Overview

Echo is an interactive virtual reality project that offers the player a sensory journey through a series of echo chambers constructed from a series of tweets. The experience bombards the eyes and ears with a number of increasingly polarized tweets. At the end, the player is presented with a visualization of the player's position on a political spectrum of a given topic. It was a group project at UC Berkeley in the fall of 2019.

• All
• Intelligibility Types
• Explanation Types

What

This is a longer card with supporting text below as a natural lead-in to additional content. This content is a little bit longer.

Last updated 3 mins ago

Why

This is a longer card with supporting text below as a natural lead-in to additional content. This content is a little bit longer.

Last updated 3 mins ago

Global Feature Importance

This is a longer card with supporting text below as a natural lead-in to additional content. This content is a little bit longer.

Last updated 3 mins ago

Decision Tree Approximation

This is a longer card with supporting text below as a natural lead-in to additional content. This content is a little bit longer.

Last updated 3 mins ago

What

This is a longer card with supporting text below as a natural lead-in to additional content. This content is a little bit longer.

Last updated 3 mins ago

Why

This is a longer card with supporting text below as a natural lead-in to additional content. This content is a little bit longer.

Last updated 3 mins ago

Global Feature Importance

This is a longer card with supporting text below as a natural lead-in to additional content. This content is a little bit longer.

Last updated 3 mins ago

Decision Tree Approximation

This is a longer card with supporting text below as a natural lead-in to additional content. This content is a little bit longer.

Last updated 3 mins ago

What

This is a longer card with supporting text below as a natural lead-in to additional content. This content is a little bit longer.

Last updated 3 mins ago

Global Feature Importance

This is a longer card with supporting text below as a natural lead-in to additional content. This content is a little bit longer.

Last updated 3 mins ago

The Problem

Echo chambers are exacerbated by personalization

The echo chamber effect is created by human nature. People’s confirmation bias result in environments where people interact with others with similar mindsets. With the affordances of the internet, people are more able to search and consume the perspectives that they’ve already agree with. Thus, the echo chamber has been exacerbated by the advancement of technology.

With targeted algorithms and personalized content, online platforms like Google, Facebook, and Amazon provide users the content that they are more likely to be interested in. The personalization not only offers users more efficient access to their desired information but also increase the companies’ ad revenue. However, while curating users’ personalized content, the algorithms also filtered out all the “unwanted” information away from you based on your behaviors. That is how the filtered bubble is created. It will be more and more difficult for diverse people’s viewpoints online due to personalization.

Filter bubbles are not going away

At the moment, the companies are still developing personalization through machine learning and artificial intelligence. Making a more accurate filtered bubble for every user increases the revenue from advertisements. So, what can we do if the big tech companies don’t want to dissolve the filter bubbles?

How might we encourage people you look outside of their echo chambers?

Pariser, Eli (May 2, 2011), “Ted Talk: Beware Online’ Filter Bubbles’”

Project Objective

Using VR as a medium

How might we simulate an experience the embraces the power of virtual reality? For this two-week project, we, a team consists of two designers, an architect, and an engineer without any in-depth knowledge about virtual reality nor developing in Unity, were tasked to create a project based on an Oculus Quest.

“We can continue many debates about how virtual, mixed, and augmented reality can be both empowering and isolating – exhilarating and nauseating. However, there is one primary element that we argue is not in debates: the power to change perspective, for better or worse.”

— Excerpt from Project Prompt

Preliminary Research

Current solutions and the power of VR

Exposing opposing perspectives can increase polarization

One of the possible solutions to the issue is trying to expose people with different perspectives. However, Chris Bail, a professor of Sociology, Public Policy, and Data Science and the director of Polarization Lab at Duke University, concluded in his paper saying, “Although we found no evidence that exposing Twitter users to opposing views reduces political polarization, our study revealed significant partisan differences in backfire effects.” Therefore, we should explore solution within the player’s own viewpoint.

Preliminary Research Insight

• 

  Fact-checking is not the solution

  When President Trump was fact-checked for multiple tweets or claims, his supporters were not swayed by the accuracy of the messages. Moreover, phrases like “alternative facts” were used to defend fact-checking. For some people, facts don’t matter. It’s emotions that drive decisions that seem ludicrous even after being disputed with facts.

