Minimal Computing and Liberal Arts Education

1. Introduction

This paper builds on the 2022 special issue of DHQ edited by Roopika Risam and Alex Gil on “Minimal Computing.” (LINK) I will begin with a review of some matters of definition and salient questions raised in that issue. Then attention will turn to a set of questions mostly neglected in the essays. What is the role of minimal computing in liberal arts education? In a dedicated learning environment, the problem of the “learning curve,” often labeled too steep for minimal computing tools, shifts. The question is not what is hard or easy to learn, but what is most worth the effort of learning. I will argue that fundamental tools and concepts will serve students better than commercial software and bespoke platforms, which may not be affordable, in existence, or applicable to different problems after graduation. Further, of all the things to learn, what most fits the larger goals of an integral liberal arts education? I will argue that symbolic computer literacy and fundamental concepts of computational methodologies are more consistent with the tradition of liberal arts education which has always viewed knowledge of languages, math, and science as integral to humanities questions. I will conclude with a brief case study of a project that required students to apply part of the TEI Guidelines in a simple text editor and collaborate in GitHub, as well as plans for a course in the liberal arts core curriculum.

2. DHQ 2022

The 2022 special issue of Digital Humanities Quarterly titled simply and appropriately, “Minimal Computing” successfully gathered a diverse range of perspectives and case studies. I commend the resistance to a rigid definition in favor of questions to ask. I would go even further and say the word “minimal” is simply an adjective by which two possibilities can be compared. This comparison does not presume that minimal is always better or that nothing could be too minimal. As the editors emphasize, decisions vary entirely with the project and team. Even without a rigid definition, minimal computing coheres in as much as several criteria for comparison often converge. I will elaborate on a set of criteria presently. I also commend the special issue for its resistance to a single archetypical example, even though the contrast between Markdown and Jekyll versus WordPress appeared multiple times. That example is funny to me because I would call Jekyll less minimal than my preferred approach using Apache Web Server and static HTML hand-coded in the plain text editor VIM. I will attempt to suggest an array of examples in which one way of computing may be more minimal than another, and a third way more minimal still.

The criteria around which the comparative adjective “minimal” converge include elegant, sustainable, accessible, fundamental, and cost effective. I was surprised to read the editors assert that the virtue of “elegance” is specific to the history of UNIX development. If it is not more widely valued, I think it should be. Achieving a task simply and efficiently seems inherently good to me, and perhaps especially in my capacity as a teacher. I often try to guide my students past a feeling of being overwhelmed to a feeling that all the pieces fit together in an elegant way. Sustainable is not always easy to predict, but I would suggest that something like TEI that has endured for decades is more sustainable than the latest beta vaporware. Technologies come and go, but a community invested in using and supporting a technology is often a good sign. Sustainabilty gets at the question of what skills will continue to benefit students long after graduation. Similarly, I would correlate accessible with open-source and free, even while acknowledging that the related descriptor “human readable” depends on the human. That point will be important for my central section on what we should be teaching our students. If a technology or method looks good on all points except human readability, maybe we should focus on learning and teaching how to read. I include “fundamental” as another important criterion because the more minimal of two choices often relies on components that also serve other purposes. A hammer can be used in ways other than driving nails and a screwdriver can do things besides drive screws. Those tools are more fundamental than a relatively specific tool like a planer. A hammer has its own learning curve, but once learned it is generalizable and reusable. A tool that students can use on completely different projects is more fundamental than a bespoke platform. Finally, I mention cost-effective last because I wish to correct for an excess of emphasis on this point. I have been told in so many words, “I thought minimal computing was just DH for poor people.” I don’t feel a need to foreground the negatives of scarcity, constraint, and lack of resources. Although most of my students are Pell-grant eligible, they have plenty of computing power in their pockets. I honestly do not know if software is as easy to steal today as it was when I was in college. I believe the point stands that free software is more minimal than Photoshop or anything made by Adobe.

