The Size of Thoughts

by Nicholson Baker

  And if you persist in wanting to perform an online subject search, prepare for real uncertainties and dashed hopes and deep screen-scrolling tedium. The big technical push in the early development of online catalogs was for “known item” searching, in part because some card-catalog studies seemed to show that library visitors don’t do all that much in the way of subject queries. Most of us come to the library, it was thought, with a specific writer or title in mind. In a monumental yearlong survey published in 1970, Ben-Ami Lipetz and his research assistants approached 2,134 pedestrians in the area of the card catalog at the Sterling Memorial Library at Yale, clipboard in hand, and politely said, “Please tell me precisely what you were about to do at the catalog the moment I interrupted you.” Eighty-four percent had a title or an author or a specific bibliographic goal in mind, and only 16 percent were interested in browsing a set of subject cards. At least, that is what the respondents said. Subject searches are somewhat embarrassing, especially if you’re a graduate student, and a graduate student at Yale (Yale graduate students used the catalog more than undergraduates, according to the survey, and faculty used it least): it sounds better to say that you want to take a second look at some curious laudanum stains on the endpapers of the Surtees Society’s Catalogi Veteres Librorum Ecclesiae Cathedralis Dunelm than to say that you’ve got to quick get some call numbers for a whole bunch of books under the heading of “Feudalism.” Subject searches are obvious confessions of ignorance; author-title searches aren’t, or aren’t so straightforwardly.

  After online catalogs began appearing, though, it was possible to analyze searches without interviews, by studying the computer’s transaction logs. And scholars like Pauline Atherton Cochrane, of Syracuse University, found that people were very interested in subject searches—more interested now, possibly, because they (erroneously) thought that the computer would be better at them than cross-referenced card catalogs were. Sadly, online catalogs are still terrible at subject searches; their noise level, in the informational sense, is incredible. Card catalogs have the sense not to shuffle together alphabetically the myriad subheadings for “labor” in the medical sense and “labor” in the AFL-CIO sense; the online catalogs I’ve seen don’t. Card catalogs don’t lump subheadings for traffic control in Alexandria, Virginia, together with ones for the lost library at Alexandria, Egypt, either. The “See also” card at the beginning of a set of subject entries in a card catalog didn’t yank you to another part of the catalog, and abandon you there; it just suggested that you might want to expand your search in various directions if you didn’t find enough where you stood. There is nothing like this yet in any but the best online catalogs, and even these have peculiarities. (If you look up “Greyhounds” on HOLLIS you won’t be advised to see also “Dogs”; and if you look up “Dogs,” and choose selection 1, which “retrieves information on the use of the above headings,” you will puzzle over this brief note: “subdivision Dogs under groups of Indians, e.g., Indians of North America—Dogs.”)

  Nor is there an equivalent for guide cards—those beauties that stick up above the rest, with typed headings—which form a sort of loose outline of knowledge built into the trays, and help you to keep your bearings, and teach you as you go. Card catalogs are “precoordinated,” whereas online catalogs are still almost entirely “postcoordinated,” which means that the burden of figuring out how the universe of subjects ought to be organized has been shifted away from the cards and onto you, the user, who must now master Boolean “AND NOT” filters and keyword trickery and crabwise movement by adjacent call numbers merely in order to block avalanches of irrelevancies.

  I have no doubt that it will all get better. That’s the wonderful thing about software: it gets better. Soon it will be possible to browse big subjects like the Bible or Film or the Mafia or Pollution in a productive way, as we formerly could, instead of despairing when we get a message saying YOUR SEARCH RETRIEVED 1,028 RECORDS. A thousand records, remember, is only a little more than one card drawer. My fingers could arpeggiate Lisztlessly through them, the lead hand’s fingers feeding card clumps to the trailing hand, scanning, rejecting, repositioning, in a minute or two. But online, a thousand records is instant death right now. “Studies of online catalog use and users have uncovered a pair of problems that seem to be endemic in today’s online catalog systems,” Ray Larson, a professor at Berkeley’s library school (or, as it is now called, the School of Information Management and Systems), wrote conclusively in 1991:

  These are: (1) a large percentage of subject searches fail to retrieve any bibliographic records; and (2) when subject searches succeed in retrieving records, they often retrieve too much material for the user to evaluate effectively.

