When was the last time you went a day without crossing paths with an algorithm? Run a search on Google or Baidu, or query Siri or Alexa, and you trigger algorithms that determine which information to feature, and in what order of priority. Visit Amazon, Alibaba, or eBay, and you’re greeted with an algorithmically generated list of products you may be interested in buying. Popular news apps behave similarly, either surfacing or burying stories based on what algorithms have deemed relevant to you. Interested in watching a movie or television show? Hop on over to Netflix, where the company’s algorithms have produced a list of content “recommended for you.” Need a ride? Request an Uber, and in milliseconds an algorithm will process location, productivity, and ratings data to determine which driver to dispatch. Looking for love? The algorithms working behind the scenes at Tinder, Match, Grindr, and elsewhere have you covered. Tired of seeing Facebook posts from that errant cousin of yours? Algorithms have much to do with the frustration (and pleasure) you feel on social media.
In less than three decades, algorithms have become a ubiquitous part of daily life, affecting all manner of human affairs in ways that may be helpful, irritating, surprising, and, sometimes, even a little creepy. This is especially true of societies that have invested heavily in digital tools and accompanying infrastructure. Algorithms seem to be everywhere these days, helped along by the explosive growth of mobile devices, apps, and wearable technologies, all of which provide remote access to powerful, back-end computational systems.
Algorithms have become widespread in popular culture, too. Since the late 1990s, algorithms have become a favorite MacGuffin for scriptwriters, which is to say, a seemingly incidental plot device that nonetheless propels the narrative forward. On television, there are episodes and even whole series in which algorithms figure as such: Numb3rs (2005–10); The Big Bang Theory (2007–19); The Good Wife (2009–16); Person of Interest (2011–16); Black Mirror (2011–); Halt and Catch Fire (2014–17); Blindspot (2015–); Mr. Robot (2015–); Billions (2016–); Silicon Valley (2016–); and surely more. The same is true of popular film, in movies including Office Space (1999), The Social Network (2010), Transformers: Age of Extinction (2014), and The Circle (2017, based on the book published in 2013), among other notable examples. You know algorithms have gone mainstream when a family-friendly feature film like Disney’s Ralph Breaks the Internet (2018) includes in its cast an algorithm named “Yesss,” voiced by Academy Award nominee Taraji P. Henson.
If algorithms are everywhere now, then so too is what you might call an “algorithmic imagination.” Ed Finn (2017) uses the phrase to describe how computational systems such as Google engage in a type of abstract thinking in predicting what you want to search for. Thus, he argues, the category “imagination” is no longer the exclusive provenance of human beings. Maybe so, but this usage misses a critical dimension of the phenomenon it sets out to identify: namely, instances in which ordinary people become aware of, and possibly self-reflexive about, their relationships to formalized, computationally based decision systems. Yesss, arbiter of trends on the fictitious website BuzzTube, is a prime example of this phenomenon, insofar as the character embodies a host of popular assumptions about how algorithms help to curate material online, and how, in doing so, they can affect one’s reputation.
It may be tempting to describe this historical time period as “the age of algorithms,” much as previous generations coined “the machine age,” “the computer age,” “the information age,” and similar catchphrases to mark earlier technological and cultural turning points. Yet, to do so would be to overlook one critical fact: algorithms are nothing new. Physicians have been talking about and utilizing “diagnostic algorithms” since at least 1960, or almost forty years before Google was founded. The expression refers to formalized processes by means of which (human) doctors either confirm or rule out the presence of illness—processes that are in some sense programmed to ensure rigor, consistency, and efficiency. Some of these processes have even begun to be automated. Such is the case in the screening for cervical cancer in which, routinely, computer algorithms now determine the initial test results. Mammography may be soon to follow.
Similarly, one could make a case for how the algorithmic feats of Amazon, Facebook, Tinder, and company duplicate, in significant respects, those of human “cultural intermediaries.” The term, coined by sociologists, refers to classes of knowledge workers whose job is to match cultural goods, broadly defined, to the tastes, needs, and desires of their clients. Examples include librarians; critics, both professional (e.g., literature scholars and film reviewers) and lay (e.g., booksellers and [erstwhile] video store clerks); matchmakers; news editors; retail buyers; museum curators; and more. At the risk of overgeneralizing, cultural intermediaries determine which goods to recommend and which to exclude by weighing a calculus of values against an awareness of their clients’ interests. Their work may not be as systematic as that of their silicon counterparts; they also tend to address a smaller-scale clientele. Even so, doesn’t Spotify essentially do what a perceptive DJ or record store clerk used to do, formally speaking?
Algorithms may be nothing qualitatively new, but, until recently, algorithm was a term belonging almost exclusively to mathematicians, engineers, and computer scientists. Today, ordinary people are trading in it too—so much so that it seems reasonable to suggest it is entering conventional usage. A search for the term in the Google Books database shows a substantial increase in the word since 1950. Not only are algorithms everywhere in digital culture; so, it seems, is the signifier algorithm.
