by
Nick Seaver
View all notes
Available query params: commented=1, untagged=1
[...] Across disciplines undergoing "cognitive turns," key concepts were thus reinterpreted as filtering methods, which protected limited minds from overload. Cognitive anthropology, for instance, reconceptualized culture and classification as an adaptive technique for coping with an overwhelming world: "We classify because life in a world where nothing was the same would be intolerable. It is through naming and classification that the whole rich world of infinite variability shrinks to manipulable size and becomes bearable" (Tyler 1969, 7).
Today [...] bandwidth constraints are largely considered a thing of the past, and recommender systems are no longer commonly framed as techniques for optimizing the bandwidth of a computer network. When people suggest that recommender systems are necessary to manage an overwhelming amount of information, they are not making a claim about digital computers. It would be technically easy, for instance, for Facebook to simply present every update from a user's friends in chronological order. The problem with this much-requested feature, Facebook suggests, is human bandwidth: users would be overwhelmed.
You must be logged in to see this comment.
[...] he has just finished an internship at a large software company known for such techniques, and he is convinced that feature learning is the future. He derisively calls feature representations like MFCCs "handcrafted," suggesting that they are archaic and ready to be replaced.
"handcrafted" (derogatory) is a good idea
The instructor asks us to look for patterns in the space, and the students find one: the movies seem to have separated by genre. Romances are on one side, while action movies are on the other. A student asks how the algorithm knew about genre, when the input only contained rating data. "It seems like magic, I know," the instructor replies, "Magically, we've measured two secret things about these movies." Those two secrets are the two dimensions of our space, although we could have instructed the computer to generate more of them. Our horizontal dimension appears to correspond to genre, as though information about the movies genres were hidden in the data, waiting to be revealed. The significance of the vertical dimension is less obvious, although the instructor suggests that it may reflect how "serious" the movies are.
What made matrix factorization "magical," like McDonald's alchemy, was its ability to uncover such cultural secrets from data that appeared to be about something else. We were being instructed not only in the fundamental spatiality of data but in what the sociologist of science Catelijne Coopmans (2014) has called "artful revelation" -- the rhetorical use of visualization to make manifest hidden patterns in data. Revelations like the one performed in class are a common way to claim authority for data analytic practices. Coordinate spaces are often presented as ways to make data intuitively understandable by people -- an easy means to see, rather than to calculate, the similarities among a set of data points. The geographers Martin Dodge and Rob Kitchin (2001, 30) call images like these "spatializations," to distinguish them from proper "maps," which they reserve for visualizations that have geographical referents. But, as we will see, this distinction is not a significant one for most people working in machine learning. Spatializations effectively summarize differences along various axes into singular, readily comparable distances, appealing to intuitions about the relationship between distance and similarity. When visualized, these spaces are almost always described as "maps."
A grad student told me a joke that captured the impossibility of developing an intuition for machine learning's highly dimensional spaces, attributed to one of the field's most senior figures, Geoffrey Hinton: "How do you deal with a fourteen-dimensional space? To deal with a fourteen-dimensional space, visualize a three-dimesional space and say 'fourteen' to yourself very loudly. Everyone does it."
[...] Richard, an engineer at Whisper, once asked me: "We have to recommend Filipino music to people in the Philippines, but we have no Filipino staff -- what should we do?" Neither he nor I were in positions to actually influence such decisions at the corporate level, and my suggestion, that perhaps his company didn't need to operate in the Philippines, was not tenable within an industry predicated on continuous expansion.
Although critics may refer to Lévi-Strauss when talking about raw and cooked data, they are rarely pursuing this kind of analysis. Instead, they are usually working in the Lakoffian mode, making and evaluating mappings between domains in terms of their adequacy. The anthropologist Tom Boellstorff (2013), for instance, digs deeper into structuralism than most writers on data metaphors, drawing on an elaboration of the raw-cooked dichotomy, where Lévi-Strauss (1966) adds a third term, "rotten," to create "the culinary triangle." In this set, rawness comes to signify unelaborated materials in general: raw materials can be elaborated culturally, by cooking, or naturally, by rotting. In a Lakoffian move, Boellstorff seeks out a data analogue for "rot" and finds it in the notion of "bit rot" -- a technical term for the degradation of data as its material substrates age.
The media theorist Maria Eriksson (2016) picks up Boellstorff's term to describe the lively, error-prone musical data that people like Ellie manage. What Ellie described as "the garden," Eriksson calls " rotten data": this data is neither raw nor cooked but exists in a state of error and decay. Such errors are intrinsic to constantly shifting data, and they persist in the analytic engines of companies like Whisper, making computational outputs more unpredictable and less rational than they may seem. By bringing in the idea of "rot," Boellstorff (2013) aims to draw attention to the persistence of "the unplanned, unexpected, and accidental," identifying a third option beyond the assumed opposites of natural stability and cultural intentionality.
Communism is also the name for the positive alternative to capitalism’s permanent and expanding exploitation, crisis, and immiseration, the name of a system of production based on meeting social needs—from each according to ability to each according to need, to paraphrase Marx’s famous slogan—in a way that is collectively determined and carried out by the producers. This positive dimension of communism attends to social relations, to how people treat each other, animals, things, and the world around them. Building communism entails more than resistance and riot. It requires the emancipated egalitarian organization of collective life.
The comrade relation remakes the place from which one sees, what it is possible to see, and what possibilities can appear. It enables the revaluation of work and time, what one does, and for whom one does it. Is one’s work done for the people or for the bosses? Is it voluntary or done because one has to work? Does one work for personal provisions or for a collective good? We should recall Marx’s lyrical description of communism in which work becomes “life’s prime want.” We get a glimpse of that in comradeship: one wants to do political work. You don’t want to let down your comrades; you see the value of your work through their eyes, your new collective eyes. Work, determined not by markets but by shared commitments, becomes fulfilling. French communist philosopher and militant Bernard Aspe discusses the problem of contemporary capitalism as a loss of “common time”; that is, the loss of an experience of time generated and enjoyed through our collective being-together.10 From holidays, to meals, to breaks, whatever common time we have is synchronized and enclosed in forms for capitalist appropriation. Communicative capitalism’s apps and trackers amplify this process such that the time of consumption can be measured in much the same way that Taylorism measured the time of production: How long did a viewer spend on a particular web page? Did a person watch a whole ad or click off of it after five seconds? In contrast, the common action that is the actuality of communist movement induces a collective change in capacities. Breaking from capitalism’s 24-7 injunctions to produce and consume for the bosses and owners, the discipline of common struggle expands possibilities for action and intensifies the sense of its necessity. The comrade is a figure for the relation through which this transformation of work and time occurs.
[...] In his memoir Incognegro: A Memoir of Exile and Apartheid, the theorist Frank Wilderson, a former member of uMkhonto weSizwe, or MK, the armed wing of the African National Congress (ANC), describes his first meeting with Chris Hani, the leader of the South African Communist Party and the chief of staff of MK. Wilderson writes, “I beamed like a schoolboy when he called me ‘comrade.’” Wilderson chides himself for what he calls a “childish need for recognition.” Perhaps because he still puts Hani on a pedestal, he feels exposed in his enjoyment of the egalitarian disruption of comradeship. Wilderson hasn’t yet internalized the idea that he and Hani are political equals. “Comrade” holds out an equalizing promise, and when that promise is fulfilled, we confront our own continuing yet unwanted attachments to hierarchy, prestige, inadequacy. Accepting equality takes courage.