• 

  Anchoring effect - the order of information matters

  “Typically, the first bit of information we receive becomes an anchor and all future evaluations are based on this anchor piece of information.” As a result of the anchoring effect or cognitive bias, it would take much longer for readers to adjust their existing opinions even when they are incorrect.

Leveraging VR as a storytelling medium

We also researched the effects of a VR experience, as well as what we can accomplish within the 2-week time frame. After some research, we found out the following qualities that we can leverage while solving the echo chamber effect.

Combining the emotional stimulation and the inescapable experience, game developers have been leveraging VR as a storytelling tool in the horror genre. At the same time, we would like to design an experience to motivate the player through “[the] lasting psychological impact after subjects return to the physical world.” 

Preliminary Research Insight

• 

  Rise of news-reading on mobile devices

  “Mobile devices have become one of the most common ways Americans get news, outpacing desktop or laptop computers,” according to a Pew Research Center survey.

  Source: Survey conducted May 30-Oct. 23, 2017. PEW RESEARCH CENTER

• 

  The habit of skimming through headlines

  Readers both scan and read deeply rather than reading from page to page in a newspaper. Given how much information people are exposed to in a day, they can only invest in a finite amount of effort and time into news content. People scan through headlines and learn more about a topic they are interested in. By choosing what to understand in-depth, readers are less likely to read articles from different perspectives.

  Data Source: Question: “Everyone has a different way of consuming news. How often do you do each of the following?” Study: “Americans and the News Media,” 2018.

• 

  Increase in social media news consumption versus direct news site

  Reading news directly from publishers website remains the most common news consumption pattern. Yet, users get their news from social media more often than before. Moreover, the number of news consumption from direct sources or social media are nearly identical.

  Source: Survey conducted Feb. 24-March 1, 2016. “How Americans Encounter, Recall and Act Upon Digital News” PEW RESEARCH CENTER

Competitive Analysis

How do existing news platforms deal with perspectives?

I have also analyzed both major news platforms and similar products that tackle the filter bubble effect to see how the current industry approach the issue.

Competitive Insights

Similar in quantity but different in perspective indications

All competitors present multiples reports on a single topic. The primary difference between these two types of news aggregators is how they label the perspectives. For Google News and Apple News, they don’t display the article’s viewpoint. However, readers can filter their preferred news reports via the name of the publishers. Meanwhile, Perspecs and The Perspective explicitly indicate the perspective of each story.

Project Scope

A mobile-first news aggregator

From the insights of people’s reading habits, I decided to design a mobile news aggregator according to the increasing mobile news consumption as well as controlling the experience in a direct news platform.

Disregarding other factors to focus on the core interactions

Focusing on challenging confirmation bias and preventing anchoring effect in three weeks, I have neglected the following factors. However, I would involve these factors in the future interactions of this product.

• Fake news and fact-checking
• Business model
• Nuance of perspectives
• Social media preview

Design Principles

Be mindful of cognitive overload and the anchoring effect

From behavior research and competitive analysis, I synthesized the following principles to guide my design process:

• 

  Balancing between challenging confirmation bias and cognitive overload

  One of the significant factors that cause confirmation bias is the limited cognitive bandwidth to deal with a large amount of information. Therefore, while challenging existing reading behavior, I have to be cautious about managing users’ cognitive load. To iterate the balance throughout the design process, I created the prototypes in higher fidelity with actual content to better grasp the users’ reactions during usability testing.

• 

  Neutral visual presentation to minimize the anchoring effect

  Knowing that the anchoring effect is a significant influence in news consumption, I wanted to make sure that the visual design and the layout of the product minimize as much bias as possible.

Ideation

Exploring ways to present multiple perspectives

To brainstorm ideas to expose opposing perspectives to users, I sketched out different concepts on papers and presented them to other classmates for feedback.

Brainstorming for ways to show multiple perspectives

Design System

Preventing anchoring effect via limited color palette

Meanwhile, I also developed a design system that aims to present perspectives as neutral as possible. Because any color could trigger users’ previous experience, I limited the color palette to only black, gray, and white for this product.

