Sorcerer’s Apprentices in an Imaginary Museum: The Critical Edition in the Internet Age

“The computer is here to stay, therefore it must be made to adapt to the librarians’ problems and needs. It must be kept in its proper place as a tool and a slave, or we will become sorcerer's apprentices, with data data everywhere and not a thought to think.”

—Jesse H. Shera, “Librarianship and Information Science,” in Fritz Machlup and Una Mansfield, The Study of Information: Interdisciplinary Messages (New York: John Wiley and Sons, 1983), 384.

“The view of the musical world the romantic aesthetic originally provided has continued, since 1800, to be the dominant view. This view is now so entrenched in contemporary thought that its constitutive concepts are taken for granted. We have before us in fact a clear case of conceptual imperialism.”

—Lydia Goehr, The Imaginary Museum of Musical Works (1995; rev. ed. Oxford: Oxford University Press, 2007), 245.

As an undergraduate guitar major at the University of Akron in the early 1990s, I entered the library with a seemingly straightforward task: to find the score of Johann Sebastian Bach’s third cello suite, which I would be learning on guitar that semester. The card catalog had by that time been superseded by an online catalog, into which I entered these terms in various combinations. But although I received numerous “hits,” none seemed to promise what I was looking for; instead, I found a confusing multitude of recordings, arrangements, and other loosely related items. My guitar professor solved the problem by steering me away from the catalog altogether, instead physically leading me to the stacks where the volumes of the Neue Bach-Ausgabe sat, promising “the actual music” in safely curated and specifically titled isolation. “Don’t even use the catalog,” he said, “just remember where these are and come here.”

That solution worked only for that composer in that library, but the larger lesson remains pertinent: libraries are most effective when searched with help, whether that help comes from a person or a reference book created for the purpose.[1] Prior to the establishment of the online catalog, of course, I would have needed to refine my search terms myself before even approaching the shelves holding the physical card catalog: the ease and power of the online catalog had lured me into a mindset that allowed for a broader semantic representation of the piece of music, “Bach’s third cello suite,” than I had actually wanted.

And now, in 2020, as an editor of critical editions, I wonder about the ramifications of that change of mindset, a universal change now that readers use general internet searches to interact with musical scores, while online library catalogs incorporate data from sources outside their own holdings. The quotes above point to two general hazards inherent in the enterprise of creating critical editions. Shera’s prescient warning regarding the proliferation of data is all too applicable to the internet age. But, lurking just behind the dangers foretold by Shera, I also see Goehr’s description of the “work-concept,” an implicit conceptual bias in the world of classical music, and one that has always haunted the validity of the critical edition as a meaningful endeavor. To no one’s surprise, I am sure, I will argue here that the critical edition has not lost its validity. However, if we are to respond to the challenges posed by the current environment (and articulated by Shera), we must also reckon with Goehr. Will the editors of critical editions be sorcerer’s apprentices, generating data in the service of data? Or custodians in the imaginary museum, shooing noisy patrons away from timeless works that do not actually exist? Or can we instead adapt to the necessary role of shepherds, serving other humans, the musicians, scholars, and readers, for whom the readily available sea of data might be a boon, given guidance relevant to the task at hand?

My example of the Bach cello suite (what I sought, of course, was the cello suite in C major, BWV 1009, usually played in A or G by guitarists) illustrates an ontological shift happening under our very noses, as more and more we turn to internet search engines to interact with music. The “work” I sought in the online catalog had become, in the seeking, a description—the “work concept” transforming into a “search term.” This change has important repercussions, but the problems described by Goehr persist; in fact, in certain senses they are magnified by the World Wide Web. Consider that the “search term” is instantiated according to horizontal relations between data sets, producing a list of results, each connected according to the calculus of the search engine: the results are those mostly likely to elicit a click from me, as determined by all the information the search engine provider has been able to accumulate about me up to and including the search term I just entered, and will therefore be most effective in directing my attention to advertisers and allowing the search engine provider to accumulate more information about me, which can then be extrapolated to others entering the same or similar search terms.

