees are being explored.

Content users

Individuals defined by subscribing institutions as designated users of the AMICO Library are entitled to make use of it within the terms of the AMICO Library Agreement independent of their location or method of access. No charges may be applied to individual use, because AMICO wants to ensure that increased usage is encouraged rather than discouraged. Users may only make agreed educational uses of the resource without further licensing. Content users are encouraged to comment and provide feedback to content creators; this dialogue is encouraged as the educational mission of AMICO members depends on meeting users needs.

Economic Roles in the AMICO Model

These social innovations are made possible because the AMICO model is designed as a non-profit self-supporting economic system. But being self-supporting in an environment with new sources of content creation and rights acquisition expenses and different costs of production and distribution, requires new economic equations and arrangements between parties. Whether implicit or explicitly, all participants in the new system must share costs differently and assume different economic burdens than in the past.

Content Holders

Content holders take on large expenses they have not previously not incurred. Not only do they bear the cost of creating the digital representations of works in their collection, but they are obliged to research and negotiate rights to works not in the public domain. We hope that over time digitization will become an extension of daily processes. Rights negotiation, however, has often traditionally been left to those requesting images. But because AMICO members are providing the documentation for licensed use, they must now take this task on themselves. We are hoping to lessen this cost over time, through consortial agreements and standard requests.

Content holders also pay membership dues to the Consortium. These dues are one of the major sources of AMICO's income and have supported the majority of very substantial start-up costs. Dues are scaled to the size of each member's annual budget. AMICO members are paying approximately the same amount to receive a license as members as they would to obtain a license as subscribers.

Consortium

The Consortium itself pays the costs of collating, validating, editing, indexing, and authoring value-added content for the Library. It also manages the publication, distribution and licensing of the Library on behalf of its members. The Consortium seeks new venues through which to disseminate the Library, and develops agreements with new communities of educational users. The risk of publishing the Library, including liability for rights offered by content owners, is assumed by AMICO.

AMICO also provides its members with services, especially technical assistance, required to make their contribution to the Library, and over time to create better educational materials. It provides a forum for discussion and debate among members, and for the development of shared guidelines and best practices. The consortium will study uses of the Library to provide members with insights into the ways in which their educational users are best served and to develop strategies for further enhancing the Library.

Non-profit distributors

Without cost to AMICO or its members, non-profit distributors provide in network facilities, distribution software, and value added tools to access and use the contents of the AMICO Library. They do this within the context of other network information delivery services, thus enabling economies of scale. AMICO provides Distributors with a free license for development purposes for up to a year before a product is released to their clients. Without cost to the consortium or its members, distributors market the AMICO Library and their own services. Distributors may recover the costs of their services in the fees they charge to subscribers that include the AMICO license fee; as a consequence, AMICO receives its license fees without having to support a billing and collection function. Distributors may also recoup development by licensing software tools developed for the AMICO Library to others.

Educational subscribers

Within the AMICO model educational institution share the costs of enabling access to a large collection of primary materials, by paying subscription fees calculated to recover collation costs incurred by AMICO and development costs incurred by distributors.18 Since costs are shared by all users, pricing has been calculated based on anticipated levels of subscription after five years. Since there will be fewer users and higher costs during a startup phase, the AMICO model involves substantial financial loses for the consortium in the first five year period. It is hoped this shortfall will be made up through philanthropy, so that debts don't need to be serviced after the consortium becomes self-sufficient.

While universities have recently made it clear that they want to "own" resources, rather than lease them, the AMICO model does not provide the subscriber with outright ownership. Instead subscribers pay an annual fee for a year's use of a growing Library. This fee structur reflects the regular enhancement and augmentation of the library, and the associated ongoing expenses. An annual term is also dictated by the fact that AMICO may not be able to license the library for more than a year at a time, because some rights must themselves obtained by members from other rights holders.

A variety of economic options for subscribers are being explored. Under one variant of the model, subscribers may be able to partly pay their subscription fee with digital content contributions. Provisions have already been made to allow subscribers to earn income from licensing tools they develop for use with the AMICO Library. And some subscribers are already exploring becoming distributors to other institutions -a university might be the distributor to public libraries or schools in its region--, furthering its own mission and sharing costs.

Users

As a matter of principle, AMICO requires that the AMICO Library be delivered free at the point of use; users must be provided with unlimited access (under terms of license) without an additional fee. Users also receive support for access from the distributor without a direct charge to themselves.

