EPUBs are an experimental feature, and may not work in all readers.
Sharing Visual Arts Images for Educational Use: Finding a New Angle of Repose
Skip other details (including permanent urls, DOI, citation information)
This work is protected by copyright and may be linked to without seeking permission. Permission must be received for subsequent distribution in print or electronically. Please contact email@example.com for more information. :
For more information, read Michigan Publishing's access and usage policy.
This article first appeared in EDUCAUSE Review 42, no. 6 (November/December 2007): 84–105 and may be found at http://connect.educause.edu/Library/EDUCAUSE+Review/SharingVisualArtsImagesfo/45225.
The “angle of repose,” as readers of Wallace Stegner’s Pulitzer Prize–winning novel of the same title may recall, is a geological or engineering term that describes the steepest angle at which a mass of loose, granular particles (such as a dune of sand) will remain stable and at rest. Stegner placed the characters of his work in the midst of U.S. western expansion—a period that, in juxtaposition with the novel’s title, was marked by sweeping change.
Today’s society is, of course, in another period of significant flux, though the transformative change now is more digital than physical. Digital technologies are forcing a reassessment of the rules governing how people relate to each other and how they access and share intellectual resources. Not surprisingly, one of the current constructs most profoundly challenged by technological advances is that of copyright law.
When I think of the angle of repose in the context of copyright law, I picture a sand dune, with the individual grains representing content of various types being shared and accessed in different ways. Like moisture for sand, copyright law provides cohesion for the “content” dune, giving it shape and slope—in other words, copyright law helps define the dune’s angle of repose. For decades, this dune has been fairly stable (subject to some occasional shifts), with generally accepted norms and practices developing around the use of copyrighted content. But recent technological advances are seriously testing the cohesion of the dune: they are adding not just a few grains to the top of the dune but truckloads of sand, allowing users not only to access greater amounts of content but to move it, remix it, and share it in new and unprecedented ways.
“As anyone who has built a sandcastle knows, if you keep adding sand to the top of the castle, at some point the wall of the castle will crumble.”
These technologies and the new products and services enabled by them are in many ways facilitating the development of society’s collective knowledge in the arts and sciences, thereby advancing the central premise of copyright. But, perhaps paradoxically, these same technologies may also lead ultimately to a retilting of the angle of repose, thus straining users’ ability to share and access content. As anyone who has built a sandcastle knows, if you keep adding sand to the top of the castle, at some point the wall of the castle will crumble. And when the castle wall falls down, you may not like the new shape—the new angle of repose—that is formed.
Fortunately, the addition of sand to the castle wall or dune also provides the opportunity to consider what the new angle of repose should be. In other words, what are the content-sharing norms and practices that copyright law should support or prohibit? It is in this context that I propose educational institutions need to actively rethink their current approaches to building and sharing digital image collections in the visual arts.
One effort aimed at such sharing is ARTstor (http://www.artstor.org), a nonprofit organization that makes available a digital library of images to meet educators’ needs in the arts and humanities. ARTstor works with many individuals and institutions—artists, scholars, photographers, museums, colleges and universities, and archives—to build collections for its digital library. The organization also develops software tools and other mechanisms that facilitate collaboration inside and outside of the classroom.
Still, while ARTstor offers one model for such sharing, educational institutions themselves need to be more actively engaged in collaboration in the visual arts as a means of facilitating education and scholarship.
Current Practices for Building Digital Image Collections
Digital technologies have required educational institutions to embrace a number of changes in how they supply images to their users for teaching. But in some fundamental respects, these teaching collections are being built and made available to users much as they were a century ago. In particular, colleges and universities are continuing to build collections to meet the needs of their own users without regard to the images that are being amassed by other institutions. By assembling these collections in isolation, colleges and universities are choosing not to take advantage of one of the greatest aspects of digital technologies: the ability to share.
“Educational institutions’ reluctance to share is due to copyright law and, in particular, to the uncertainty surrounding the U.S. copyright doctrine of fair use.”
In large part, educational institutions’ reluctance to share is due to copyright law and, in particular, to the uncertainty surrounding the U.S. copyright doctrine of fair use. Questions about the scope of fair use—and its application to teaching collections in the visual arts—are particularly problematic because fair use has played a critical role in the development of these collections for instruction in art history. A brief overview will illustrate the importance of the fair use doctrine in this context.