Whatever image of the “work” ultimately emerges after entering any given search term—such as “Three blind mice,” “Dákiti,” or “Beethoven’s Fifth Symphony”—now owes its bewilderingly multifaceted complexity not so much to the ubiquity of those terms in our culture as to the incredibly wide net that the speed and facility of an internet search allows the engine to throw out. While my search for “Bach cello suite no. 3” in the library card catalog owed its futility mainly to the popularity of the piece and the composer, internet searches by their nature will produce similarly varied and confusing results for almost any musical composition; the very power and flexibility of the medium demands it. Who or what, then, can, like my guitar professor, take us by the hand and lead us through this sea of data? The “imaginary museum” described by Goehr now resembles a mini-mall, or perhaps a chaotic open-air market, where datapoints are valued equally in their instantiation but not ordered according to categories such as “true,” “false,” “before,” or “after.” For example, attempt a search for the Jesse Shera quote that heads this essay: you will get many results, but few that actually mention Shera, and far fewer that give the title of Shera’s essay or the collection from which it came. Most, instead, refer to the mathematician John Allen Paulos, including a familiar social media “face and quote box meme” giving Paulos’s image above the quote (see fig. 1).

Fig. 1. Shera quote erroneously attributed in a social media posting.

To be fair, Paulos credits (but does not cite) Shera in his own use of the quote.[2] But the journey back from the Paulos meme to the Shera essay is not easily done via websearch. The connections that are important to the search engine parameters have to do with the pithiness of the quote, its wide applicability, and its capacity to attract attention; mine is by no means the first blog entry to open with it.[3] These connections are horizontal, or synchronous, and the diachronous (i.e., occurring in sequence) steps associated with creative activity—inspiration, notation, transmission, quotation, and documentation—are elided or obscured, not important to the business at hand.

So, do we need a new museum? Shall we take the work-concept, so aptly defined and illuminated by Goehr as a historically contingent, cultural creation, which by the end of the last century was already losing whatever efficacy it might have had as a tool for the organization and promotion of classical music, and reify it, creating a stronger edifice that can withstand the onslaught of data that trembles in readiness at each mouse click? Is this, in short, the job of the critical edition in the internet age?

It is not. As producers of critical editions we need instead to focus on context, in service to the reader, as a guide through the potentially overwhelming availability of data. This has always been the case; the difference is that now there is no way to duck the issue without become lost in the data ourselves. The circumstances that necessitated both Shera’s warning and Goehr’s analytical takedown of canonical ontology have only been exacerbated by this data-hungry environment. Shera’s essay, published in the 1980s, was written in the context of a methodological split in librarianship, between information technology, as the term was then used, referring to the (primarily American) trend towards reproduction of library materials via film (microfilm, microfiche, etc.) and the European approach, which favored documentation over reproduction. The American trend was powerful enough to envisage a library in which such reproductions took prevalence over physical books; a radical concept at the time, but certainly not so now. As a society we now access our information from screens, via globally shared pools of data, in forms that have been digitally reproduced, much more than by any printed media. The only question is, are we the custodians and beneficiaries of that data, or have we become the “sorcerer’s apprentices” foreseen by Shera?

And what is the status of the “work,” now that any piece of music is likely to begin its relationship with any individual human being as an internet search term? Have we gone so far from the 220-year-old “work-concept” that Goehr’s “imaginary museum” can be put in its own museum, safely ensconced as another search term for busy students to enter in their late-night bibliographical web crawls? But, like Shera, Goehr’s admonitions cannot be set aside so lightly. The move from “imaginary museum” to “imaginary mini-mall” has not freed us from the philosophical burden of the work-concept, nor from the responsibility to acknowledge and counter that burden, especially as regards the critical edition.

In fact, critical editions have long been the site of negotiation between reproduction and documentation—the very issue that prompted Shera’s catchy warning—and the conceptual issues surrounding that negotiation are the very heart of Goehr’s analysis of the work-concept. The critical edition exists by virtue of its superiority to scores that lack the critical apparatus, in such cases where that apparatus is considered important; for example, where there might be some question as to the true nature of the work, defined by the presence or absence of various notational elements. The nineteenth-century concept of Werktreue, targeted by Goehr as the germ of the Romantic work-concept, would, by this definition, also seem to be the defining characteristic of the critical edition. As Goehr put it: “Too few persons worried about the implications of thinking about music in terms of works precisely because, socially and musically, the work-concept seemed to be the only packaging device around.”[4]

And this is why we need both Shera and Goehr’s warnings. In particular, we should attend to Goehr’s use of the term “cultural imperialism”: the work-concept, according to Goehr, is not only a cultural construct, it is regulative. It creates an illusionary idea and compels conformity to that idea, all in the guise of ontological certainty: we make decisions about a piece of music, not because we feel that the music should be that way, but because we assume that a “piece of music” is that way. These decisions can range from the placement of a dynamic marking to the inclusion of composers of color in an anthology. And, like the algorithms that govern the terms brought up by a web search, the work-concept is more synchronous than diachronous: the “work of art” that it defines exists outside the restraints of time, of messy cause-and-effect relationships created by human activity such as notation, performance, and publication.

