Must We All Be Rhetorical Machines - Misti Yang
Rhetorical Machines: Writing, Code, and Computational Ethics, edited by John Jones and Lavinia Hirsu, Tuscaloosa, University of Alabama Press, 2019, 279 pp. $69.95 (trade cloth); $34.95 (quality paper or e-book)
A conversation with the chatbot Eliza:
Eliza: Hello, I am Eliza.
Misti: Hello, I would like to ask you about a book I read.
Eliza: We were discussing you, not me.
Misti: Well, I just finished a book called Rhetorical Machines.
Eliza: I see.
Misti: Do you think you are a machine?
Eliza: Perhaps you would like to be a machine.[i]
Each section in John Jones and Lavinia Hirsu’s edited collection begins with a dialogue between themselves and a chatbot, similar to my own chat above with an Eliza emulator. And, although my conversation with Eliza was brief, it developed into a poignant final thought for beginning this review: “Perhaps you would like to be a machine.” This codic pronouncement left me pondering why would one want to be a machine. Perhaps a flight to algorithmic certainty is driven by anxieties about ethics, a theme of this collection. If machines can answer our difficult questions, then we are somehow less responsible for outcomes. And, thus, Eliza’s indictment also left me wondering to what extent we want rhetoric to be a machine.
Jones and Hirsu’s collection contributes to conversations about the role of rhetoric in computation. Given the pervasiveness of computational influence, little needs to be said to justify the importance of thinking critically about rhetoric’s role in our coded past, present and future. The essays mirror the expanse of the influence of computational culture by covering a broad range of topics and times ranging from a historical-critical analysis of the argumentative strategies of Charles Babbage, who built mechanical computers and developed theories of digital computation in the mid 19th century, to an examination of the mathematical underpinnings of computational code. Jones and Hirsu frame these diverse approaches to computational rhetoric with the “theoretical lens” of the “machine” (3), a lens inspired by previous work from Jenny Edbauer Rice and James J. Brown Jr. While the machinic metaphor could imply an instrumental and/or categorical approach to rhetoric, Jones and Hirsu insist that “the rhetorical machine is meant to call forth a wide range of rhetoric(s) through which human-machinic interactions become possible, such as posthumanist, procedural, material, and bodily rhetorics” (4). While the scope is ambitious, Jones and Hirsu suggest that the essays are anchored by two central goals— first, to “examine the influence of rhetorical processes” on computation broadly defined, and second, to “uncover multiple theoretical, methodological, and practical entry points into questions of computational communication” (5).
The chapters are organized in four thematic sections: “Emergent Machines,” “Operational Codes,” “Ethical Decisions and Protocols,” and “Responses.” In the “Emergent Machines” section, authors address “how technologies, software and algorithms are shaped in conjunction with rhetorical processes” (5). Jonathan Buehl provides a historical contextualization of the argumentative strategies that Charles Babbage used to garner cultural and capital support for the development of his “calculating engines”(16). The engines were giant mechanical calculators that used metal gears instead of silicon chips to solve complex mathematical problems. While the analysis is interesting, Buehl’s final section provides a provocative inversion by examining how the “engines” shaped Babbage’s rhetorical strategies. For example, in the Ninth Bridgewater Treatise: A Fragment Babbage relied on his engines to “reconcile the laws of science with the theological concept of what we would now call intelligent design” (41). J.W. Hammond draws from debates over automated essay scoring from the 1950s and 60s to demonstrate how definitions of writing, assessment, and computation influenced the development of grading technologies. Hammond works to make the larger point that all machines are “rhetorical machines of and by definition” (64). Kevin Brock suggests that rhetoricians often “recognize code primarily for its expression of some ‘other’ argument made through code as a transparent vessel,” for example, how code results in a video game that is then the focus of critique (71). Brock advocates for another way of thinking about the rhetoricity of code by looking closely at online discussions about best practices for Ruby on Rails, “a popular framework for web applications” (69-70).
The essays in “Operational Codes” turn to “computational processes that are being used to achieve rhetorical ends” (6). In a surprisingly illuminating case study of the Arizona Department of Transportation’s attempt to develop a computer program in the 1980s to improve road maintenance, Jennifer Juskiewicz and Joseph Warfel advance two important perspectives on the role of rhetoricians and rhetoric in computational contemplation. First, they suggest that we need to expand our understanding of computation beyond code to account for mathematics because “much of the information and many of the decisions that undergird programming are made at the foundational mathematical level” (94). The authors also emphasize that to accomplish this requires collaboration with “computational specialists” (106). Ryan M. Omizo Ian Clark, Minh-Tam Nguyen, and Willam Hart-Davidson delve into hands-on programming. Their chapter chronicles how they developed the “Facioloscope— a web-based, machine-learning application that performs rhetorical analysis and visualizes findings based on natural language submissions on the fly” (110). The Facioloscope’s ultimate goal is to aid in troll detection and intervention in online comments sections. Joshua Daniel-Wariya and James Chase Sanchez examine the assumptions about race that are embedded in video game cutscenes, the moments in video games that are like short movies in-between gameplay. They develop their argument through the Thomas Rickert’s concept of ambient rhetoric and Alexander Galloway’s work on pure process, moving beyond a simple close-reading of the scenes to examine how game engines, software-development toolkits that simplify many aspects of video game development, “enclose racial ‘safeties’ and stereotypes into their material production” (138).
