Core Post: Issues Surrounding Praxis - Kam Copeland

At the beginning of my presentation during our last formal class meeting, I initiated my discussion with a YouTube comment that presented a necessity to transform these intensive theoretical discussions we have, within the confines of academia, into practice. Several weeks ago, in our reading entitled “The Virtual Barrio @ The Global Frontier,” Guillermo Gómez-Peña beautifully interrogated how the web has been represented in a seamless utopian fashion at some academic conferences, despite the degrading, traumatic experiences oppressed groups have undergone within the medium and in the making of new technologies (as detailed by Lisa Nakamura).[1] In this trajectory of my favorite readings during the past few weeks, Safiya Umoja Noble’s chapter, “Searching for Black Girls” expands on how oppressed groups (specifically Black women and girls) have been demeaned by Google’s algorithm. Noble further describes how representatives of Google fail to assume accountability despite removing immediate offensive images, mainly during times in which critical tweets have gone viral (damage control?).[2]  Additionally, Noble describes how Tanya Golash Boza argues that “critical race scholarship should expand the boundaries of simply marking where racialization and injustice occur but also must press the boundaries of public policy,” emphasizing the necessity to translate extensive discussions into practice.[3] In my discussion on how some Black queer web series forums may serve as an example of a queer operating system in practice, I primarily focused on how these critical concepts may be materialized outside of our space of discussion. Specifically focusing on the space of discourse, I find it interesting that many of our discussions (in this class and others) focus on structural issues related to oppressed groups; however, the ambitions of this emphasis do not translate into significant changes within SCA. Perhaps one of the most recurring noteworthy cases surrounds the undergraduate course “Introduction to Cinema”. Although recent discourse around the John Wayne exhibit and its endorsement of White supremacy that maligns innumerable oppressed groups has recently gained some level of institutional recognition, it’s interesting how the large Intro to Cinema class, which functions as the face of the Department among many inquisitive freshmen, has consistently embodied Wayne’s philosophy. For example, during a recent lecture, D.W. Griffith was described as a “good man” and was applauded for his contributions without any interrogation of his harmful representation that literally stimulated the three-fold rise of the Ku Klux Klan and murder of many Black citizens. There are infinite examples in this course and beyond (some cases are far worse...can save for the comments). How in the world can we discuss changing institutions beyond the realm of academia if a significant change cannot even be made within our own Department (over several decades)?

--Kam

---

[1] Guillermo Gómez-Peña, “The Virtual Barrio @ The Other Frontier,” in Electronic Media and Technoculture, ed. J Caldwell (New Brunswick, NJ: Rutgers University Press, 2008); Lisa Nakamura, “Indigenous Circuits: Navajo Women and the Racialization of Early Electronic Manufacture,” American Quarterly 66, no. 4 (2014): 919–41.

[2] Safiya Umoja Noble, Algorithms of Oppression: How Search Engines Reinforce Racism (New York, NY: NYU Press, 2018).

[3] Noble, 80; Tanya Golash-Boza, “A Critical and Comprehensive Sociological Theory of Race and Racism,” Sociology of Race and Ethnicity 2, no. 2 (2016): 129–41.

AI Now Symposium

In addition to the authors for this week, AI NOW had their annual symposium last night in New York. I wanted to share out the footage both to promote their work and to build on our capacious understandings of algorithms as well as the primacy of critical fluency and response.

Core Post 3: Social Patterns of Exclusion

Both the Noble and Bucher readings this week seem to aim at a more material understanding of the impact of the digital. Bucher offers a somewhat tangled (not to discredit her approach, but it seems definitely less straightforward than Noble’s chapter) engagement with the social lives of algorithms/algorithms as social life while Noble straightforwardly argues that algorithms have disastrous material consequences on (mostly) always-already marginalized communities.