Starting in the early twentieth century, as art history gained a place in the curricula of many institutions, including museums, and as projection technologies developed, educators around the United States began to use slides to illustrate lectures. Although there is little information available about the number of slides used for teaching over the years and the rate at which such practices grew, starting in at least the 1940s, slides were being produced “by the thousands” for use in teaching. More recently, a report from 1997 stated: “It is conservatively estimated that a semester long art history course may use two thousand slides. . . . At a mid-size institution where there are ten art history courses being taught each semester, professors will use forty thousand slides a year.” In addition, a former visual resources curator at a large university told me that the visual resources department at her institution—which meets the image needs of the entire campus (not just the art history department)—regularly supplies the faculty with approximately 80,000 images annually.
Despite the significant need for slides among educators, especially art historians, few commercial sources existed for such images. Some slides were available from museums, galleries, and historic sites or monuments. Others were provided by a relatively small number of photographers who distributed slides (and now digital images) to the educational community, typically for a modest fee to cover their travel and photography costs. But these sources collectively supplied only a small fraction of the visual arts images needed for teaching (particularly since such images served as primary source materials in a broad, and ever-expanding, canon).
To supplement these sources, faculty and their educational institutions began creating their own, usually departmental, slide collections, either through their own photography or through the direct photographic capture of visual materials appearing in books, postcards, and other printed materials. The latter practice, known as “copystand photography,” was widespread in at least the early 1940s, with the development of machines and techniques that facilitated the rapid creation of glass negatives and lantern slides from the printed materials. Thus, the practice of copystand photography developed—and eventually flourished—out of necessity. Obtaining images in this manner had the added benefits of being relatively inexpensive (which was important, given the small size of art history departmental budgets) and of being able to meet the rapid pace of faculty and curricular demands.
Not surprisingly, educational institutions maintained these copystand collections, along with purchased slides, for use and reuse in other courses or in the same course when next offered, sometimes during the next term or the next rotation cycle. Institutions employed “slide librarians” (sometimes also referred to as “visual resource curators”), who were charged with (among other duties) developing, cataloging, maintaining, and providing access to these slides. The slides were incorporated into teaching collections that now—after more than sixty years of such practices—number well into the hundreds of thousands at some institutions.
Today slides are being phased out, and digital images are being used in their stead. Although the format for art images may be new, the challenges and needs of educators are the same: faculty still need images (and lots of them); budgets are still strained; and the appetite for images is still large, having now spread well beyond the art and art history departments. Resources like ARTstor, offering hundreds of thousands of images of art, architecture, and other visual materials, have developed over the last decade to meet institutions’ image needs for pedagogy in the visual arts. But as in the days of slides, no one resource can supply all of the images needed by faculty (think of the professor who wants an image of a Henry Moore sculpture, from a particular angle, and with just the right light, to illustrate a point). So institutions continue to respond to faculty members’ specific needs for images through the use of copystand photography—except that today, digital cameras and scanners capture the specific and essential image to describe or prove a point in a lecture.
Just as copystand practices developed by necessity then and now, so has educational institutions’ reliance on the copyright doctrine of fair use, both to create these collections and to use them in the classroom. As the demand for images in the classroom increased markedly, copyright law did not keep pace. As a result, obtaining permissions from the associated copyright owners has been impossible in many instances and sufficiently difficult in others that a requirement to clear all rights (fortunately, a requirement never clearly dictated by law) would have impeded pedagogy.
“Copyright law has largely been driven by a need to protect text first, with all other formats following.”
One of the reasons for institutions’ reliance on fair use was the historical development of copyright law, which has largely been driven by a need to protect text first, with all other formats following. For many years, for example, the copyright owner had to provide notice of copyright on a work to make sure it was copyrighted. This worked fairly well in the print media, and even though the laws have changed, authors and publishers have fairly standardized means of indicating their copyright ownership in a textual work. But the same traditions did not develop in the visual arts. There may be multiple reasons for this. According to one source, indicating copyright on a work was traditionally considered too crass and commercial for creators of fine art. Other artists simply ignored the copyright registration requirements as not applicable or necessary for art that was already unique and singular, that was not one of a multiple or an edition. In addition, there are few physical spaces on visual objects where one can identify copyright ownership without interfering with the integrity of those objects. And for some artists, indicating copyright ownership would be antithetical to the aim of their works (think of Marcel Duchamp’s Fountain signed “R. Mutt.”). Regardless of the reasons, innumerable artworks, photographs, and other visual images do not identify their creators (and likewise do not indicate copyright), making the permission-seeking process daunting, if not impossible, in many cases.
