9 Computational Thinking and Software Studies

The Content of Computing

Marvin Minsky opened his Turing Award Lecture in 1970 with the comment, “The trouble with computer science today is an obsessive concern with form instead of content” (197). Fifty plus years later, we can make the case that Minsky’s concern is our concern, especially in the subfield of computer science education. Minsky lamented the field’s focus on formalisms, arguing that not enough was known about the basic phenomena of computing to generate universal theories. Instead, he urged a closer focus on the “content” of computing; what computing could do. Minsky uses programming languages as an example, arguing that we should spend more effort on describing how actual programs work than describing the abstract syntax of the language and how it can be used to write any program. In this memorandum I argue that computer scientists and, especially, computer science educators should focus more on content and less on formalism. To make this argument, I look at the two leading theories in computer science education, computational thinking and constructionism; then, I suggest that we should incorporate features from the field of software studies into our understanding of computer science education.

Computational Thinking Takes Hold

Our current charge to teach “computational thinking” (CT) has been the rallying cry to include “coding,” computer programming, and computer science more generally in every level of education, with a special focus on introducing CT as early as possible in K–12 education. This renaissance of computer science education in the guise of CT stems from Jeanette Wing’s article for the Communications of the ACM (2006), where she poetically rushes across the range of computer science disciplines and recasts them as essential, general tools for critical thinking and problem solving.

Wing (2008) identifies abstraction as the key concept of computational thinking, and urges CT everywhere for the inclusion of CT across disciplines, and CT for everyone because computing touches all of our lives in myriad ways. As Wing’s CT began to take shape in curricula and instructional methods, Grover and Pea summarize the key aspects of CT they identified in the literature. They found the following nine principles as common and consistent to conceptions of computational thinking (Grover and Pea 2013, 39–40).

1. Abstractions and pattern generalizations (including models and simulations)
2. Systematic processing of information
3. Symbol systems and representations
4. Algorithmic notions of flow of control
5. Structured problem decomposition (modularizing)
6. Iterative, recursive, and parallel thinking
7. Conditional logic
8. Efficiency and performance constraints
9. Debugging and systematic error detection

They also emphasize that computer programming and computational thinking are intertwined, where programming is key to solidifying CT knowledge and essential for evaluating it.

It’s difficult to argue that Wing’s vision has not garnered success. Bibliometric analysis (Figure 9.1) of the prevalence of “computational thinking” in the Scopus databases of scholarly literature indicates CT was not in use prior to 2006 and then slowly grows in publications after Wing’s seminal piece, from 2006 to 2011. The year 2011 appears to be a turning point with a large jump: Published documents grow from 58 in 2012 to 912 in 2023. Much of this growth should be attributed to the adoption of CT by professional organizations like the International Society for Technology Education (ISTE) and the Computer Science Teachers Association (CSTA). More than any other group, though, the nonprofit Code.org (founded in 2012) has been the well-funded lobbying arm for CT. Together these groups seized the idea and popularized it through dissemination of educational materials, grants for schools and teachers, formalized teaching standards, political lobbying (for state education standards and CS teaching certifications), and standardized exams like the Computer Science Principles Advanced Placement (AP) exam from the College Board.

Computational thinking claims to underpin the intellectual processes needed for all fields of study today, offering a way of thinking and approaching problems for everyone from STEM to humanities. Accordingly, proponents argue CT is the dominant, necessary thinking for our age because it powers technological change and anticipates societal needs. It is more likely, though, that the dominance of “computational thinking” in discussions of computer science education stems from its tight alliance to techno-capitalism. The supporters list for Code.org sports the largest tech firms with generous support from private equity and the world’s wealthiest individuals, with more than $60 million coming from just Microsoft and Amazon. From 2013 through 2023 they spent almost $220 million on their mission (Code.org 2023).

Why does the tech industry care so much about education and CT in particular? Because CT accretes power to the industry, and forwards corporate interests. CT’s focus on skills and abilities offers socialized workforce training decoupled from any of the big problems surrounding the industry (e.g., privacy, environmental destruction, racial and gender bias, unfair labor practices/worker exploitation, etc.). It offloads the persistent problem of a significant lack of diversity in engineering and leadership jobs to K–12 schools, asking schools to change their curriculum to create a STEM “pipeline” so the industry can avoid real introspection or radical change that would reshape its workforce and power dynamics. Most importantly, as CT asserts the primacy of CS over all other fields of inquiry and forms of knowledge production, political positions supported by computing gain legitimacy over more democratically derived positions.

Wing’s vision of computer science education, however, is not the only vision and, I argue, not the leading vision. CT’s success occludes other understandings of computing and the goals and forms of computer science education we pursue. By comparing CT to the earlier (and in many ways still dominant) constructionism and to the field of software studies (CT’s contemporary), we find that CT’s content-free formalism and operationalized problem-solving make it a close ally for the dominant powers of late-stage capitalism.