A word’s leap from one semantic context to another is rarely an innocent occurrence. When a term suddenly changes definition or assumes new meanings, or when it experiences a rapid expansion of its user base, it is often behaving along the lines of what the cultural studies scholar Raymond Williams in 1983 called a keyword. Typically, keyword denotes an important term that performs a representative function. The keywords of an academic article are supposed to stand in for, or point to, its major themes. While this definition is somewhat consistent with Williams’s thinking, it is not exhaustive of it. Instead, keywords are terms whose semantic twists and turns betoken shifts in material and social reality and, thus, in our capacity to be and act in the world. An exemplary instance is the word culture, which, until about 1800, referred almost exclusively to the raising of plants and animals in agrarian settings.
Could it be that the word algorithm is undergoing a comparable shift today, or even perhaps that it is helping to usher in different modes of thinking about, conducting, and expressing ourselves?
If algorithm is on our collective minds and in our vocabulary to a newfound degree, and if indeed it is engaging in some deep existential work, then perhaps it makes sense to figure out what it means, more or less definitively. According to computer scientist John MacCormick (2012), an algorithm is “a precise recipe that specifies the exact sequence of steps required to solve a problem.” Procedure figures prominently in this definition as do, implicitly, the values, objectives, and pathways that must be painstakingly prescribed—programmed, if you will—for an algorithm to work not only effectively, but efficiently. Little wonder that in addition to recipe, the metaphorics of algorithm often include words like plans, instructions, flow charts, blueprints, tricks, and templates. Collectively they emphasize form, and thus the extent to which apparently idiosyncratic choices and behaviors may in fact follow predictable patterns. They also provide insight into why algorithms have a tendency to serve up the same types of products, services, and ideas again and again, and to classify groups of people on the basis of racist, sexist, homophobic, classist, ageist, ableist, and other deplorable stereotypes.
The preceding definition of algorithm may provide some context for current dreams and anxieties about algorithms. But, like the word culture, algorithm did not always mean what it means today. It is better to approach prevailing understandings not as definitions but as temporary settlements, or articulations: that is, as semantic resting places carved out at this unique moment in history. This of course begs the question, where else has algorithm come to rest? And what, if any, older senses and meanings endure today as “traces without an inventory,” as Antonio Gramsci (1971) stated in The Prison Notebooks? The question is not “what does the word algorithm mean?” but instead: “in what semantic and social history do you participate whenever you cross paths with an algorithm?”
Apropos, algorithm refers to a person: Abu Jafar Muḥammad ibn Mūsā al-Khwārizmī. He was a mathematician and astronomer of the ninth century CE who lived and worked for most of his professional life in Baghdad, then the capital of the Persian Empire. There he was a member of the *House of Wisdom, a think tank (to impose an anachronism) established by al-Mamun, caliph of Baghdad, to promote knowledge and learning and to assert the intellectual superiority of the Persian Empire. Algorithm is a name, moreover. I mean this not only in the simple, nominative sense but also in the sense of a “principle of authority” as captured in the ancient Greek root, onoma. It is to this sense of the word name that one appeals when one utters phrases such as, “stop in the name of the law.” The authoritative sense of algorithm comes chiefly from an unattributed eleventh-century Latin translation of al-Khwārizmī’s manuscript on arithmetic, which repeats the phrase “dixit algorizmi”—“algorithm said”—throughout. Significantly, this manuscript was the principal source through which both the word algorithm and its alternative spelling, algorism, wound their way into the English language. In other words, al-Khwārizmī was at intellectual ground zero of the “golden age of Islam” and, indeed, a critical player there. More to the point, algorithms may be established authorities both in and beyond the West today, yet they are not strictly of the West, historically speaking.
Thus, they bear witness to the complex and often troubling movement of people, goods, and ideas across the surfaces of this planet, and also then to the territorialization of the planet into distinct, albeit imagined and unequal, cities, empires, regions, and more. Algorithm isn’t just a person, therefore, but also a place, in addition to a language associated with that place. To wit: though al-Khwārizmī worked in Baghdad, his family hailed from Khwarizm, now the modern-day city of Khiva, located on the border of present-day Uzbekistan and Turkmenistan, an area known for exquisite textiles. Although al-Khwārizmī wrote in Arabic, he or his ancestors were likely to have spoken Khwarizmian, a linguistic distant relative of modern Persian. The language would have indicated their belonging to an ethnic minority that was conquered by the Persian Empire in sometimes brutal acts of political and cultural repression. Algorithms, therefore, have long been implicated in the machinations of culture, power, and politics; that is their long-standing predicament. Observers of digital culture have only now rediscovered it.