Technical Roles in the AMICO model

The AMICO model is primarily an innovative economic and social response to evolving technology opportunities. These responses, however, influence approaches to technology and its implementation. Here we will discuss only those technological approaches that are essential to the model, rather than looking specifically at the various ways in which AMICO is doing its work.

Content Holders

Despite more than a decade of study and many excellent efforts, cultural institutions still lack standards for the digital representation of most types of primary materials. Building on the best work to date, AMICO members have developed a shared data specification, and are exploring best practices in digitization. AMICO does not see itself as a standards setting body; shared specifications are essential to normalize output from divergent systems. Until such time as community-wide, broadly implemented standards are available, an AMICO data specification will be necessary to achieve coherence in a jointly constructed digital resource. Sharing this common target is a benefit from the perspective of members, however. Collaboration provides a form of protection; they will then be able to deal with the future from a common foundation.

Consortium

The consortium is engaged in preliminary experiments to exploit new technologies that enable educational use of the AMICO Library, relating directly to the architecture of the AMICO model.

First, the new social relations that AMICO is trying to forge require open lines of communication between parties which don't normally know, or speak with, each other. To facilitate communication AMICO has created a set of electronic discussion lists which enable it to host dialogues between its members, between its members and distributors, between its members and subscribers, and between its members and end users. Each forum involves numerous topics and often engages the attention of working groups specifically created to address evolving issues. These are linked to in-depth web pages that support document exchange and reference.

Secondly, AMICO is committed to developing an open disciplinary knowledge model and to taking a lead in implementing emerging standards for making metadata of its Library available on the World Wide Web. In both cases, AMICO recognizes that the universe of relevant educational resources will be larger than that provided by any one source. Enabling interoperability will enhance the value of the AMICO. A shared knowledge model and common metadata support its use in conjunction with many distributed resources.

Thirdly, AMICO is committed to the development of documentation standards and openness. It has been involved from the first in development of metadata standards, and has made its Library available as test data to support these processes. As examples, AMICO data was used in the modeling the relationships between various information types in the Dublin Core (DC).19 Discussions. Methods to create DC-compatible metadata for multimedia files are being pioneered by AMICO on behalf of its members.

Finally, AMICO is exploring new uses for term lists developed initially for authority control of data entry in cataloging systems. Bearman and Trant have long advocated placing such vocabularies and thesauri between the searcher/user and the databases rather than enforcing consistency in data values found in records from diverse sources. To make such enhanced searching work, of course, the terms from the data must be represented in the vocabularies. AMICO is developing matching routines and adding terminology to reference vocabularies and thesauri. In addition, because the members contributing data to AMICO have their information resident in local databases that may not be able to effectively use terminology resources, AMICO is exploring ways to return enhanced data, incorporating the results of these comparison routines, to its members.

Non-profit distributors

AMICO's distributors are innovating in a variety of ways in their provision of access to the AMICO Library, and in their integration of this resource with others. The terms of AMICO Agreements require user authentication through various methods that are not tied to where the user is located. Since some users must be authenticated by means other than IP addresses, all users can be authenticated at the subscribers site and tokens can be passed to the distributor which encode information about the users for subsequent demographic analysis. While not in itself a prime objective of the distributors, their collection of use data that can be analyzed by AMICO and others will benefit all publishers and users of digital libraries in the longer term. Supporting sophisticated user studies is one of the major objectives of the AMICO University Testbed Project: understanding the nature of uses of primary digital resources.20

AMICO's distributors are also being strongly encouraged to implement searching through expanded, regularly updated, vocabularies and thesauri. Because AMICO data comes from many institutions with different cataloging traditions, different types of users, and different source collections, vocabulary and terminology used in management of the collection varies substantially. Rather than attempting to impose rules on all members, AMICO prefers to introduce term explosion and synonym searching functionality into searches between the users query and the resources being searched. Sophisticated versions of such functionality might engage users and authority or reference databases in dialogue prior to formulating the final search. AMICO's own work with common vocabularies will support this aspect of enhanced distribution.

AMICO distributors are also being encouraged to create tools appropriate to the different kinds of end-users, and types of uses, purposes and work methods that characterize their clientele.

Educational subscribers

Educational subscribers are considered partners with AMICO in making resources available for educational uses. As such, they are permitted under the terms of AMICO licenses to locally mount all or some of the resource and any of the metadata for value-added purposes, and are encouraged to develop ways to integrate resource metadata with locally provided facilities. In return they are obliged to manage content mounted locally, reporting on its use.