First approach

Dissolving the filter bubble via an “I Disagree“ button

Knowing people read based on confirmation bias, I designed an experience where users can actively “disagree” with what they read on the platform. Even though users might end up reading the articles that support their beliefs, they are exposed to opposing perspectives. Furthermore, their clicks on the “I Disagree.” button would provide data to train the targeted algorithm to dissolve the filter bubble.

Logic of "I Disagree"

Initial User Flow

The homepage of the platform regularly shows the opposing perspectives to the users based on their browsing history. Users can disagree with the given perspective after reading the headlines or the abstract. After disagreeing with the given viewpoint, they can read and select reports that are more aligned with their views as their reasons to object to the original perspective.

Main Features

A neutral headline of a topic

Through whether curation or automation, the headline presents the unbiased view of the subject.

Curated perspective on the topic

Presenting a single perspective, which usually does not align with the user’s views according to their browsing history, a headline of an article is featured for the specific topic. The article will not be labeled with any publishers to affect users’ perceptions at first sight.

To disagree or to read more about the topic

Users can choose to read more about how the perspective covers the subject, or they can disagree with the perspective. Disagreeing with the initial perspective, users will proceed to other news reports that might better align with their opinions.

User Feedback

Intrigued but concerned about practicality

From the user tests with people at the School of Information, most people found this concept intriguing, but they also wondered about the practicality of the product. Some of the feedback talked about this user flow might be “too cognitive load heavy.” Some users concerned that “people might be worried about collecting their opinions.” Most importantly, this product required more testing with the target users when a user said:

“I don’t know how other people will feel, but if I say I disagree I probably want to be over and done with the article.”

Final Results

Multiple scannable perspectives at first sight

To mitigate cognitive overload while presenting a more comprehensive view on a topic, I designed a layout that allows users to grasp different perspectives from the homepage. Similar yet different to the industry-standard presentation, these perspectives are not only shown without the label of their publishers but also randomly ordered to minimize the anchoring effect. Before reading a specific view, users are provided with a neutral headline as well as a section for neutral context extracted from all reports on the topic page.

Logic of the Final Approach

Final User Flow

Final Features

A scannable homepage featuring diverse perspectives

• Scannable perspective at first sight
  Deciding when to present multiple perspectives is essential to the project because 69% of people scan through their daily news headlines more than once per day. Therefore, I designed an experience where users can scan through different perspective right from the homepage before they dive deep into any particular point of view.

Constructing an impartial coverage

• Understanding the context before reading perspectives
  ‍Before reading different news reports in detail, users can read about the common knowledge (assuming there is common knowledge) for better context.
• Arranging perspectives in randomized order
  ‍To reduce the anchoring effect, I listed all reports of the subject in random order under the context section. Even though the listing order might still affect users’ perceptions, showing different coverages on a single page allows users to grasp the general sense of the topic.

Minimizing existing opinions on original sources

• Removing the publisher from the articles
  ‍Through visual design, I wanted to reduce users’ prejudice toward publishers from the opposing perspective. Taking away the colors and logos of the publishers, I assumed that the users would be more focused on the content and be less likely to be influenced by the image of the sources. However, this prototype is designed for accountable publishers and disregarding the possibility of fake news.

Final Feedback

More aligned with current behavior while worrying about cognitive load

Overall, people seemed to like this approach more than the “I Disagree” concept because the navigation is similar to how they currently navigate news platforms. Some people liked the design that forces the users to read everything when labels and publishers are covered. On the other hand, they worried that users might be lazy and would return to reading only their preferred perspectives.

“Visually, this is a fantastic design — seeing how the intention of neutrality comes through in the typography, layout, and color choice was fantastic. The basic interaction works as well — switching up POV ordering is a great idea. I’m not sure you answered the deeper question of whether this approach will have the intended effect — but that’s fine for this project.”

— James Reffell, Instructor & Design Director at Clever

Retrospective

Designing for behavior change is difficult

Mockup for the final product

Next Step

Moving forward in the world of filter bubbles

I will continue to iterate this product at the School of Information in the future. I want to tackle the issue from another point of view with additional practical considerations.