Compounding this problem, there are few registries that identify copyright owners, and those that do exist have largely been inadequate in providing notice about the copyrights in visual artworks. So even when there is a registry, the task of associating an object or an image with the proper rights holders is very difficult and time-consuming. Moreover, there are no meaningful rights clearinghouses or centralized licensing mechanisms on which individual educational institutions can rely for images used in teaching. Indeed, even the U.S. artists’ rights organizations—which represent thousands of artists—have not been in the business of granting permissions to individual educational institutions or faculty for the use of images in the classroom.
As a result, in the vast majority of cases, if a faculty member or visual resources librarian wants to obtain permissions to use images for teaching, he or she will have go back to the individual sources for each image (assuming they can be identified), obtaining clearances in many instances from the photographer and, if the object depicted in the image is under copyright, from the artist of the work itself. Once the copyright owners are identified, the often arduous and sometimes impossible task of locating them has to be undertaken. And even assuming they can be found, most artists, estates, and photographers do not traditionally license images for teaching, and there is no streamlined permissions process. It is not that they are uninterested in having their works represented in the classroom (and becoming part of the teaching canon) but rather that they are unaccustomed to addressing these permissions questions in the teaching context. Therefore, when permissions are sought for use of those images, there is often no response, and when there is one, obtaining permissions is rarely a simple, quickly negotiated solution. Indeed, because there is no centralized licensing process, obtaining permissions requires addressing the varying needs of artists, museums, and photographers, with some being concerned about the context in which their works will appear, others worrying about the quality of their works, and still others wanting to make sure the environment is limited to education. As a result, seeking permissions often entails a series of conversations rather than a simple, one-time transaction.
Now imagine seeking permissions for thousands of images—or hundreds of thousands of images—to support the teaching needs of art history faculty, as well as a short turn-around time so that faculty can develop and modify their lesson plans as needed, and the importance of fair use to teaching in the visual arts becomes obvious. As one report estimated in 1997, the average slide librarian would need forty years to obtain permissions for a single year’s worth of acquisitions at a mid-sized university (6,000 images).
The Consequences of Having to Rely on Fair Use
Given that there is no efficient licensing mechanism for the vast majority of visual arts images for teaching, that many copyright owners of visual artworks cannot be identified or easily located, that the use of images in the classroom has become a mainstay of pedagogy in art history (and other fields in the humanities), and that learning in the arts and humanities would be severely hampered if permissions were required, one might think (as I do) that educational institutions’ copystand practices clearly fall within fair use. Indeed, that is why ARTstor relies on fair use in making some images available through the ARTstor Library.
However, because uncertainty continues to surround the fair use doctrine and because there has been no legal determination that such practices fall within this copyright exception, most educational institutions have relied on the doctrine rather uneasily, developing a kind of “don’t ask, don’t tell” policy. This has meant, not surprisingly, that institutions have resisted the idea of sharing their image collections with other institutions and instead have opted to build collections to meet their own users’ needs.
At first blush, this might seem like the wise course. After all, Viacom’s decision to sue Google for more than $1 billion because of copyright infringement on YouTube is just one of the latest examples of the ongoing legal challenges to peer-to-peer models. Moreover, the one major effort to develop guidelines for copyright owners on what practices constitute fair use in the educational context—the Conference on Fair Use (CONFU), convened by the U.S. Patent and Trademark Office—failed to achieve consensus.
Nevertheless, there are several disadvantages to maintaining the current, siloed approach. First, there is the obvious problem that some collections will remain locked within one institution’s walls. The unavailability of these collections has the potential to adversely affect the richness of scholarly discourse and education in the arts and other fields, since any individual institution’s collection will have far less depth than a shared resource. The lack of access to these images may be aggravated by the unavailability of other materials that, though not perhaps significant in and of themselves, gain value because of their ability to provide context. This “the whole is greater than the sum of its parts” type of work is what has been so important for Web sites like Wikipedia. Compounding this problem is many students’ assumption that if content is not online, “it doesn’t exist.” Therefore, by continuing with the current, siloed model, institutions may be “curating by default” what students are learning in the arts and associated fields.
