Review by Geoffery Gimse, University of Wisconsin-Milwaukee

Printer-friendly versionPrinter-friendly versionSend by emailSend by email

What Algorithms Want: Imagination in the Age of Computing

Ed Finn
Cambridge, MA: MIT Press, 2017. 272. Book.

#TCVol7Gimse


Cover of What Algorithms Want: Imagination in the Age of ComputingThe title of Ed Finn’s book, What Algorithms Want: Imagination in the Age of Computing, highlights an ambitious subject, the possibilities of an algorithmic imagination. This topic may lead the reader to expect a deep dive into the nature of algorithmic design and a look ahead to what comes next. To be sure, Finn provides this, yet, more importantly, he uses the algorithm and algorithmic imagination as way to look at the evolution of modern culture itself, a hybrid culture that is neither entirely human nor machine. For Finn, the algorithmic imagination cannot be fully considered without such a cultural positioning (186). In so doing, he gives the reader a glimpse into the future of the algorithm and, perhaps, culture itself.

Finn begins with a discussion on the nature of algorithms and how they have changed over time. In his first chapter, he sets out to define what an algorithm is and how its definition has become linked to broader social and cultural understandings. He traces the evolution of the algorithm through developments in computation, cybernetics, and mathematics while at the same time interlinking different critical analyses of the impact and nature of those developments. This movement between techno-historical background and critical analysis allows Finn to map the transformation of the algorithm within a distinctly cultural context. This transformation is one that he sees as a movement from a pragmatic definition of algorithm to a computationalist one. The pragmatist's definition of an algorithm, for Finn, begins as a "way of solving a problem" (17). Of course, the problems that modern algorithms seek to resolve involve more than code and often directly collide with a wide variety of cultural factors. In these instances, the pragmatist's algorithm becomes a systematizing force that seeks to categorize and define social and cultural constructs in the interest of resolving that problem (18). The algorithm then, from the pragmatist's view, is a description and categorization, with abstraction, of the broader world. Finn argues that while much of modern technology has been created via this pragmatic view, a new "computationalist" understanding of the algorithm is developing (21-22). The computationalist view is less interested in the idea of the algorithm as a description of the world and more interested in mapping the world into algorithmic structures.

This shift from algorithm as a description within a machine to algorithm as a complex system embedded within a culture is an important one. It has turned algorithms into powerful forces for accessing and changing the very nature of the culture. Algorithms themselves become complex cultural machines that drive modern society by abstracting the social, political, and economic structures upon which the world runs making them accessible and "effectively computable" (24-25). In the computationalist sense, algorithms operate much like language and carry with them, in their implementation, the same capacity to alter conceptions of reality on a fundamental basis (39-40). It is in implementation that the goal of the code comes in contact with the broader culture. The algorithm, as Finn describes it, sits at the gap (47). Throughout the rest of the book, he uses discrete examples to highlight the code-to-culture gap and the roles algorithms play within it.

In the second chapter, "Building the Star Trek Computer," Finn open with a fascinating and relatively little-known exploration into the military history behind Apple’s Siri. He then connects this history to Google's attempt to design predictive algorithms based on its vast data troves. What Finn sees in both Google's and Apple's attempt to develop digital assistants is an ongoing transformation in the relationships between people and computers. It is here where he turns to science fiction highlighting both the Star Trek computer and the operating system from the movie Her. In all of these instances, fictional and real, Finn highlights the shift from algorithm as a system that collects and stores of data and information to a system that lives with, shares, and predicts the needs and interests of the individuals around it. This interconnection between the human and the machine once again seems to be an excellent description of the gaps that Finn describes within implementation. What is intriguing about Finn's portrayal of the gaps in this chapter, is that he does not see this solely as a limitation of the machine but of the human, as well.

Chapter 3,"House of Cards: The Aesthetics of Abstraction," builds not so much on these limitations but on a movement toward interoperability. In this case, the gap in implementation becomes a space in which neither machine nor individual can work effectively on their own, but when working cooperatively can produce significant and effective results. The Netflix algorithm, a product that is a direct combination of human and machine labor, seems to offer Finn a ready example. As Netflix changed from a mail-order, DVD rental service to a digital streaming service, the needs of the sorting algorithm shifted to meet the changing needs of the audience. An algorithm that was capable of not only of describing existing content but of predicting and responding to a user's interests and needs was required. Netflix did not rely solely on machine processing to build this new algorithm, however. It employed human marketing agents as taggers who worked in concert with specialized algorithms that tracked its users and their interests to help develop and shape potential predictive categories. Netflix then used this same apparatus to custom design and tailor House of Cards both in terms of content and marketing to its audience with tremendous success. Finn argues that this "cyborg apparatus" made Netflix's recommendation system so successful by creating the ability to recognize, manage, and take advantage the gaps between machine and culture. This type of service that focuses on traversing that gap is what Finn terms a form of "algorithmic arbitrage" which he believes will only become more vital and important in the days ahead (110).

The role of the “cyborg apparatus” becomes even more evident as Finn moves from his discussion of Netflix and into a broader analysis of the role of work and play in algorithmic spaces. In Chapter 4, "Coding Cow Clicker: The Work of Algorithms" he highlights Ian Bogost’s creation of Cow Clicker. Developed as a social critique game, it became an unexpected, viral sensation. While designed as a parody of social media games, it evolved into a version of the very media it sought to critique. Finn shows how algorithms blur the lines between work and play (p. 121), leading to the rise of an algorithm-driven, gamified, sharing economy epitomized by companies like Airbnb and Uber. It can also been seen in Amazon’s Mechanical Turk, which is perhaps one of the more powerful examples that Finn explores. Mechanical Turk’s capacity to blend human and machine labor, what Finn calls “hybrid labor” (144), in a virtual workshop demonstrates how algorithms operate not just as processes for machines, but for culture as well.

While Amazon’s Mechanical Turk provides an excellent example of hybrid labor, Finn uses algorithmic stock trading and bitcoin to illustrate how algorithms have become further intermeshed into definitions of value, work, and financial interest. In chapter 5, “Counting Bitcoin,” he explores the impact of these changes. It is hard to imagine a better example of the computationalist view than replacing an economic currency with an algorithmic one. While Finn points to bitcoin as a potential next step in such a movement, he shows how this change has already taken place in large financial markets where human traders often play a secondary role to machine-driven algorithmic traders.

Throughout the book, Finn does an excellent job of making technically complex topics accessible and positions them in terms of social and cultural relevance. In his coda, however, he brings us back to the title of the book. What, after all, do algorithms want? He suggests that the space of implementation may never provide for a direct method of access to the algorithmic imagination and that researchers and artists, alike, are just beginning to understand what such an imagination may entail. More importantly, for Finn, is that fact that this imagination extends not just to the algorithm but to the human as well in a form of collective imagination (185-186). In this hybrid space, the human and algorithmic build upon and change one another. Algorithmic desire becomes linked to its human counterpart and neither can be considered wholly separate from the other (190). For Finn, there is possibility here and danger. He pushes humanities scholars to explore these concepts through new styles and approaches to algorithmic reading. Building on the art and scholarship that has come before, he hints that that an understanding of algorithmic imagination and desire may in fact provide modern humanity with a better understanding of itself.

 

Biography

Geoffrey Gimse is a doctoral student in professional and technical communication at the University of Wisconsin-Milwaukee. His research interests include the social impacts of digital architecture, data structures, media management systems, and participatory culture in online and offline spaces. His current work examines how different digital tools and development strategies impact user publics and the people within them.

 

© 2017 Geoffrey Gimse, used by permission


Technoculture Volume 7 (2017)