While I do agree that there’s a certain reciprocity between the social life of algorithms and the algorithms as social life (as per Bucher), Noble’s piece clearly highlights the genealogy of the uneven distribution of such a system. As Noble argues, algorithmic systems (whether they rank, populate or recommend) are similar to more historical (some are still simultaneous) systems and this seems to bifurcate any of the reciprocities highlighted by Bucher. Generating uneven distributions and even create systems of oppression for certain marginalized populations, seems to be the inherent goal of algorithms as they are tools (among others) of the competing capitalist market.

As someone with a similar professional training in information studies (I studied with Safiya Noble at UCLA) and who has worked with a multiplicity of information systems (in archives, in libraries and with search engines), I’ve learned and seen that these material consequences come in all form: at the UCLA special collections, for example, maybe students expressed their discomfort with some of the language used in the archive’s finding aid. While every institution can usually come up with their own unique language, most use the library of congress’ classification system. This reproduces not only the violence of the state onto marginalized subjects but also it also affects the type of research being done: as a lesbian, would I enjoy encountering the violence of my preferences being classified as a mental illness in my own institution of knowledge? While the archives propose a slower and maybe less far-reaching example than algorithms, the principle seems to remain the same: violent results yields underrepresentation.

Another question that I had while doing the reading this week was: Can a system predicated on language ever do anything else than reproduce the limitations of language itself?

Also, I’d be interested in talking about how this approach to the digital (a much more humanistic, borderline social scientific) differs from the articles we read two weeks ago? What happens to the political in both cases (because I do think that Galloway and Chun, for example, were quite provocatively political in their polemic approach)? Is the reader situated differently by these two approaches? What works and what doesn’t when one decides to approach the subject of the digital from such different angles?

Core post 2: Algorithms in contemporary art

This summer, I happened to run an exhibition at the Hermitage Museum (Saint Petersburg, Russia) that was devoted to the appliance of artificial intelligence to art. All the works by artists from all over the world were based upon the same principle: an artist supplied an algorithm with data – in this particular case, with traditional pieces of art such as “classical” paintings; generators then analyzed that data and created new images (so basically this experiment falls under the rubric of "learning algorithms").

Here are some of the artifacts that I thought might be interesting to look at if we want to see how the appliance of “learning” algorithms works in contemporary art. They illustrate, in an interesting manner, some of Bucher’s insights too (e.g., “What a model learns depends on the examples to which it has been exposed.”)

1. “Memoirs from Latent Space” by Turkish artist Rafik Anadol: the artist trained the newest algorithm of the generative-adversarial network (GAN) in 1.3 million images of architectural photographic memories from the buildings of 9 architects and 11 different historic eras. The network analyzed them and generated unique (never existing) images of facades: https://www.youtube.com/watch?v=yCC5GgakAeE&feature=youtu.be 

2. “The Belamy Family Portraits” by French art collective Obvious: the GAN studied 15,000 portraits painted by European artists from the 14th to the 20th century, and created new images based on this data. It is interesting that the GAN technology is based on interaction between “generator” and “discriminator” mechanisms: the former creates, the latter constantly compares new portraits with those by the generator searching for the unlikely; the generator’s task is to deceive the discriminator and make him think that a new image is a real portrait: https://obvious-art.com/gallery.html

3. “Summer gardens” by Italian artist Quayola: he filmed floral compositions manipulated by high winds, and employed computer algorithms to analyze motion, composition, and color schemes in these shots. The machine synthesized and processed macro shots to single out abstract paintings, and the result turned out to be something alluding to French impressionist paintings: https://vimeo.com/195022642

Since being exposed to these pieces of art, I’ve been haunted by the question – what is actually the creative process based on the interactivity of a human and a computer algorithm? And, as a follow up to Bucher’s piece, to what extent these methods include and depend on human decisions and choices?