Users

Users of the AMICO Library need access to information which responds to their purposes, rather than simply in pre-configured packages. AMICO anticipates users of all ages, with diverging levels of interest and knowledge. The information provided by the AMICO is likewise suited to different audiences. Making the Library work for these different users is one of AMICO's greatest challenges. We are exploring methods of integrating user profiles (both explicitly declared by users and implicitly constructed by the systems from user behavior) with metadata about the level of content, to customize user interaction with the AMICO Library.

Another way build value in the AMICO Library is to locate it within and environment of easy to navigate links to many other kinds of resources that provide context. AMICO's development of a disciplinary knowledge model articulated in SGML/XML and made available to publishers of numerous related resources will support navigation of links across and among resources. Over the coming years we hope to augment this disciplinary knowledge model with views specific to particular kinds of users and user situations.

Experience to Date

License Negotiation

Despite the extensive analysis of mutual interests undertaken by the museum and university personnel involved in the MESL project, and the resulting framework for licensing agreements between museums and university educators, the process of negotiating the AMICO Library University Agreement with the 18 testbed participants has been a protracted one. In the MESL model, museums agreed to provide very high resolution images, permit local storage of such images on systems outside the technical control of the university but within its sphere of social control, and supply detailed textual records from internal control systems. In return, the universities agreed to permit only educational uses, to have in place strict policies regarding the misuse of licensed intellectual property, and to report on uses.

Between the time these agreements were reached in principle (spring of 1996), and the advent of licensing for the AMICO Library in the fall of 1998, the context for licensing electronic resources in most American universities had been transformed by the relatively sudden growth of such opportunities. As a consequence, librarians and general counsels have developed strict guidelines about the kinds of terms they agree to in licenses. Through international discussions, and often in consortia with other institutions, university electronic resource managers articulated principles for licensing electronic resources which conflicted those of the MESL project. Specifically, the sheer volume of licenses made universities unwilling to engage in monitoring and reporting, regardless of the trade-offs. Thus, when AMICO circulated what it felt was a very liberal license to its university testbed participants in the fall of 1998, it found that many were unwilling to assume obligations to report on "local mounting" of images, to ensure that images and data would be displayed together as agreed by MESL participants, or report on modifications made to image files.

Prior commitments to the university testbed will lead to compromises between positions of both AMICO and the universities in 1998. But it is increasingly evident that one by one negotiation of licenses with what are seen as exceptional terms is a burden neither AMICO nor its subscribers can bear. AMICO members mayonly be able to guard against inappropriate use in an unmonitored environment by limiting the amount of data available to educators whose institutions will not agree to usage reporting. Alternatively, some technical means of control may have to be implemented, despite the fact that this was agreed to be undesirable by both parties in the MESL discussions. Some combination of two-tiered license and watermarking may emerge as the basis for licensing after the testbed year, to the regret of those who tried to find a more collaborative solution that enabled access to high-quality images.

Start-up Problems in the Business Model

Difficulties have also arisen in trying to craft an economic system. The model drafted by AMICO members at the outset of their collaboration shared costs throughout the system. Members of AMICO, in addition to providing educational representations of works of art in their collections, would pay a "membership fee" to the Consortium in return for access to the Library. Because the cost of providing the Library exceeded the income that could be generated from members, a license fee would be charged to others based on presumed intensity of use and value to their educational programs. A formula was developed, that tiered user fees based on presumed intensity of use: annual fees would be calculated for universities @ $0.25 per student, elementary and secondary schools @ $0.10 per student and public libraries @ $0.01 per card holder. Based on this formula, AMICO could become self-sufficient within five years if it obtained subscriptions from 10% of the target market.

The model is attractive for a number of reasons. It is simple: the costs of creating educational materials are born by museums as part of their ongoing programs, and the costs of providing access to that library are paid by the subscribers who want to use it. It is non-exploitative: museums pay to support the consortium and have access the Library, without the expectation of income from the licensing however successful it might be. Subscriber costs are relatively modest and could fall if penetration grows beyond the hoped for 10% of market share. It is proportionate. Large and small institutions should find participation within their means; the size of museum budgets and the number of designated users in subscribing institutions are the basis for calculation of fees.