Another obvious disadvantage to not sharing is the duplication of efforts and other costs across institutions. It is well recognized that the development of software systems to maintain and access digital content is expensive. Building a digital image library means not only scanning the images but also developing an infrastructure to hold the images, licensing or building software to access the images in the database, hiring a database architect to manage the database and to map collections to the database structure, migrating the software as other technologies emerge, devising metadata schema and tagging the images so that they can be found, and implementing access and archival policies for the images. With these high and continuing costs of “going it alone,” some greater degree of collaboration across the community to redistribute or share these expenses might make sense.
Third, an institution’s collection of copystand images may not be of particularly high quality and may not have correct attributions. Although scanners have improved in recent years, sharing images across the educational community could allow educational users to access the best-quality images and data available within the larger pool of content. As discussed in more detail below, it is the poor quality of images that is most troubling to many artists, photographers, and museums today—rather than the lack of revenue generated from the use of those images in teaching (which is often mentioned in other contexts).
Fourth, and perhaps most important from the perspective of copyright, the reluctance to share with other institutions may affect educational institutions’ collective ability to rely on fair use in this shared setting. By making the decision not to share collections with other educational institutions, colleges and universities may be suggesting to courts that such sharing is not fair use. In determining what constitutes fair use, courts may look at community practices to help guide their analysis, and if those community practices do not accommodate sharing, courts might believe that this was a reasonable limitation of that doctrine.
“The collective understanding of copyright is being informed only by the Napster and Google disputes.”
This leads to another concern about the current, siloed approach: the problem of not engaging content owners in the dialogue about sharing digital visual arts images for teaching. The absence of meaningful dialogue has had adverse consequences. One such consequence is that the collective understanding of copyright is being informed only by the Napster and Google disputes, even though the music industry has a very different “angle of repose” from that of the visual arts community. In discussions I have had with artists and their representatives, for example, many of them have expressed an interest in making images of their works available for teaching and study but fear that those images may be used inappropriately by educational users. Not surprisingly, those concerns are based on the press surrounding illegal music downloads—even though the ways in which digital visual arts images are being accessed and used on campus are very different from how popular music is being shared.
Likewise, educational institutions frequently conduct risk assessments about the use and sharing of visual arts image collections as if copyright owners in the visual arts have the same interests as the record labels. This may lead to “knee-jerk reactions” of risk aversion even though museums, artists, and photographers may want their works represented in the teaching canon, and it may lead also to the “locking down” of collections even when the copyright risks of sharing those collections are low. For example, ARTstor sometimes has to provide assurances that it will take all the legal risks of sharing unique, institutional collections that would be of value for teaching—even though those collections consist of images of older, historical works that are very likely in the public domain—because doing so is the only way that institutions will agree to make the collections available to the broader community.
Another disadvantage of the current, siloed approach is that content owners often have unique, unpublished image collections that would be of tremendous scholarly and educational value but that are not being shared with the educational community both because of fears of abuse and because of the lack of any easy mechanisms for sharing. For example, ARTstor engaged in discussions with one prominent artist’s estate for more than a year before the estate was comfortable with the idea of sharing its high-quality images with the educational community. But now that those hurdles have been overcome, the estate has expressed an interest in sharing some never-before-published works as well.
Finally, if content owners are not engaged now in a dialogue about the sharing of digital image collections, it may be increasingly difficult to do so in the future. The divide between copyright owners and users is growing—not decreasing. As this polarization continues, the likelihood of being able to initiate constructive dialogue and to calm fears—on all sides—seems increasingly remote.
Toward a New Angle of Repose
There are a number of reasons why a new model based on sharing could work. The new model draws some parallels with the angle of repose in a sand dune.
One of the interesting things about a sand dune is that if each grain of sand is not acting as a loose, granular particle but is instead bound together with the other grains—because, for example, more moisture has been added—the angle of repose changes. In other words, because of the greater cohesiveness of the sand grains, a much higher sand dune can be built, and the dune can withstand increased pressures as more sand is added to it.
Similarly, I believe that the best approach to building digital visual arts image collections for teaching is to create a new angle of repose, one brought about by greater collaboration, both among educational users and between educational users and content owners of images. Educational institutions should be actively rethinking how they are accessing and using copyrighted images, and they should be exploring—with those copyright owners who are most receptive—an approach that seeks to address copyright owners’ interests as well as users’ needs, in an environment that encourages increased, shared access to visual arts images for teaching and study.