  Larson cites an earlier study of his own in which he found that the average number of online records retrieved in his sample was 77.5, whereas the average number that users actually took a look at was 9.1. The “futility point” in online searches—the point at which you give up and go with what you have—is, because of screen fatigue and the sequential lethargy of system response, much lower than in card-catalog searches. The life of the mind suffers as a result.

  Not so for experienced onliners, you may contend—those who have developed a feel for Library of Congress subject headings, and may even have the now indispensable four-volume Library of Congress Subject Headings open beside their keyboard—but if this is the case it is because, as Professor Larson neatly points out, “part of the experience acquired by ‘experienced users’ is frequent search failure and information overload when subject searching.” Larson goes as far as to say that online catalogs “are, in effect, conducting a program of ‘aversive operant conditioning’ against subject searching by their users.” You could, in fact, go Larson one better by hypothesizing that online catalogs are acting to reinforce the tendency toward mindless academic hyperspecialization, since any attempt to venture into areas in which one is a novice, using the library’s basic search tool, is met with the sharp electric shock of a long search warning. Certainly it is truer now than ever that if you want your scholarship to be read, you had better find a way to maneuver it into the early letters of the alphabet, because online, nobody’s getting much past the “G”s and “H”s.

  Software engineers are clever and adaptable people. They have, more than most of us, profited speedily from their mistakes. We have leaped ahead from, say, the late-seventies computer catalog at the University of Toronto Library, which reportedly included a program to rotate any “Sir” beginning a name entry to the end of the name: the program not only moved all the stand-alone “Sir”s but bestowed involuntary knighthood on writers like Ernest Sirluck (who was a professor at the University of Toronto at the time and an editor of, among other things, Milton’s Areopagitica), turning him into “Luck, Ernest Sir.” Any day now, information retrieval will be a new and amazing experience: “For the first time the catalog user will be freed from the tyranny of the linear sequences of A-Z and 0–9,” Michael Gorman headily writes, in an anthology called Closing the Catalog. We will cast off our letter fetters, he seems to promise, and soar. Envisioning what Gorman calls the “New Jerusalem” of the online catalog, I can fantasize about one that would include an aging-and-fading component, so that the older a given database record is, the yellower (or pinker, or ocean-bluer) the screen would be at its corners, thereby offering us some of the instantaneous secondary information that cardboard offers us now. (Word-processing packages would also benefit from what might be called AGE or OLD utilities—for Advanced Geriatric Engine and Optional Latent Dogearing, respectively.) And the more times a bibliographic record is called up by the users of the database, the darker will be the accumulation of random “grime pixels” in the top margin—though never so dark that they would interfere with legibility, of course, and every tenth retrieval might remove one grime dot rather than add one, since handling wears away previous deposits, too. We will be able to tour mind rooms full of three-dimensional representations of catalog cabinet
s by gesticulating with our data gloves like armchair Shivas, so that we will have some intuitive sense of the size of the collection we are interrogating, just as we do now when we walk into the lobby of a library we haven’t visited before and size up its rows of card cabinets; we will have ways to maintain a sense of where we are in the database, as we do now through drawer and cabinet labels, through guide cards, and through our subconscious feel for where the rest rooms and the circulation desk are; we will have screens whose resolution will be the equal of Library of Congress printed cards, or even of handwritten cards; we will be able to look at five or six records at a time, and to move forward and backward through 826-unit clusters of retrieved records with the rifflingly variable speeds we now attain over semi-pliant paper. All this and more will be ours in the years to come, assuming there is money to pay for it. But it isn’t ours yet, and Michael Gorman and his colleagues in Closing the Catalog were foretelling the ways that online techniques would free information retrieval from various alphanumeric tyrannies almost fifteen years ago. Today, if I take a stool in front of a University of California MELVYL screen and type, for instance, BROWSE SU CENSORSHIP (meaning “Please show me the subject headings relating to Censorship”), this is what I get back:

  LONG SEARCH: Your search consists of one or more very common words, which will retrieve over 800 headings and take a long time to complete. Long searches slow the system down for everyone on the catalog and often do not produce useful results. Please type HELP or see a reference librarian for suggestions.

