As part of the Association of College and Research Libraries (ACRL) Digital Scholarship Section (DSS) Open Data Week 2019, I participated in an open research discussion group: “Open Data Activism in Search of Algorithmic Transparency: Algorithmic Awareness in Practice,” led by Montana State University (MSU) researchers: professor Jason Clark and research assistant Julian Kaptanian.

At the time of the discussion, Clark and Kaptanian were in the process of concluding an IMLS-grant-funded research project entitled “Unpacking the Algorithms That Shape User Experience.” The ACLR DSS presentation built off modules and workshops that Clark and Kaptanian had run in the past year as part of the IMLS research project, exploring how building user competencies and empowering technology users on a personal level is a form of activism. You can learn more about the grant project here.

A ‘Symptom’ of Technology

Clark and Kaptanian grounded the discussion by characterizing algorithms as a ‘symptom’ of routine technology use. Like a cough to a cold, algorithms can be the less than desirable phenomena that shadows the data generated from our daily computer-mediated transactions. However seemingly inexplicable, algorithms have real consequences.

To illustrate this point, Clark recounted how online platforms amplified the incorrect online identification of the Las Vegas shooter in 2017 by pushing 4chan reddit users conspiracy theories in the online search results following the mass shooting. Within hours, an innocent man faced online harassment and blacklisting, an ordeal to which Google and Facebook simply responded ‘the algorithm did it,’ begging the question: what’s behind an algorithm? (Figure 1.)

Clark defines algorithms as the “computational processes embedded into our software” that in turn “predict, recommend, and speculate about our interests” in our all digital interactions. This is to considerable effect and risk as Gillespie warns in “The Relevance of Algorithms,” “as we have embraced computational tools as our primary media of expression, and have made not just mathematics but all information digital, we are subjecting human discourse and knowledge to these procedural logics that undergird all computation.” (Gillespie, 2014, p. 168) Clark then asks what if the “ghost in the machine” was understood by technology users and an “interrogation of algorithms” was a fundamental element of the digital environment? (p. 169)

Open Data and Algorithmic Awareness

Clark grounds this call to action in the EU General Data Protection Regulation (GDPR) “right to explanation” or “meaningful information about the logic involved” in an algorithmic decision. Clark posits GDPR is an international opportunity to demand algorithmic transparency and therefore positions digital literacy as a form of activism.

Clark and Kaptanian led a series of “algorithmic awareness” exercises that they piloted with MSU undergraduate students. First, they broke down the core functions of an algorithm including searching, filtering, ranking, and parsing information through illustrating the “weighted graph” of how Facebook ranks your connections online which in turn shapes your Facebook feed, a theoretical concept which is readily understandable to a user of social media.  

In the next exercise, Clark and Kaptanian aim to demystify the technical aspects of the algorithm by utilizing ‘pseudocode’ through which participants are asked ‘program the library’ or code different goals with the possible actions within a library to reach those goals. For example, the goal of ‘research’ could be achieved by the possible actions: ‘reference desk’ and ‘computer lab.’ They also introduced ‘methods’ as an added layer for achieving the task, like ‘emailing a librarian,’ as a tangible approach to the if, and, or logic underlying all code (Figure 2).

Despite the moderator’s best efforts to explain the technical structure of code through the tangible and familiar spaces of the university library, the exercise in practice proved difficult for the participants, requiring double the time to complete than the suggest 5-7 minutes. However, it was the follow-up questions to the exercise that proved the most valuable in understanding the limitations or code. Kapitanian asked the group who we envisioned as our audience in generating our code and we all answered differently. Some participants envisioned students, other’s faculty, staff or even an outside visitor. Collectively we came to discuss, however bound by brackets and formulaic syntax, our ‘code’ was still limited by our embodied experience enacted within social structures. Therefore, despite their neutral appearance, algorithms and the information they retrieve are subjective (Bates, 2006, p. 11-12).

The presentation concluded with a brief discussion of data profiling, and the steps users can take to understand what personal data is stored in platforms like Facebook and Google by walking participants through how to view and download their digital profiles. For many in the discussion, this exercise was nothing new and limited because their Ad Personalization feature in Google was already turned off.

‘Pedagogy for an Algorithm’

Moreover, the discussion “Open Data Activism in Search of Algorithmic Transparency: Algorithmic Awareness in Practice,” both highlighted the urgent need to build algorithms awareness into digital literacy efforts and while offering tools for educators and students to build that competences and ultimately framing that as activism.

The survey circulated at the close of the event, further emphasized these points. The survey solicited the level of resources and education around algorithm awareness at my current institution as well as asked at what grade level I thought it appropriate to introduce digital literacy.

My response: elementary school, or as soon as students begin to start to seriously engage with the internet.

At a moment when society is attempting to take a step back to fully understand the ‘ghost in the machine,’ it is important to see the opportunity in building digital literacy as a safeguard against current risks and advocate for change or open data in the future.

Works Cited

Bates, M. J. (2006). “Fundamental forms of information.” Journal of the American Society for Information and Technology 57(8): 1033–1045. https://pages.gseis.ucla.edu/faculty/bates/articles/NatRep_info_11m_050514.html

Gillespie T. (2014), “The relevance of algorithms” in Media Technologies: Essays on Communication, Materiality, and Society, eds. T. Gillespie, P. Boczkowski, and K. Foot. Cambridge: MIT Press, 167–194. https://www.microsoft.com/en-us/research/wp-content/uploads/2014/01/Gillespie_2014_The-Relevance-of-Algorithms.pdf

Figures Cited

Figure 1.

Wakabayshi, Daisuke (@daiwaka). “Google statement on how 4chan thread identifying the wrong man as the shooter showed up “in the news.” October 2, 2017, 10:32 AM, Tweet.

Figure 2.

Clark, Jason, 2019, Github algorithmic awareness pseudocode template. https://github.com/jasonclark/algorithmic-awareness/blob/master/modules/one/library-pseudocode-exercise-template.py