Several factors make me think that this approach could succeed. First, unlike the markets for music, popular movies, and books, the market for digital visual arts images—especially for those images that would be most useful in the educational context—does not have many big commercial players. Instead, the world of digital visual arts images is mainly populated by individual artists and their representatives, by individual photographers, and by museums. And all of the artists, artists’ estates, photographers, and museums that ARTstor has talked with—including major artists and estates (such as the Lichtenstein, Warhol, and Pollock estates), more than fifty museums, and many photographers—have been supportive of making their works more broadly available for teaching and study. They want their works to become part of, or to remain in, the teaching canon. They want more recognition and exposure for lesser-known works, and they want higher-quality versions of their works available for teaching and scholarship. And they see the distinctions between commercial and classroom use—both in terms of fulfilling their own missions and in terms of generating revenues.
The visual arts community is different from the entertainment and book markets in a number of other ways as well. In the commercial context, for example, there has typically been a clear differentiation between traditional content owners and consumers (most people, for example, are consumers of movies but do not create their own commercial films). But in the visual arts context, the distinctions between users and creators often blur. There is a long, albeit sometimes contentious, history of appropriation in the visual arts, with many artists using others’ works for their inspiration. And of course museums straddle the divide between image users and image owners, since they produce publications using others’ images and also license images of works from their own collections for a variety of uses such as publications and commercial merchandising. Given that the lines between users and owners of content often blur in the visual arts world, consensus may be more easily reached here than in the commercial context.
Additionally, there are signs that a collaborative approach in the visual arts community could work. In the most recent and compelling example, the Metropolitan Museum of Art approached ARTstor about distributing, for scholarly publications, images of works from the museum’s collections. In this case, the Metropolitan Museum sought out ARTstor because it wanted to do something positive to address the state of scholarly publishing in the arts and the high fees that scholars have paid to use images in publications. The museum’s images are now being made available through ARTstor to both ARTstor participants and nonparticipants for use in scholarly publications, free of charge. While this effort is groundbreaking in many ways, one of the most heartening aspects is that, in this charged copyright environment, a leading museum has “crossed the copyright divide” and taken the initiative to help meet the needs of scholars, even when the museum is incurring costs to do so.
Another reason a collaborative approach might work is that such an approach benefits copyright owners in ways that the current, siloed approach does not. In the various discussions that ARTstor has had with museums and with individual artists and their estates, for example, one of the most important issues has been ensuring the quality of images in teaching collections—how well those images represent their artworks—and having correct attributions and cataloging data to accompany those images. One of the problems with the current copystand practices is that many of the images in these collections are of poor quality or do not have accurate credit lines or consistent data associated with them. A shared system that engages artists, museums, and photographers might allow educational users to meet some of these core interests of copyright owners while also better fulfilling their own pedagogical needs.
All of these factors lead me to think that there is some hope for a collaborative approach, one in which ARTstor and similar resources would be but nodes of a broader educational network that supports the shared use of images for teaching and scholarship. This approach would mean engaging in an effort to bring together the two halves of this relatively small community to see if some common understandings could be reached and building on a model that openly embraces fair use in a shared context. Of course, many other aspects would need to be addressed, such as technological, curatorial, and archival questions. But I believe that such collaborative efforts are the way to create a new angle of repose—one that is uniquely formed around digital visual arts images and that meets the needs of both educational users and copyright owners.
Gretchen Wagner has been the General Counsel of ARTstor since its founding in 2002. She devised and implemented ARTstor’s intellectual property policy, which ARTstor has relied upon in building relationships with a wide range of copyright owners and constituencies, including artists, photographers, museums, archives, scholars, and publishers. She has overseen the licensing of more than 100 collections for the Library, representing over 1.8 million images, while also relying significantly on the US doctrine of fair use in sharing this content with more than 1,000 academic and cultural institutions. Before joining ARTstor, Gretchen was the Assistant General Counsel of The Andrew W. Mellon Foundation, where she helped articulate the Foundation’s intellectual property policy and drafted numerous agreements, including with the Government of China and prominent cultural and academic institutions in Europe and the United States. Gretchen obtained her B.A. from Yale University and her J.D. from Columbia University School of Law, where she was an editor of the Columbia Law Review and a Harlan Fiske Stone scholar, and received recognition for her pro bono services to the community.