  If I type in BRO SU ROME—HISTORY, I get the same thing. So, too, if I’m curious about AIR POLLUTION or BIRTH CONTROL or PHILOSOPHY OF HISTORY or HISTORY OF PHILOSOPHY, or BIBLE—HISTORY, or even INFORMATION STORAGE AND RETRIEVAL. If I need some books on LINGUISTICS or FOLKLORE and ask MELVYL what it has, and I do so during normal working hours, I get a slightly different message:

  PEAK LOAD RESTRICTION: Your search consists of a common word which would retrieve over 6,400 headings and would slow down the system. During peak load periods, your search cannot be completed. You may reissue your search to make it more specific, or try again during the evening or early morning.

  And if I try to narrow the search by typing, say, FI XS ROME—HISTORY (meaning “Please show me only those subject headings that begin with the words ‘Rome—history’ ”), I will be sure to miss many excellent books, including Robert Brentano’s Rome Before Avignon, which is cataloged under “Rome (Italy)—History—476–1420.” Nor will title-keyword searching solve my Roman history problems with MELVYL: to get one possible sample of things I will miss if I place too much confidence in keywords and exact subject headings, I can type FI SU (ROME HISTORY) AND NOT XS (ROME HISTORY) AND NOT XS (ROME ITALY HISTORY) AND NOT TW (ROME) AND NOT TW (ROMAN) AND LANG ENGLISH, which translated means, “Please show me all the books with subject headings that contain ‘Rome’ and ‘history,’ but whose subject headings do not begin either with ‘Rome—history’ or ‘Rome (Italy)—history’ and that do not contain either the word ‘Rome’ or ‘Roman’ anywhere in the title, and that are in English.” Even after this narrowing-down (which takes 97 search cycles and should only be performed after midnight, when few are roaming the system), I will see over 130 books, among them Peter Brown’s Power and Persuasion in Late Antiquity and Gillian Clark’s Women in the Ancient World—books I would have missed had I been too uncritical a keyword devotee. And MELVYL is, in fairness, a much more flexible and powerful system than many, especially in its ability to marshal long chains of Boolean exclusivity: if you do an analogous FI KSH search (KSH refers to Keywords of the Subject Heading) on Harvard’s catalog (which cheats by defaulting to a limited, initial-word subject search very similar to MELVYL’S XS, and which disallows commands longer than one line), HOLLIS too will choke on topics like ROME—HISTORY, or ETHICS—HISTORY, or AGRICULTURE—HISTORY, or COSMOLOGY, or BOOKS AND READING, or COINS, or TEXTILE (but not TEXTILES), or TOBACCO, or LIBRARIES, or our friend INFORMATION STORAGE AND RETRIEVAL. Each time, HOLLIS comes back with its HELP OVERFLOW screen, saying in boldface: “Your search retrieved more than the maximum number of items the system can display.”

