Barnwoggler Explored

With the explosion of interest in generative artificial intelligence this year, since the public launch of ChatGPT in late 2022, it seems like a good time to revisit Robert C. Binkley’s story “Barnwoggler’s Invention”, and take note that the revolution we are living through was foreseen, dimly, by a grad student employed in the Hoover War Library in 1926. In the story a librarian invents a machine that absorbs the text of library materials and generates publishable research papers. The story, brief sketch though it is, is interested in two questions: 1) what might a machine like this be like? and 2) is it ethical to publish the machine’s output as one’s own work? The answer to the second is no: “A man’s name on an article was supposed to stand as a guarantee that the writing was the work of his own hand.” Barnwoggler’s downfall therefore follows, even though “it seemed impossible, for technical reasons, to convict him of plagiarism”. This post will look at the first question: what technical novelties prompted Binkley to imagine the invention? And how does the story relate to the technical innovations of Binkley’s own career?

To celebrate the coincidence of AI I had ChatGPT generate the four articles that Barnwoggler produced with his invention, complete with hallucinated bibliographic references. I hope Barnwoggler’s invention did a better job.

The question I was left with in the first Barnwoggler post (before I found the text of the story) was: “In how much detail did [Binkley] imagine this machine?” The answer is in the paragraph that describes the development of the equipment.

Barnwoggler’s invention developed by successive improvement from a comparatively simple mechanism which he secretly devised in order to save time in the cataloguing of books. From an electric cataloguing machine he developed a book-detector, from a book-detector an automatic assimilator. These were the intermediate stages. In its final form the invention combined the principles of electric typewriter and player piano, radio and photostat, findex card system and adding machine. Barnwoggler could then take the book-detector, run it through the stacks with its delicate gauges set on a certain problem, arrange the selected books in the assimilator, turn on the current, and out would come page after page of manuscript ready for publication.

In its final form, then, the invention consisted of two devices: a portable book-detector that could be wheeled through the stacks to gather books that were relevant to a selected research topic, and an assimilator that would somehow read in the content of the books and print out a finished academic article.

Input

The input mechanisms are deliberately vague. Radio and photostat are the models, providing respectively the ability to operate remotely (as the book-detector does, controlled by “sensitive gauges”), and a form of input via photography.

Output

In mentioning the electric typewriter Binkley probably had in mind the Hooven Automatic Typewriter (see the advertisement above). This device used a player piano roll to encode text, and drove a conventional manual typewriter to do the actual typing. The coding on the roll could mark places to stop and let the operator type custom text such as an address. The result was indistinguishable from a letter produced by a skilled typist. This uniformity was valued by advertisers as a way to get past the recipients’ mental spam filters, which were already tuned to reject obvious form letters. Binkley refers to this kind of service in New Tools, New Recruits for the Republic of Letters: “One of the most curious developments in the graphic arts today is the effort of the advertisers to make printed matter look like typescript, while the creators of books that are not in sufficient demand to warrant publication are demanding a typescript that will look like print.”

Processing

The Findex card system and the adding machine, and indeed the sense of text as something that can be broken down and reassembled to produce new meaning, represent computation. The Findex is particularly interesting since it connects with the directions of the documentalists, which Binkley did not pursue in his published writings.

Robinson Findex cards used a system of aligned holes and slots to encode data and enable basic boolean queries. You started a query by inserting a long needle through a selected hole in a pack of cards in a special drawer, and then turning the drawer over. Those cards which had a slot cut between the selected hole and the next hole would be allowed to drop the length of the slot, and could then be caught on a second needle before turning the drawer back to its original position. A boolean AND could be done by entering two needles before turning the drawer, so that only cards with a slot for both holes would fall; an OR could be managed by using a needle to transfer all the fallen cards for two searches into an empty drawer; a NOT could then be done by searching the second drawer and transferring selected cards back. Findex cards found their principal application in personnel management, and Stanford used this system to place students in part-time jobs as early as 1921-22: “… the training, experience, and general qualifications of the students were taken into account in recommending them for work. The use of the Robinson Findex record cards makes this grade of service possible with a minimum amount of effort and time.”

Context: Librarians and Scholars

There is also a foreshadowing of Binkley’s ideas on the relationship between the roles of researcher and librarian. When he moved to Western Reserve University he asked the college librarian “how many cards or how many drawers of cards” were in the 900 section, and of those how many were devoted to European history.¹ He chaired the Library Committee at Flora Stone Mather College for most of his time there (1930-40), and was closely involved in service and collection policy both in the College and in the university libraries. When he first arrived in Cleveland, during a brief period before the Depression devastated its budget, he was involved in Cleveland Public Library’s collection policy as well, coordinating its collecting of European government papers with that of the Western Reserve University libraries. The success of the Cleveland Union Catalogue project, implemented as a WPA white-collar project, depended on his coordination of the developments in microfilm technology with both the various libraries of the Cleveland area and the WPA hierarchy.

The librarian identity he developed at the Hoover Library was therefore something he carried with him into his academic career. In a letter of recommendation for one of his students for a fellowship at the Newberry, he wrote:

… I observe that a man is wanted whose training is in research rather than library technique. As one who has worked in both kinds of position – for I was reference librarian at the Hoover War library before I became a university teacher – I have the highest respect for this policy decision that you have made. I am anxious to see appointments made under this policy not only in your library, but elsewhere in the country, as successful as possible.²

This seems to contradict Barnwoggler’s explanation for his varied publications – “I have always said that library training is the best possible preparation for research work of any kind.” Binkley did not in fact have formal library training, but on-the-job experience in the newly founded Hoover War Library within the Stanford library system must have been an excellent apprenticeship.