Culture, yes, but also computation: algorithm additionally refers to a system of numeration. The aforementioned manuscript by al-Khwārizmī not only introduced the word algorithm into European *lexicons; it also was instrumental in popularizing the use of Indo-Arabic numerals. They were called, even into the early twentieth century, the “numbers of algorism.” And here it is important to stress that although numerals may seem more or less given (2 is two, after all, right …?), they are nonetheless specific tools for representing quantity. And, like any tool, they have their own affordances. Have you ever tried performing long division with Roman numerals? This is tantamount to saying that algorithms may be helpful in performing certain types of tasks, yet they also encourage path dependencies that make it difficult to imagine alternative ways of performing those tasks, or alternative outcomes.
Similarly, algorithm refers to technology, particularly to commercial technology. The first recorded instance of the word in the English language occurs in Chaucer’s Canterbury Tales (1387–1400), specifically “The Miller’s Tale,” where the author refers, in Middle English, to “augrim [algorithm] stones.” The reference is to a calculating device resembling an abacus or, more precisely, to one whose stones had been modified by etching Indo-Arabic numerals—the numbers of algorism—onto them. The resulting object embodied both dominant (manipulable) and emergent (symbolic) systems for calculating quantity, generally in commercial settings. In other words, Google, Netflix, and company were hardly the first to monetize algorithms. Chaucer, the great humanist author, happened upon the political economy of algorithms six centuries ago.
Despite their centuries-long public presence, algorithms—or rather, their inner workings—have tended to be shrouded in mystery. Protected as they often are today by intellectual property laws, users rarely know what “ingredients” make up the proverbial “secret sauce.” One also hears from engineers and computer scientists about how the very algorithms they have created function in ways they do not fully comprehend. The mysteriousness has everything to do with the fact that algorithm also refers to a *code. Until the twentieth century the number zero was regularly referred to as “cypher in algorism,” and it was al-Khwārizmī’s manuscript on arithmetic that helped to introduce zero into Western systems of numeration. Previously, the idea of a numeral “signifying nothing” was apt to seem antithetical to the very concept of number, which denoted only positive quantity. Significant, too, is the manuscript’s introduction of the concept of place value (1s, 10s, 100s, etc.) and the use of zero as a stand-in, or placeholder, in the event of a nonvaluation. This may be the basis for what Gillespie in 2016 identified as the “synecdochical” function of algorithm, in which algorithms become convenient proxies for an otherwise opaque network of actors and actions.
Lastly, algorithm refers to the giving of order. Another manuscript for which al-Khwārizmī is known helped to popularize the principles of al-jabr, or algebra, also translated as the “art of reckoning,” or of “restoration and balancing.” The book set out to proceduralize mathematics, laying out a series of rules by means of which to discover the value of unknowns—placeholders, like algebra’s ever-present x—in all manner of problems. As the word algebra diffused into Moorish Spain, it coalesced into the word algebrista, now an archaic form of the word for orthopedist, or bonesetter. The metaphor is an apt one, for the purpose of al-Khwārizmī’s algebra was to cull all the pieces of a disjointed system and set them aright according to predefined rules. This proceduralism, however, is also why some observers considered algebra to be a lesser form of mathematics. The philosopher Edmund Husserl dismissed it as “a mere art of achieving results according to technical rules,” a view that carries over into critical accounts of contemporary algorithmic culture and its tendency to encourage filter bubbles.
What to take away from all these latent and manifest meanings of algorithm? First, an algorithm is a topos or enunciative ground from which to produce authoritative statements about the world. Paradoxically, an algorithm is both a principle of authority and that which grounds the principle itself. Similarly, algorithms are both placeholders for some unknown aspect of reality and the procedures by means of which to fabricate the missing aspect. All this circularity may get at why Gillespie in 2016 referred to algorithms as “talismans”—objects of the human world that appear to be endowed with magical capacities, as if beyond our control. But lest we conclude that we live in a world in which algorithms now perform their work autonomously: the figure of al-Khwārizmī is a startling reminder of the raced, classed, gendered, colonized, and sexualized human bodies that subsist, persist, and insist in every algorithm and in every query. Much the same might be said of the etchings on Chaucer’s augrim stones, which epitomize humanity’s imprint on algorithmic tools—and, indeed, the imprint they have left on us, in the form of an algorithmic imagination.
If algorithm is a keyword, its status as such hinges on the latter insight—namely, on a popular awareness of the degree to which code and computational decision making now orient human affairs. Its status also hinges on our bearing witness to the ways in which long-standing human repertoires are, today, overlaid with and mediated by technical infrastructure. The larger point is to appreciate both the functional and the semantic proximity human beings have shared with algorithms for more than a millennium; it is also to accept that we are hardly the first generation to inhabit an “algorithmic culture.” Algorithms may be programs for solving problems, but they are better imagined, in the abstract, as sociotechnical assemblages—temporally and culturally unique entanglements of people and technology that are nonetheless historically freighted.
Ted Striphas
See also bureaucracy; computers; data; databases; encrypting/decrypting; information, disinformation, misinformation; programming; quantification; social media; surveilling