I believe these criteria often converge on a distinction by which one tool can be called more minimal than another. I can offer several examples that illustrate the comparison. Starting with web sites, WordPress is more minimal than Flash. Jekyll is more minimal than WordPress. Direct-authored static HTML is more minimal than Jekyll, especially if authored with a text editor. Hand-edited JavaScript is more minimal than jQuery. Hand-edited CSS, when sufficient, is more minimal than JavaScript. Semantic HTML, when sufficient, is more minimal than CSS. Leaflet is more minimal than ArcGIS. R and Python are more minimal than SPSS. A command line interface is more minimal than a graphical user interface. Remember I am not asserting that minimal is always better for everyone.

One additional matter of definition that I wish to adopt from the special issue is the distinction between metaphor and symbol. In the context at hand, Risam and Gil use those terms to contrast a graphical user interface with code. (LINK) Coding can be called more symbolic and more minimal than learning to use the graphical user interface of a bespoke platform. Miriam Posner appropriately acknowledges the problems with valorizing or requiring the ability to code as the mark of a real digital humanist. (LINK) By no means would I support devaluing the work of a digital humanist who relies on collaboration rather than direct coding of machine-actionable languages. The point is important not only because of the gender correlation of coding ability in generations such as mine. I think it is safe to say that the special issue addresses an audience of DH professionals and project managers rather than teachers. If I am leading a team of peers I need to be careful not to valorize disproportionately an attribute that correlates with gender. As a teacher, however, I would never say that women should not be taught science just because of an imbalance in previous generations. Shifting the focus from project design to teaching turns several questions of minimal computing upside-down.

3. What is the Role of Minimal Computing in Liberal Arts Education?

My central concern in this paper is to consider a simple question. What is the role of minimal computing in liberal arts education? That is, what should we be teaching our students? Reframing the question in this way causes us to re-evaluate the advantages and disadvantages of minimal tools. Chief among these is the concern of the “learning curve” associated with minimal tools. The majority of the essays in the special issue used the phrase “learning curve” specifically, always with a negative connotation. Of course in liberal arts education, learning has a positive connotation. The concern of the steepness of the curve remains valid, as with any topic in the liberal arts. Pacing is a different question than difficulty. We want our students to learn new and difficult things. The question shifts to which of the hard things to learn are worth the effort. I suspect most of us would agree that learning at least a second language is essential to a liberal arts education. Language acquisition has fundamental benefits to forming minds that cannot be replaced by Google translate. If you’re like my students, you might be slower to recognize Calculus or Advanced Algebra as essential for the education of a humanist. If you are able to appreciate the value of math for the liberal arts, it should be a small step to appreciate symbolic thinking as valuable when teaching Digital Humanities.

In my experience Digital Humanities can be seen by academic colleagues as leaning toward the trendy and fleeting. In these situations I believe it can be helpful to talk about education in terms of liberal arts and interdisciplinarity. The liberal arts have always been premised on the value of bringing together in one person the ability to think in different ways. Even as we begin to specialize, the liberal arts assist our ability to collaborate with others and their complementary specializations. Interdisciplinarity also has a consistently positive connotation in conversations about curriculum. I suggest that the intersection of the interdisciplinary liberal arts and digital humanities, significant by any account, is most profound in the area of minimal computing.

I discussed several criteria as useful to differentiate one computing tool as more minimal than another. I believe the same criteria can be useful in making decisions about teaching and curriculum. How do we want our students to be different having taken our class? How will they be different after graduation? Ten years after graduation? With those questions in mind, I would suggest that symbolic thinking is an end in itself. I would not call the Text Encoding Initiative an end in itself, but I do think it can aid critical reading. Can thinking about the divisions and milestones of text help us think about literature? Can coding our conventions to be machine-actionable help us be aware of our assumptions? Can a markup language that throws an error over a missing comma help students appreciate precision and attention to detail in human-to-human written communication? I propose the answer is yes.