Context: Quantifiability of Research

Barnwoggler’s invention was founded on his sense of the quantity of source material needed by his professors to produce an article (one booktruck) or a book (five trucks). In this he anticipates two threads in Binkley’s thinking about the nature of historical scholarship: 1) scholarly activity is quantifiable, and 2) historical research proceeds by working “unworked” materials and is therefore determined by the availability of fresh sources. We see them in an evaluation of the research possibilities at the Hoover War Library he wrote from New York for Ralph Lutz in 1928, on the research problems that could be addressed using the collections of the Hoover Library:

There seem to be two types of research problem, which might be called the topic and the research project. The topic is usually suggested by the concurrence of some unworked source material (printed or not) with the general reference background of the Library. Every important acquisition or publication of sources provides a number of such topics. The Herron and Tisza papers for us, like the House papers for Yale, suggest out of hand a number of such topics. The research project corresponds in a more general way to the equipment and acquisition policy of the Library, and includes within itself a number of topics logically related to each other. To this class belong Fisher’s field of war relief and food control, and Golder’s Russian field.³

Late in his own career Binkley developed these ideas in an attempt to encompass whole research careers in a single analysis. On 2 Oct. 1939, he wrote a letter to Robert Crane of the SSRC, running up a line of thought which he hoped to discuss with Crane when they met in New York a couple of weeks later. The letter was circulated as an appendix at the meeting of the JCMR in December 1939.

In it he proposes a method of making scholarly activity measurable. Where this is usually only done at the project level in relation to grant funding, he proposes to consider it at a macro level: the “life times and part life times of scholars and clerks”. His inclusion of “clerks” harks back to “The Cultural Program of the WPA” (published earlier in 1939), with its invocation of the clerici of the Carolingian court, the educated white-collar workers who formed the paradigm for the amateur scholars Binkley hoped to foster in his own teaching and in the WPA projects he sponsored. At the time this letter was written Binkley had applied for a grant to study the possibilities of academic support for amateur scholars. This led to the formation of the Committee on Private Research at Western Reserve University, whose 1942 report was published by William Dix.

Binkley proposal to Crane was an extension of the JCMR’s early work on categories of research materials. The question about a project should be the extent and quality of the bibliographic control of the source materials: the better the bibliography, the more time the scholar will spend on interpretive research instead of discovery of materials.

Now I am going to propose that we take the one unit which limits us, whatever else may be without limit, – namely, the scholar’s lifetime – and ask ourselves how the scholar’s time should be distributed through the whole hierarchy of controls over original data. We may discover that in some cases the system of controls is so weak and inadequate that the scholar’s operations have an almost random character; in other cases the system of controls is so clear and so ample that the scholar’s time is very systematically divided between the consultation of controls and the consultation of the underlying data which is controlled.

The better the bibliographic control, the more time in the scholar’s lifespan will be devoted to systematic interpretation.

From this point of view, the contribution of Barnwoggler’s invention to the scholar’s productivity is more in the book-detector than the assimilator. It builds on Barnwoggler’s understanding of the necessary quantities of source materials and provides a perfect control of the relevant bibliographic resources, at least within the scholar’s local research collection. The WPA projects that Binkley sponsored, including the local archives lists of the Historical Records Survey, the great newspaper digests and the Cleveland Union Catalog, all aimed to use the work of the clerks to enable the research of the scholars. Binkley’s own research career took off with his publications on the availability of sources for the study of the Paris Peace Conference; he saw his bibliographic articles in Political Science Quarterly⁴ as a “starter’s gun” for research on the conference by others.⁵

It is ironic that the line of thought in his letter to Crane should develop when Binkley’s own lifetime was so limited: he would die six months later.

---

1. Doc. 4318: RCB to Elizabeth Richards, 1930-10-03.↩︎

2. Doc. 5878: RCB to John Windle at the Newberry Library, 1937-09-20. The student was John Kolehmainen, who went on to a productive career in Finnish-American history; it appears he did not get the Newberry position.↩︎

3. Doc. 1602: RCB to Ralph Lutz, 1928-04-24. The copy of this memorandum in Binkley’s papers is marked “not sent”. In some form, though, it must have reached Lutz, who wrote to Binkley a few weeks later: “I was mighty glad to receive your formidable letter on research problems in the Hoover war Library. It is splendid and arrived in time for consideration at the meeting.” Doc. 1588: Ralph H. Lutz to RCB, 1928-05-22.↩︎

4. Robert C. Binkley, “New Light on the Paris Peace Conference I: From the Armistice to the Organization of the Peace Conference,” Political Science Quarterly 46 (September 1931): 335–61, and “New Light on the Paris Peace Conference II: The Organization of the Conference,” ibid. 46 (December 1931): 509–47.↩︎

5. Peter Binkley, “‘If You Could Send over Your Documents to the Photostat Department…’: Paris Peace Conference Documentation and the Advent of Microfilm” (Open Ideas 2021, University of Alberta Library, 2021), p.5.↩︎