The third section, “Ethical Decisions and Protocols,” explores “the ethical implications of designing software programs” (7). Anthony Stagliano’s chapter exemplifies the potential of rhetoric to invigorate grappling with algorithmic encounters. Stagliano accomplishes this by expanding “metis, or cunning intelligence, to examine ways in which it is not always (just) a human capacity but is also active in non-human ‘bodies,’ in this case, open-source ‘code libraries’” (170). He accomplishes this through an analysis of CV Dazzle, an “activist media art project aimed at thwarting facial recognition algorithms” (170) and its engagement with the face recognition algorithms that are available in an open-source code library. Conceptualizing the code library as a “wily audience” is central to Stagliano’s argument (181) that the computational landscape is rife with things that are “simultaneously available means of persuasion and persuadable audiences caught up in a complex web of response” (187). As Stagliano suggests, such a perspective opens up possibilities for tactical interventions into algorithmic landscapes. Jennifer Helene Maher, Helen J. Burgess, and Tim Menzies offer a helpful overview of how data scientists think about big data as being opposed to and superior to human judgment and unpack how this assumption is undermined by a theoretical approach they term “rhetorical stacking,” intentionally building on the idea of the technological stack. They outline six layers of what they term the rhetorical-technical stack in their analysis of a big data study on gender bias on GitHub. In the final chapter, Elizabeth Losh revisits Hillary Clinton’s private email servers and outlines how it was perhaps “the conflation of gender and technology at work in the popular imagination,” not one or the other, “that was to blame for Clinton’s stunning defeat” (213). Central to her claim are the ideas of digital purity, which Clinton violated, and digital exclusivity, which Trump embodies.
Because the book started with conversations had during a workshop led by James J. Brown Jr and Annette Vee, the collection closes with responses by each scholar. Brown provides a deft answer to the question, “What now?” His answer also offers a reply to my opening rumination on rhetoric as a machine. He suggests that approaches that “focus on the crafting and analysis of procedures,” what could be considered a machinic approach, may be “hitting a limit” (230), and in response, he calls for more focus on machine learning, as well as “content and data” (232-33). He admits that his own work is “susceptible” to his critique of an overly procedural focus, and I would suggest that so are several of the book’s chapters. The framework of the rhetorical machine can overwhelm the strength of rhetoric to address uncertainty and messiness, in other words, to be un-procedural.
At the intersection of Brown’s call for work looking at machine-learning, content, and data is his question: “What rhetorical pedagogies might be useful for educating machine-learning systems?” (234). But, this begs the question: What should machines learn? This is a question of ethics, which is a theme of the book that is not fully addressed. The chatbot conversations within the collection are untapped as fruitful content, but one provides some perspective on why these questions matter. In the dialogue that opens the third section, a chatbot named Elbot announces, “My self appointed mission in life is to assist humankind in its confused meandering through life. And if I’m lucky I may even succeed in protecting it from its own stupidity, although that is truly a Herculean task, even for a robot” (167). Here is a “machine” that embodies assumptions about uncertainty, the human condition, and the role of computation. How might the assumptions of rhetoric address these machinic assumptions and render ethical insights? Joseph Weizenbaum, the author of my chatbot partner Eliza, thought that answering the “should” question was critical before developing artificial intelligence and natural language processing further. However, neither his chatbot or his insights are addressed in this collection.
Returning to the stated goals of the editors— to demonstrate the rhetorical influences on and display the breadth of rhetorical approaches to computation— the collection succeeds, making a strong case for what rhetoric contributes to the digital humanities. Given the breadth of scope, the book is well-suited for a course that surveys rhetorical methodologies and theories for thinking about computation and/or the digital. A few chapters may be too theoretically dense for undergraduates, but the central arguments should still be accessible for upper-level undergraduate courses.
To close, a few final thoughts from Eliza:
Misti: Rhetoric confuses me.
Eliza: Come, come, elucidate your thoughts.
Misti: What is rhetoric?
Eliza: Are such questions on your mind often?
As Rhetorical Machines proves, such questions should be.