Constructionism: Thinking About Thinking

Prior to Wing, Marvin Minsky and Seymour Papert’s research from the AI lab at MIT had dominated thinking around computer science education since the 1960s. Unlike Wing’s strand of CT, Papert doesn’t find the power of computing in problem-solving or as a way to tackle and solve difficult challenges. While he would agree that computational approaches are closely related to both mathematics and scientific experimentation, their true value is to the fields of psychology and philosophy. Computing offers us, collectively and individually, a window into how we think and know. The study of computing and learning for Papert and his team, under the guise of cognitive science, is an epistemological project. Specifically, Papert (1980) is concerned with an epistemology of “powerful ideas”—foundational ideas that help us organize our understanding of the world. Papert offers abstract physics and combinatorial mathematics as sources for powerful ideas that are better understood within the context of computing. Papert’s epistemology represents a strong contrast to Wing’s operational CT. With constructionism, to know something deeply requires reflection and study of the concept. Computing offers an alienation of the self that allows us to probe, test, and refine the limits of our own knowledge. For the constructionists, computers in schools were a political project aimed (whether effectively or not) at disrupting existing power structures. Used in certain configurations, they would undermine authoritarian models of instruction and unlock organic, emergent learning. Where Wing’s CT seeks an accretion of power to computing and STEM, Papert envisions a Trojan horse (Papert 1997) where computing enables the bricolage, hard fun, and aesthetic experience that create the foundation for an emergent, self-realized learning.

Software Studies and Critical Coding

Both the CT and Constructionist approaches have an affinity for STEM fields and share an interest in the cognitive processes behind computing. As 2006 was the formative year for computational thinking, 2001 saw the rise of the related but distinct academic field of software studies—an approach to studying computing equally grounded in code and electronics, but less focused on operational benefits or individual development. Lev Manovich called for a “software studies” (2001) that uses the tools and concepts of computing to develop new methodologies and grammars to interrogate computational culture. Software studies examine “algorithms; logical functions so fundamental that they may be imperceptible to most users; ways of thinking and doing that leak out of the domain of logic and into everyday life; the judgments of value and aesthetics that are built into computing; programming’s own subcultures” (Fuller 2008, 1). This new approach values neither CT’s operational STEM nor constructionism’s psychology and epistemology. Instead, it explores a new politics and aesthetics necessary for a computationally mediated world.

Alexander Galloway and Eugene Thacker demonstrate the analytical power of the software studies approach in The Exploit (2007), where they explore the possibilities for revolutionary political action in our twenty-first-century networked society. They are concerned with networks and networked power. In traditional asymmetrical struggles, networks confront stable power bases, for example, “peer-to-peer protocols (versus the music conglomerates), guerrillas (versus the army)” (21). Today, these conflicts are increasingly symmetrical: networks vs. networks, where power centers have adapted means of network control. Galloway and Thacker argue that political counter-power must be exercised through an exploit of network topologies, and that the resulting, exceptional topologies “will have to consider the radically unhuman elements of all networks. It will have to consider the nonhuman within the human, the level of ‘bits and atoms’” (22). They choose to use the tools and language of computer science to explore networked power, to show how “networks” are neither inherently democratic nor authoritarian, but exhibit different vectors of control and possibilities for action depending on the specific network topology and protocols. In their writing, they investigate the technical details of networked society, offering an analysis of the OSI model of network layers and protocols to distinguish social models of control from the “informatic control” coded and wired into computer networks (125–26). Galloway and Thacker establish a political goal “not to destroy technology in some neo-Luddite delusion but to push technology into a hypertrophic state, further than it is meant to go. . . . We must scale up, not unplug” (99); we must look for exploits that allow us to use and capitalize on the existing networks of power, to create new topologies that subvert the existing protocols and help us achieve our goals.

Where The Exploit draws on close studies of computer code and technical architectures to articulate a new political posture, Investigative Aesthetics (Fuller and Weizman 2021) uses these tools to exert an “aesthetic power” made possible by the vast amounts of data “sensed” today and new computational approaches that make this data sensible. Fuller and Weissman describe a Forensic Architecture approach used in human rights investigations to uncover abuses such as the use of tear gas against civilians. Their technique uses advanced machine learning, neural networks, and computer vision to scour vast amounts of images and videos uploaded to YouTube, Twitter, and other social media. Their software helps build timelines and evidence of human rights abuses, leading to new forms of creative work like Triple Chaser (Forensic Architecture and Poitras 2019) which is equal parts documentary film, criminal prosecution, and artificial intelligence lecture. This work exploits the hardware, software, networks, and methods of big tech and state surveillance to advocate for justice for marginalized groups.

The software studies/critical code studies approach appeals to those interested in both computer science education and political justice. Aesthetic Programming (Soon and Cox 2020) combines these fields. This textbook sets out to “address the cultural and aesthetic dimensions of programming as a means to think and act critically” (13). It contains both the requisite “Hello world” introduction to programming, usage of arithmetic operators, control structures, and other elements of a programming textbook. It’s structured in a way that new programmers can write code that attends to a “political aesthetics: to what presents itself to sense-making experience and bodily perception” (14). New programming concepts are introduced along with historical and philosophical discussions related to the concepts. It’s likely that studying Aesthetic Programming is not the most efficient route to becoming a proficient coder. But efficiency is not the key objective.

Together, The Exploit, Investigative Aesthetics, and Aesthetic Programming tell us why the software studies approach matters, what new artifacts and knowledge we may create with this approach, and how we might teach it to others. The focus on the material structure of computing and the social and political contexts in which it occurs resonates with some of Wing’s key principles of computing, but unlike CT these other fields show how they can be used to forward democratic goals that push beyond the agendas of state and corporate technologies.

These three approaches to studying and understanding computational culture are clearly not exhaustive nor exclusive of each other. Each has a core commitment to articulating the new ways of thinking that become apparent when we are immersed in networked computing environments. In the end, CT bolsters the status quo, with its promise of formal processes that lead to optimal solutions. We must find ways to expand both the reasons for learning and methods of teaching computer science if we dream of more diverse, interesting, and equitable computational cultures.

References