Black Box
Problem Statement

Algorithmic decision-making affects us in subtle and pervasive ways.
How might we prompt critical reflection on the everyday impact of algorithms?

Nate Smith
Spring 2019
10 Weeks

Our every digital move is tracked, analyzed, and used to refine algorithmic models that not only predict, but shape, our needs and desires. Even as we train these algorithms, they in turn train us, altering both our behavior and the way we think about the world. To what end are these capabilities being employed? This is the question we aim to engage with through the Black Box exhibit, an interactive space that puts you into contact with the other you—your algorithmic “Data Self.”

Algorithmic Arbitration
The field of digital humanities provides useful conceptual tools for thinking about our relationship with technology. Algorithms, in particular, pose a problem for those trying to think critically about the intersection between the digital and physical worlds. These tools are incredibly powerful yet remain opaque even to their creators. They occupy a liminal space between the elegant abstraction of mathematical models and the messiness of reality, a space where decisions are made on our behalf. We must grapple with this algorithmic arbitration not just as designers, but as humans. 

The power and influence of tech companies has become a matter of public concern. Countless articles decrying the “move fast and break things” mentality of Silicon Valley continue to be published. That mentality allowed for the spread of misinformation and adoption of lax privacy standards. Yet few of these articles grapple with the deeper import of the pervasive nature of algorithmic tools and the cultural work they do, molding us into new forms of subjectivity. 

Investigating Perceptions
We conducted a workshop with our peers to examine their assumptions around algorithmic systems. The data from our workshop activities broadly coincided with notions in the popular media. Emphasis was given to the negative influence of recommendation engines and issues around privacy.

Design Principles
︎ Clarify Models

Help users build an accurate mental model of the impact of algorithms

︎ Question Assumptions

Provide new language to question assumptions around technology

︎ Embrace Nuance

Nuance is important—algorithms remain incredibly beneficial tools

︎ Encourage Imagination

Prompt audience imagination for other ways of relating to algorithms

Determining what we wanted to convey was the easy part—doing so effectively presented challenges. We engaged in a lengthy process of presenting concepts and getting feedback from our instructors and peers in order to clarify a compelling design direction. 

One early concept presented visitors with a curated “media diet” of information or entertainment to which they would not usually be exposed. 

Another approach overlayed projections of media feeds on a video of the user, creating a “media mirror”

Final Concept
Referring to algorithms as “black boxes” has become common due to the lack of transparency around how they produce results. Our final concept draws on this metaphor both visually and conceptually. We aim to investigate the inside of these black boxes.
     Upon entering the black box, the audience is confronted with a visualization of their “Data Self.” This is a construct we came up with to convey the idea that, inside a digital service like Google Search or Netflix there exists a model of the user. This Data Self is an accumulation of inferences draw from the user’s decisions and is used to anticipate their needs. Just as the user shapes the Data Self, it in turn shapes them. To show this, our exhibit contains three elements:

︎ Monolith

A central pillar holds two screens, one with a visualization of the Data Self and the other with a scrolling essay on algorithms. The essay is also being read aloud as you walk around the pillar.

︎ Interface

Our audience interacts with the an ambiguous, amorphous interface, a malleable surface whose manipulation mirrors the way we are shaped by the media with which we interact.

︎ Room

The room itself is a dark, intimate space for reflection with projected light guiding movement. 

Achieving the desired result for the visualization required much trial and error. We used Arduino sensing and visual effects produced via JavaScript.
Putting together the exhibit meant long hours in the woodshop. Our goal was to be ready for the year-end UW Design senior exhibition.

Step Inside the Black Box
Your Data Self is the accumulation of information gather about you through everyday digital interactions, organized in a constantly-evolving pattern that determines the way you experience the online world—and the way it perceives you. Our exhibit confronts the ambiguity of our relationship with our Data Selves. We remain unaware of how our actions impact the algorithmic assumptions being made about us, even as these assumptions modify our behavior. Stepping inside the Black Box makes us question the price we pay for the digital services we take for granted. 

Read our essay︎

Designing a compelling interactive exhibit presented a new set of challenges for me. I had never worked with a physical space as a canvas and had to adapt to new considerations. The process of translating theoretical concepts into a cohesive and legible experience was a rewarding, if at times difficult, part of this process. Overall, this project marked a fitting capstone to my evolution as a designer and thinker throughout my time in college.