  None of those general headings deserve a system rebuff. All of them are eminently reasonable ways to begin a search—a search that we who have research papers, or dissertations, or mere essayistic tirades due might like to begin in the next few days, not two or five years from now, when the online catalog has been improved to the point where it can comfortably accommodate this sort of inquiry. If there is already a big wooden machine in the library that is able to point us quickly in a few directions without calling us users of “common” words and thereby hurting our feelings, maybe we should keep it. The fact that the card catalog is no longer the necessary first stop in a visit to the library shouldn’t doom it. If nobody uses it for months at a time, stuff it away in the rare-books wing, or sequester it on a seldom visited subbasement floor, crowded humbly among the Z shelves (Z is the call letter for bibliography and library science), like Mike Mulligan’s obsolete steam shovel. Or it could be pushed off to the far end of the reference room (one row of cabinets bolted on top of another to halve its footprint, the entire structure festooned with warnings that its information is not current), where it will gather exactly the same amount of dust as other big and usefully out-of-date library catalogs: Cutter’s, or the British Museum’s General Catalog of Printed Books, or the magnificent pea-green-and-gold wall of the 756-volume National Union Catalog, Pre-1956 Imprints.

  There are still some librarians who do what they can to save the card catalogs under their care. The librarian of the small, separate library-school library at Berkeley, Patricia Vander-berg, stores her two card cabinets in the stacks, out of range of the administrative eye, but she has kept them. They were edited by a now retired librarian named Virginia Pratt. Ms. Pratt “did some special things to that catalog,” Patricia Vanderberg told me. The university, however, which is in a self-mutilating mood these days, plans to merge the library-school library with the main library, and before the merger there will be a severe culling; the card catalog will almost certainly be thrown out. But for the time being, if some afternoon, depressed by thoughts of cardnage now in progress, you were to travel to the stacks of the library-school library, you could restore your good spirits by contemplating an unharmed, unshrunk, fully operational (though frozen) subject catalog.

  Fifty drawers compose it. They are made of blond wood. Pull out one of the “C” drawers, the one labeled “Catalogs.” Here are several guide cards: one for “Catalogs,” one for “Censorship,” one for “Children’s Literature.” Before you have read or touched a single card, you have learned something important about the library-school library, and even possibly about librarianship in general: the edges of the cards following the guide cards for “Censorship” and “Children’s Literature” are dark with handling, whereas the ones following “Catalogs” are not. Thus “Censorship” and “Children’s Literature” have over the years been of more interest to library-school students than “Catalogs” has. This is a surprise—at least, it was a surprise to me. Maybe it’s one reason we’re in this pickle.

  And abruptly you realize, looking at these expressive dirt bands, that even the libraries, like Harvard and the New York Public Library and Cornell, who microfilmed or digitized some of their cards prior to destroying them, have—by failing to capture any information at all about the relative reflectivity of the edge of each card—lost something of real interest, something eminently studiable. Who knows what a diligent researcher who photographed (from above, on a tripod) each close-packed drawer of Harvard’s Widener catalog with a high-contrast camera might find out, were he to correlate his spectrographic dirt-band records with the authors that, as distinct clumps, exhibited some darkening? Of course the “Kinsey” cards would be thoroughly dirt-banded—but which others? This is, or was, a cumulative set of scholarly Nielsen ratings for topics at twentieth-century Harvard that is perhaps mor
e representative than any other means of surveying we have. Instead of tossing its catalog out, Harvard ought to have persuaded a rich alumnus to endow a chair for dirt-band studies.

  If, though (back at Berkeley’s library-school library), you now take a glance at those two other major subjects in the still-extant subject catalog, the ones more popular than “Catalogs,” and begin by flipping the “Censorship” guide card toward you, you will note that Virginia Pratt has prepared some helpful “See also” material, typed with a red typewriter ribbon on several different models of typewriter:

  Censorship

  SEE ALSO

  Libraries—Censorship

  Liberty of the press

  Expurgated books

  Prohibited books

  Condemned books

  Books and reading for youth—Censorship

  Government information

  Children—Books and Reading—Censorship

  Book burning

  Pornography

  A second card goes on, still in red:

  Censorship (continued)

  SEE ALSO

  Freedom of information

  Audio-visual materials—Censorship

  Libraries—[Place name]—Censorship

  Young adults—Books and reading—Censorship

  Textbooks—Censorship

  School libraries—Censorship

  And then in black there follows:

  For additional material on this subject see:

  Vertical file: Intellectual freedom