Core Post: Algorithms & Their Others

“What is generally asked of an algorithm is that it produce a correct output and use resources efficiently” (Bucher 23) or “‘to produce a desired outcome’” (Kitchin in Bucher 21), “encoded procedures for transforming input data into a desired output, based on specified calculations” (Gillespie 1). I am curious the relationship of desire to algorithm and procedure, how the propositions and inputs and systems are set up willing the outcomes into ‘right’ and ‘wrong’ echoing in the repeated emphasis on the ‘correct’ answer in these definitions, by which these technologists likely mean the answer that solves the problem they’ve proposed. However, this discipline’s methodological orientation toward problems and correct solutions may predispose them toward particular types of solutions that feel or seem correct based on criteria that come with their own constraints and biases—we might look to the feedback loops of Gillespie’s items 1) “patterns of inclusion,” 3) “evaluation of relevance,” and 5) “entanglement with practice: how users reshape their practices to suit the algorithms.”

What would it mean to consider ‘solutions’ that come out of different desires? Already in these feedback loops of algorithmic thinking, how can we begin to imagine these? Is it as binary as all that either-or? Here’s where I take the algorithm for a walk.

Desire lines are unpredicted trajectories across distance, unanticipated data. If desire is an artifact of distance, if longing a length, an impulse to move toward that implies and requires space. One approach to desire is to quench it, and the procedural is an effort to foreclose, to solve. “Machine learning algorithms reduce this uncertainty by making predictions about the likelihoods of outcomes. Put differently, machine learning is about strengthening the probability of some event happening, based on evolving information” (Bucher 28). I am interesting in this Xeno’s paradox of uncertainty, probabilities, distances that cannot entirely foreclose or finish. Where Bucher says, “an algorithm as ‘a strategy or plan of action’ [...] always in becoming since events are not static but unfolding” (paraphrasing Chun 28), this version suggests a process-oriented algorithm—almost an A Thousand Plateausean mood through which to reframe: becoming-fill-in-the-blank. I went back to Chun’s original, and unfortunately I think Bucher misreads... Chun was critiquing such thinking, ends reflecting back to justify means. But I’ll leave the thought, because what if we were to imagine—in addition to what Bucher argues, that “algorithms have the ability performatively to change the way events unfold or, at the very least, change their interpretation” (28)—that the algorithm itself is changed, always imperfect and in progress? This is much closer to how machine learning is working these days anyway, and makes me think of how Luciana Parisi (2013) talks about anticipatory systems in Contagious Architecture, systems that rather than being merely interactive, “sense and anticipate (or productively prerespond to) changes in atmospheric pressures, moods, sounds, images, colors, and movements, incomputable data have infected the general ecology of media systems” (25). To anticipate, to respond to another’s desires, one must be able to adapt, and are there queer potentials and desires possible in such systems? Could they be imagined?

How do we contrast this question of different desires and works-in-progress with something like “power that works from below [... that] becomes indistinguishable from life itself by sifting into ‘the capillaries of society’” (Lash in Bucher 34)? How do we know different desires within proceduralized systems predicting us under our skins, algorithmic power as “immanent life force”? But are we willing to grant it that much power, or is that only to desire the algorithm itself instead—to make it the fetish object instead of a means of investigation or access?

Links for Presentation/Discussion

Simplification: Nick Montfort, @one_algorithm

A bot (poem) that simplifies to the point of boredom the mystique of algorithms into a procedural, almost bureaucratic walkthrough of its permutations until it quits after having exhausted them all. 

Complexification: Jer Thorpe, You Say Data, I Say System

“The next time you read a story with the word data in the headline, swap it out with data system. When you see a data visualization, think of it instead as a data system visualization. If the government proposes new policies around personal data, think about them instead as policies about people, and the data systems which they inhabit.” 

Ancient Babylonean Algorithms: The Earliest Programs

“Babylonian Programming: [...] they represented each formula by a set-by-step list of rules for its evaluation, i.e. by an algorithm for computing that formula. In effect, they worked with a 'machine language' representation of formulas instead of a symbolic language.” 

Each algorithm closes with “This is the procedure.”