But the model has a number of structural weaknesses that are becoming evident and may prevent its "fair" implementation. First, and most critically, the Library begins as a relatively small resource. It will grow in size, and hence in value, over time. This is a counter-incentive to early subscription. With the growth in value AMICO will hopefully attract more subscribers, but initially subscribers may feel they are paying a premium for access. To keep initial subscription costs reasonable, AMICO has forecast a negative balance sheet for the first five years. Specific ways to cover that negative balance have not been identified, although philanthropy is hoped for. Slow adoption strains the ability of the consortium to operate, and could require curtailing of services to its members, in turn slowing the growth of the library.

Subscribers, aware of their power in this situation and also recognizing the relatively lower value of the library in the early years, are seeking discounted fees. The most common approach is to propose a consortial license which, in effect, gives preferential terms to members of the consortium. Since AMICO is not itself a distributor of its resource, it views consortia which are willing to be distributors differently than some other nonprofit licensing entities such as JSTOR which has refused to offer any consortial discounts. AMICO is willing to negotiate limited-term discounts for distribution consortia, that they offer the AMICO library to 100% of their members (perhaps more than AMICO could ever reach individually). Nevertheless, the consequence is to reduce the overall income per subscriber, change the business model, and either increase required market penetration or postpone the time at which AMICO will become self-sufficient.

Archival 'Ownership'

The issue of "perpetual use" has risen to prominence in the two year lag between the end of the MESL project and the general availability of licensed access to the AMICO Library. Stung by the abandonment of back files by publishers, and uncomfortable with the prospects of paying for access but owning nothing in the end, libraries have taken the position in most licensing discussions that they must obtain rights to use every year of their subscription to a resource after they have stopped subscribing or if the publisher goes out of business. In the MESL discussions, museums emphasized the irrelevance of this provision to them, since like libraries they are in the business of ensuring long-term preservation of and access to their collections. In addition, the museums pointed to the benefits of a constantly updated and authoritative Library of digital representations, and emphasised that the digital representations are simply surrogates, and should not be confused with the real work of art housed in museum. Ultimately MESL universities grudgingly accepted these arguments, but this consensus would doubtless have broken down in AMICO discussions, but for technological change.

One of the reasons that AMICO members wanted to provide an annual license to the Library was that they wanted to ensure that updated documentation on individual works was regularly distributed. They also wanted the images to do justice to the original. The "minimum" specification of 1024 x 768 pixels and 24 bit color images (based on full screen display on a 17" monitor) seemed adequate at the time that AMICO was being planned, but eighteen months later it is clear that higher resolution monitors will soon make this level of display seem as inadequate as 640x480 seems today. Increasingly, this factor of technological obsolescence and its consequences is beginning to hold weight. University libraries may reconsider their insistence on perpetual use simply on the basis of the size of the AMICO Library; the testbed library (~20,000 works) amounted to 65GB of data leading many subscribers to realize that they will not want to mount the entire resource locally anyway.

Another major reason for the annual nature of AMICO Library distribution is that the museums are providing rights which they in turn often need to obtain from artists and artists estates. In obtaining these licenses, there may be fees that must be paid and/or assurances that must be given on an annual basis. Hence it is often not within the rights of the AMICO to grant license to the use of the resources for more than a year at a time even if they wished to.

A second alternative could possibly be offered to university subscribers that reflects this underlying constraint. A termination option could be offered in the AMICO Library Agreement which allowed universities to obtain long-term rights for educational use of some of the AMICO library data to which they had subscribed for a one-time fee. Such a separate license would apply only to those works in the AMICO Library for which the museum members had full rights. This one-time agreement would be between the universities and the individual museum members of AMICO; in effect it would be the same kind of academic use agreement made outside the context of AMICO. However, the acceptability of this "fall-back" mechanism to both the museums and the universities has not yet been explored.

Rights Acquisition

MESL made no contribution to our understanding of negotiating licenses with contemporary artists and artists estates for educational use. Seeing rapidly to build a test library, MESL museum participants avoided twentieth century works. But AMICO cannot very well take this stand. Not only does avoiding modern art result in a seriously deficient educational product, it means that many AMICO member institutions would effectively be unable to contribute since large parts, and in some cases all, of their collections are modern.