Thinking about digital humanities education in terms of liberal arts education can also help us with the difficult question of balancing liberal arts and professional education. The mission statement of my university mentions “integrated liberal arts and professional education.” Thus, I can sympathize with my colleagues who insist that employers want graduates to know how to use commercial tools such as Photoshop and SPSS. Yet, I think we can and should communicate to employers that the ability to do the same thing with Gimp or R will make for a quick transition to commercial products. Furthermore, the skills developed to use R will come in handy the next time a new problem emerges that cannot be readily addressed with specialized software. Although I wish to de-center the poverty argument for minimal computing, I do think there are deep problems with stacking “lab fees” for temporary access to commercial software subscriptions that disappear when the course or degree program ends. If we teach our students to use commercial products, we create consumers for life. If we teach our students to use fundamental tools, we create builders for life.

4. Case Studies

To add some specificity to the general claims above I would like to describe two case studies that illustrate some success in as much as they emphasize the fundamental tools of digital humanities. The first has been implemented and the second is still being planned. The first fits the format of an elective course or paid student research, while the second aims to fulfill a core curriculum requirement.

The first case study was technically paid student research, but the amount of time and size of the student cohort make it very comparable to the experience of my colleagues who teach digital humanities electives. The major activity was to consult digital images of a fifth-century manuscript and an 1861 print edition. Students encoded the “text” using the Text Encoding Initiative. By “text” I do not mean just strings of characters. The manuscript and the print edition differed in conventions such as space between words, punctuation, and capitalization. The print editor often normalized orthography and silently corrected deviations from standard grammar. The print edition had its own limited conventions for indicating gaps and uncertainty, to which the students added their own assessments. The manuscript lent itself to division of folio, column, and line, but made no written claims about literary structure. Chapter and verse numbers were added even after the 1861 print edition. Students thought about all the information that one might wish to encode when encoding a text. They thought about how the TEI guidelines attempt to address that information. They entered those codes with all the angle brackets into a validator that would turn red and throw an error message over any violation of TEI or XML. The error messages were not so helpful that it did not take a troubleshooting tree to isolate a misplaced forward slash. The students shared their work through GitHub. They were able to consider multiple ways of doing things within the rules of XML and TEI.

My evaluation of the overall success of the learning experience is positive. Students certainly experienced frustration when a validator told them something was wrong with their work. This frustration, overall at least, seems to me balanced by the experience of success when validation was successful. Students gained great confidence in their ability to code. I don’t have reason to believe that any of them went on to write XML or HTML with any regularity. I do think they understand that web pages are communicated in strings of characters that represent meaning and visualizations other than words. I do think they understand that when they are reading a text there are features and conventions of which they were not previously conscious.

The second case study is still in development and has not yet been implemented. Thus, I am especially interested in your suggestions here. The challenge is to design a course that lives up to my goals about the liberal arts while fitting within the divisions of my university’s core curriculum. In my case, the course must be rooted in the academic study of Theology sufficiently to meet my university’s expectations to teach students to think about timeless questions.

The course I am designing is tentatively titled, “Godish: The Languages of Machines, Humans, and God in Translation.” The course would include enough traditional human linguistics to help students understand that communication is hard, but possible. The course would include enough fundamentals of machine-actionable language to convey that humans and computers are good at different things. Communication across human and machine realms is also hard, but possible. The anchor in traditional theology is Theology of Revelation. Christians, along with Jews and Muslims, believe in a God who is fundamentally transcendent, radically super human, and yet reveals Godself to humans. Despite some differences in emphasis, we share a belief that God’s self-revelation is communicated through human prophets in human language and human scribal cultures. Students should internalize the broader idea that divine revelation is powerful but humbling as it requires translation across language, culture, and time in addition to the transition from divine to human platforms. I welcome all ideas and references that could support these learning objectives.

Thank you.