Sarah Ciston, Uncertainty Index

How does the body digest the computational and conform to it? Conversely, how does the algorithmic procedure (fail to) capture the non-binary, doubtful, less-than-ideal human interface? How do Gillespie’s “cycles of anticipation” converge into grotesque portraits of subjectivity that culled from “creepvenient” interfaces?  

Core post 3: Algorithms as self-affirming fortune cookies

When thinking about power, one cannot escape Paulo Freire and the Pedagogy of the Oppressed (1968). The power structures he emphasizes are so applicable to digital technologies, mechanization, and algorithms. As sources of revolutionary –and sometimes political– struggles, we find ourselves yet again before power structures of the oppressed and the oppressor. Man-made algorithms turned mechanical, but cannot escape their sociotechnical factor, both of those responsible for creating it, and those that feed it its knowledge, and reinforcing the embedded power in the algorithm. Even when disembodying Freire’s dialectical materialism, stereotypes are reinforced by these machine-learning processes. I understand that algorithms are created by humans who have certain understandings of the world, but I can’t help but unveil the sociological –and public– impact of these search results. “We have to ask what is lost, who is harmed, and what should be forgotten with the embrace of artificial intelligence in decision making,” reads Safiya Noble’s Algorithms of Oppression. It is a social and collective process, both of thought and organization, that reifies the marginalized, that finds as a first response to black girls search results of pornography. We take pleasure in looking; the gaze is unescapable, where diasporas are inflected, intersectionality is overthrown by vulgarity and the carnivalesque (Bakhtin, 1968), why do machines keep repeating a when looking into Black Popular Culture that scholars and activists have fought against for years? Who mediates this search engine we all rely on? Should it be mediated, since that would be mediating ourselves. We try to understand their complex nature, embedded with meanings and ways of seeing the world, they are part of the social fabric (Bucher, 2012), which makes the disentanglement barely impossible. I recently considered the power of the gaze related to algorithms, into the conception of publics –or production on calculated publics, in Gillepsie’s terms– and how something is perceived and the point when something else becomes public –particularly when talking about places and Google Street View. Nevertheless, there is also the possibility of a lack of gaze, when an algorithm decides something will no longer appear as a result, an not becoming part of the public. Balka’s (2011) shadowed bodies are also a direct impact on the gaze, that which can also be harmful, even when it comes from below in a capitalist society. The possibility of shadowing bodies makes objectiveness a concept complicated to grasp, for it is still related to how people use data, what they decide what to search for, and how they see themselves as a source of publics (Gillepsie, 2014). We find patterns of inclusion, we exclude, becoming part of the algorithm with the hope of being included, to be seen and feel relevant. We, as the machine-learning systems, become chameleonic and volatile, because we know that there is no impartiality in algorithms, even by acknowledging algorithm logic as self-affirming. We try to find affirmation, a connection to that trust system we are embedded to, to make sense of it all. Even so, we also rely on magazine horoscopes and fortune cookies for affirmation. That is, perhaps, what sets us apart from the machine.

Critical Mediations: The Unruly Canon - Conference

Critical Mediations, the annual graduate student conference in Communication and Cultural Studies at Annenberg, is happening on campus at ANN (Wallis Annenberg Hall) tomorrow! We guide this year’s conference with a series of questions: how does the meaning of a cultural object change once it has been deemed ‘canonical?’ How are these determinations made? When writing (and re-writing) the genealogy of Communication and Cultural Studies, who is given a platform, historical or otherwise, at the expense of others? How can canonical texts be reread or reclaimed for contemporary, potentially oppositional use? Your classmates, Paulina and I, are part of the organizing committee and we think many of the panels will be of interest to this class :) Miles will be starting the day off as the first speaker for the Pop Culture's Canon(s) panel, and we will be ending with a keynote by Dr. Paula Chakravartty from NYU. 

You can see the full list of panels and talks here: 
https://criticalmediations.org/schedule/

If you're interested, please register here: 
https://criticalmediations.org/registration-2/

We hope to see you there! 

Ed & Paulina 
Critical Mediations