This issue is bound to become even more pressing now that legislation in the U.S. has been extended to offer the same term of protection - 70 years from the death of the artist - that has been afforded under European law. Effectively, this means that most art created in the twentieth century is covered under copyright and requires that museums obtain rights from artists and their estates in order to be able to offer use of the images of modern art in their collections. The costs to museums of acquiring the rights they need to make modern works available for use in education are going to be tremendous. Locating rights holders and conducting negotiations is a difficult and expensive project (and, of course, is the underlying reason why the AMICO Library itself was considered to be a critical resource for scholars and students).

Preliminary discussions about collaboration regarding rights have led to proposals that AMICO members, when requesting such rights from artists, ask for limited use rights with respect to any of the artists' works on behalf of all the AMICO institutions. In this way, museums with close relations to individual artists and estates could take on the task of explaining AMICO and the protected nature of AMICO licensed distribution on behalf of other members. AMICO may also enter into direct talks with artists rights organizations to streamline the rights negotiation process.

The implications of needing to acquire rights, however, are present for many similar cultural and scholarly undertakings and will have significant cost implications for electronic publishers of recent historical materials as well as for subscribers. Current calls for authors to retain rights to the works they publish will only exacerbate this problem, further distributing ownership. Subscribers need to appreciate that the rights being assembled by publishing ventures are one of their major sources of added value. At the same time they need to understand that those who offer subscription licenses often do not "own" the information themselves; regular re-negotiation is one more reason why they cannot grant a subscribing institution "ownership" of a digital resource.

Distribution Services

Ultimately, the largest issue in the implementation of the AMICO Library model is the ability to identify and attract distributors. The most radical feature of the model is that it purposefully creates redundant and competitive distribution mechanisms, rather than having AMICO serve and support access to its own publication. In principle, this model was developed to enable niche markets to obtain the AMICO Library in ways tailored to their needs. It also spreads the "first copy" costs in ways that are more economical. AMICO bears the significant costs of assembling a coherent dataset. Distributors leverage existing investments in information distribution systems to make new content available. One of the traditional ways of capitalizing on investment in the digital economy has been in the creation of derivative products, that make the same information available to larger market segments, trading lower prices for increased market share. AMICOs investigation has shown that the investment needed to distribute the AMICO Library to those larger markets is greater than the cost of assembling the research data set. In many ways, primary and secondary educators and public library users will require a much higher level of authored content and assisted retrieval tools than does the higher education sector, which is happy with access to a large database, in a somewhat raw form.

So, while the overall cost of a system involving multiple distributors to different niche markets may be higher than that of a single distributor (because of some redundancy in development) it is imagined that this would not unduly increase the costs to subscribers, and may be the only way to reach them. AMICO has agreed to work with is a not-for-profit or governmental entities, each of which, in different ways, can underwrite some of the costs of providing the AMICO Library to its constituency. By explicitly allowing, and encouraging, distributors to cross-license software developed for AMICO delivery to other distributors of the AMICO Library, and encouraging subscribing institutions to license software to make the AMICO Library more useful to end-users, AMICO hopes to lessen some of the costs of this redundancy.

Nevertheless, the true test of the model will be whether AMICO can encourage a variety of distributors to offer a range of products for accessing the educational multimedia content of museums. On this score, the jury will be out for at least another 24 months.

Conclusions

Networked electronic information distribution has changed not just how we publish, but what we publish. Despite technological demands in terms of data storage and band-width for delivery, there is a rapid growth in the availability of visual and multimedia information, from a greater number of sources. In such a diverse environment, authenticated sources of scholarly information will become increasingly important.21 Traditional models of publishing must adapt to this changing environment. New forms of information distribution, and the availability of new kinds of information not previously published, require new economic, social and technological relationships between information creators, aggregators, distributors and users. New social, economic and technical methods will be required to exploit these new opportunities for publishing primary materials.

Collecting institutions have different incentives for venturing into electronic information distribution than traditional publishers, largely rooted in their own educational missions, and guided by non-profit philosophies. They also have different levels of capitalization and different requirements for income. A sustainable system for digital distribution of primary documentation of our cultural heritage must reflect a new sharing of costs between content holding institutions and educational users. And it requires innovations in social relations, economic rewards and techniques. Such a system has been developed and is being implemented by the Art Museum Image Consortium. If its complex elements, that are being engineered and introduced nearly simultaneously, can be maintained for a few years, we believe that AMICO will provide a self-supporting model for similar collaborative ventures. We don't underestimate the difficulty of that task.

Footnotes

Informatics: The interdisciplinary study of information content, representation, technology and applications,
and the methods and strategies by which information is used in organizations, networks, cultures and societies.