For a time, the critical edition as a commodity had enough currency, in the world of academic scholarship, that publishers could sidestep uncomfortable questions by presenting Urtexts whose evident scholarly credentials provide their own authority. But perspicacious scholars and critics have long pointed out the dangers of assumed authority in historical inquiry. I recently had the pleasure of sitting in (via Zoom) on Marianne Kordas’s music bibliography class at Andrews University, where she discussed some of these same topics with her students, specifically the problems in finding suitable scores via a general web search. As part of an introduction to the general idea of the critical edition, I brought up Winton Dean’s 1965 review of Fritz Oeser’s critical edition of Georges Bizet’s Carmen.[5] Oeser’s stated goal in that edition was to present Carmen in its “authentic” form, including passages existing in manuscript that had not thus far appeared in published versions, as they had been cut during the course of preparation for performance.[6] Dean pointed out that most of those cuts were made by the composer himself: the first edition of the Choudens vocal score, which was printed contemporaneously with the first performance, was prepared according to a score arranged by Bizet, and from proofs that Bizet checked. Now, it is possible that some of the cuts, even though made by the composer, were made unwillingly, only in response to the inferiority of the performers and conductor available to him, and that an ideal version of Carmen should therefore restore them all. But there is no actual evidence that the cuts were not instead made for artistic reasons. Oeser, however, assumes that any contextual contingency can only represent a diminishing of the “true” work—his work-concept cannot allow the intrusion of cuts made for the purposes of performance, even those made by Bizet himself. This puts Oeser in the unusual position of disagreeing with Bizet about what the “authentic” version of the opera actually is, an irony not lost on Dean. Dean’s prescription for avoiding the trap of the “work-concept” in this particular case is worth quoting in full:

He [the editor] needs to bear at least four considerations in mind. (i) He should distinguish clearly in the text between what was performed in March 1875—the only production during Bizet’s life—and what was not. (ii) He should endeavour with the aid of all available evidence to determine the composer’s wishes. (iii) He should separate fact from conjecture and avoid a priori assumptions about the motives and intentions of everyone concerned in the production of the opera. (iv) Having adopted a policy as to what will be included in the main text, the notes, and the appendices respectively, he should abide by it.[7]

So, as early as 1965, we have both an implicit recognition of the “work-concept” problem, if not in so many words, and advice on dealing with it. Notice that not only does Dean’s prescription engage with the details of historical and chronological context that the work-concept avoids, it would, if followed, also provide a foothold amid a large quantity of data. Like scattered iron filings aligning along magnetic poles, the data available for any given piece of music can, under the influence of a well-constructed critical apparatus, be organized according to chronological, historical, and creative context, with assumptions, doubts, contradictions, and areas of incomplete knowledge duly noted.

Furthermore, Dean’s emphasis on context accords well with Jesse Shera’s admonition, not twenty years later, to the librarians newly armed with data reproduction and transmission technologies. Shera posits three components for the concept of the library: (i) acquisition; (ii) organization (the only one, he notes, where information science makes a positive contribution); and (iii) information and service—this last being the library’s raison d’être, what he calls “the sphere of maximum context.” The library, Shera reminds us, “is a creature of society, evolved to meet the needs of human beings working toward the solution of certain problems.”[8] Notice the emphasis on context in both of these solutions: like the work-concept, the “search-term” tends to spread the data evenly, with little to organize it, excepting those things that are of use to the data providers. But a human, be she librarian or editor, can provide a vehicle to survive in the sea of data, a perch from which to survey and utilize it.

And what of Lydia Goehr’s advice? As a philosopher, she is perhaps less focused on solving the problem than she is on analyzing it, but she does realize the potential for good that could come from a clear-sighted recognition of our devotion to the Romantic “work-concept.” With typical philosophical rigor, however, she shies away from easy answers, noting that “To fully dismantle the force of a concept, one needs a global paradigmatic shift, of at least equal complexity to the shift that founded the force of the concept in the first place.”[9] But with the internet age, perhaps such a shift is indeed taking place. And although the imaginary museum may yet loom as a portal into which our musical heritage could disappear, tended only by sorcerer’s apprentices in service of global data aggregators, there is another option for the critical edition. The critical apparatus of bibliography, source descriptions, editorial methodology, and critical notes, if carefully and ethically maintained, can now assume the role it should always have had. We must take the advice of those who have seen this moment approaching, and turn our attention to context: not only to the historical context of the music and the bibliographical context of its sources, but, most importantly, to the people who will read, perform, and study the music, and how best to serve them.