Phase VI: Filter Bubbles & A Pivot in Thesis

An Updated Proposal

This is a continuation from this post.

It is no longer a secret that our current technoscape doesn’t reflect the opinions and politics of majority of this world; on November 8th, this was made abundantly and very rudely clear. Vast parts of the world and their ideologies are hidden from us (I cringe at the sharp distinction, too). The ways in which information is presented to us also mediates different types of thinking, and therefore has politics. How might we leverage design to start recognizing this?

The reason for my contextual pivot is that my research into Machine Learning (ML) systems and bringing ethical awareness to the field, combined with the recent election has led me to shift my focus onto Social Media (SM) platforms, where massive amounts of info shared from peer to peer create interesting friction, especially in light of the election. Because of the technical constraints around ML, I felt that it would be a more useful approach to look at the values around people’s perception of ML, rather than try and work with the algorithms themselves. I also believe that communication design combined with lightweight speculative design methods would be the best approach to explore the questions around values.

I’d like to start by analyzing the grid systems in which information is presented to us on Google search or different Social Media platforms, conduct research on how people mind-map the information that they’re shown, speculate on how they ought to be shown, in order to:

1. Promote thinking about the world in ways that respect spatial and temporal qualities, then create toolset/methods with which one might do this in the context of Machine Learning & algorithm models funneling us down singular paths, and;
2. Ideate on how it might be designed & presented and test it to speculate for an alternate future.

I then decided to seek out mentors that I thought might help me solidify my thoughts*

*thanks to Ahmed Ansari, Stacie Rohrbach, and Cameron Tonkinwise for their insights and guidance.

Some literatures that could help define my thoughts are:

• Bernard Stiegler’s symbolic misery: vast parts of society that are effectively invisible
• Peter Sloterdijk’s concept of Spherology (we live in a sphere composed only of known things) and Immunology (we’re immune to ideas that are novel to us)
• Alec Borgmann’s concept of technology and the character of contemporary American life (Verbeek has a nice summary section on Borgmann’s argument on democracy & capitalism)

Discussions around these concepts led me to question:

• how does technology shape the ways in which we’re thinking about the future?
• search algorithms and machine learning: how do the biases shape what we believe in? and what kinda future we see for ourselves?
• reimagining machine learning: what would happen if it weren’t trained on data?
• can we build scenarios or maps of whose views are invisible and unknown to us? and to speculate on what CS might do in the future, in building ethical ML systems?
• Google presents things in grids/lists that aren’t representative of the real world — could we build mind maps/conceptual cartography of said real world? Can it be a cartography? of other knowledges that respect space/temporal/etc, and analyze the platforms in their nitty gritty details?
• I need to take into account the nature of a tweet modifying the way we treat that information, and same with how we treat information on a Google search or a facebook feed (think in terms of metaphor). What does this mean politically? How does it shape / mediate reality?

Platforms mediate the way in which things are said, their tones of voice, etc. It’d be interesting compare & contrast the different ways that they are said.

Slack’s voice is friendly but also short-lived and extremely egalitarian in that it treats all threads of conversation equally, while also lending the conversations a sense of levity. Twitter’s voice is terse, in that the user doesn’t have to take the time to argue why he feels and thinks in a certain way — the platform doesn’t support this kind of nuance, like Facebook does. How does the medium create a whole new kind of politics? It would be interesting to think of the UI elements as metaphors (i.e. does the comment feature remind you of casual gossip? a formal chat?) Could we leverage design to create a toolset to think more deeply about the ethical implications of the technology of social media? And what would the design research methods look like?

Machine Learning is a great case study, in that it applies to everything. Its various applications include many parts of the current tech scene, from newsfeed algorithm models to self-driving cars. Looking at it from a socio-technical point of view, how does Machine Learning mould and mediate the diff perceptions of people?

Seeking Mentorship on the Subject

While chatting about my recent and abrupt pivot with a colleague, I was recommended to speak with Stacie Rohrbach, CMU IxD program’s chair of Communication Design, as her latest class had worked on a similar project and prompt. I explained to her my predicament and reason for this shift, and she provided me with the process documentation from her class’s project.

Then, together we explored questions like: why don’t we know what it is that were the deal breakers for the conservatives? The liberals (myself included) could see past Hillary’s flaws because they seemed trivial in light of Trump’s fundamentally immoral ones, but what broke the conservatives?

This got us ideating on a couple of different ways to research the bubble phenomena:

• would the first step in researching for this design problem asking both side of the spectrum: “what were the things that you could look past and not?”
• could I possibly find groups of people that voted for the opposing sides (i.e. 5 from democratic and 5 from republican)? And then, what if I could show them the news feeds from the opposing side (give the democratic voters a feed from the republican side, and vice versa) and then have them look at them together? Depending on what research methods to conduct next, it may give me an understanding of how people read these opposing content, to get to the heart of the matter which is: why do they live within their own bubbles?
• what’s it like if they’re confronted with their bubble, then the others’ bubbles, then compare them? It’d be curious to gauge their reactions, and how their interactions change.
• could I measure their reactions on a scale system, by using methods of semantic differentials to qualify the subtleties of research results?
• and lastly… what do I, as a designer, do with the results?

We see complacency everywhere we browse, and contents that would only magnify our existing beliefs. We were blindsided by this election (I’d like to believe) not because we are ignorant, but more because we simply didn’t know what the other side was thinking. So, how do we use design interventions to benefit those instances?

Reference: a bi-columned feed is interesting, like this WSJ blue+red feed experiment.

Coincidentally, there was an amazing exhibition going on at the Miller Gallery on the subject of Speculative and Critical Design, curated by Ahmed Ansari, Deepta Butoliya, and Katherine Moline, which made me request a meeting with Cameron Tonkinwise, the Provocateur of the exhibition. During our chat, Cameron raised the following questions:

• what would be the value in basing the findings to the ‘human values’? who would be the judge?
• what if I did a research by design? speculate on 4–5 diff extreme scenarios and asked the panel of the thesis jury: “what do you think of this?” how do you think ppl might rethink? how will it reshape their behavior? how would you do it? what would be the consequences of it?
• carbon credit: “You’ve read 4 democratic-leaning articles today, make up for it by reading more rep articles.”
• what if, upon posting and sharing, the person tags that post and specifies why they’re reposting/sharing this? i.e. I only agree with this 50%, but still think it makes some sane points. This way, the others will know that the post does or doesn’t come with the poster’s full support.

Reference: the slate article about switching fb feeds with each other for 5 wks prior to the election.