Comments on this article can be sent to the author at firstname.lastname@example.org.
Philip C. Beam, “The Color Slide Controversy,” College Art Journal, vol. 2, no. 1 (January 1943): 37. See also Donald N. Wilber, “The Nature, Care, and Projection of Color Transparencies,” Bulletin no. 7, September 20, 1940 (Color Slides Cooperative, Princeton, N.J.), 3.
Virginia M. G. Hall, “Fair Use or Foul Play? The Digital Debate for Visual Resources Collections,” presentation, Association of College and Research Libraries, New England Chapter, conference, November 7, 1997, available at http://www.vraweb.org/resources/ipr/papers/fairfoul.html, 3.
Ibid.; Christine L. Sundt, “Testing the Limits: The CONFU Digital-Images and Multimedia Guidelines and Their Consequences for Libraries and Educators,” Journal of the American Society for Information Science, vol. 50, no. 14 (1999): 1329. See also Patricia Sloane, “Color Slides for Teaching Art History,” Art Journal, vol. 31, no. 3 (spring 1972): 276.
Sundt, “Testing the Limits,” 1329; Beam, “Color Slide Controversy,” 37; Wilber, “Color Transparencies,” 3. See also Lester C. Walker Jr., “Low Cost Slide Production for Teaching Aids,” College Art Journal, vol. 13, no. 1 (fall 1953).
Some institutions have also sought to rely on the classroom exceptions to copyright, such as the TEACH Act (see section 110 of the U.S. copyright statute), although the limitations contained in the TEACH Act have led many institutions to rely on fair use instead. See “The TEACH Act Finally Becomes Law,” http://www.utsystem.edu/OGC/IntellectualProperty/teachact.htm.
A few efforts to obtain permissions from all of the copyright owners to use visual arts images for teaching were abandoned as impractical. See, e.g., Lisa J. Hawkos, “Copyright Compliance in the Visual Resources Collection,” VRA Bulletin, vol. 28, no 2 (summer 2001).
The Art Museum Image Consortium (AMICO)—a consortium of museums that made available a digital library of art images for teaching—may have been the first organization to enter into agreements with the U.S. artists’ rights organizations (starting in 1999) for permission to make images of their artists’ works available for teaching. Following that precedent, ARTstor—which makes the ARTstor Digital Library available in countries where there is no fair use copyright exception—entered into agreements with these organizations as well (following lengthy negotiations), and a few educational consortia may have followed suit. To my knowledge, the U.S. artists’ rights organizations have not entered into any agreements with individual educational institutions for the use of visual arts images in the classroom.
I am not suggesting that the reproduction of every image would be fair use: reproducing, without permission, those images that are sold by photographers to campuses for educational use might unfairly harm those photographers’ markets. But educational institutions have developed community-based practices that both reflect their own needs and discourage the copying of images that would unfairly affect such markets. Two resources that describe such community-based practices and help individual faculty and visual resource curators follow those practices are the Copy Photography Computator (developed by Allan Kohl) and the VRA-developed Digital Images Rights Calculator, both posted on the VRA Web site: http://www.vraweb.org/resources.html.
Not surprisingly, because campuses are not sharing their copystand collections, few institutions have developed mechanisms to share their unique collections, even when those collections are free of copyright issues.
CONFU did not result in meaningful dialogue because the meetings were government-generated and therefore fostered a sense that everyone needed to be protective of their interests and, according to one CONFU participant, because of the failure to reach consensus with the commercial intermediaries invited to participate in those discussions, even though many of those intermediaries were not themselves the copyright owners of artworks or photographs.
Because there is relatively little appetite among students for visual arts images (in contrast to their appetite for popular music and videos), it seems unlikely that such images are being shared by students in great numbers through peer-to-peer systems or social software sites on the Web. Instead, students access the collections of individual faculty or institutions as they would many other institutional resources: through specific access points made available by the institution, for specific assignments and study purposes.
I am not suggesting here a completely licensed model, since I do not think that would be feasible, let alone advisable. Instead, institutions should be exploring how a shared system could actively support fair use for teaching in the visual arts while adopting measures that seek to respect copyright owners’ commercial markets and other interests.