Year
2021
Client
Schibsted · AHO
Recognition
🏆 Visuelt Nomination (2021)
Scope
Interaction design
Anti-Filter Bubble
Designing transparency into algorithmic news
Anti-Filter Bubble was a collaborative project with VG, one of Norway’s biggest news outlets. VG was testing algorithms to personalize its front page, but they also wanted to make sure readers stayed informed, not trapped in their own bubbles. The goal was to find a balance between transparency and trust in how news is shown. I designed tools that made bias visible, broadened what readers were exposed to, and encouraged them to see different sides of a story.
Three proposals
1. VG Explainer
Explainers often drown among other articles, making it hard for readers to catch up on complex stories. However, when the explainers are presented contextually on the front page, it can be a subtle way of prompting users who to visit a dedicated explainer page.
Short summaries in text and video make it easy to grasp the story, while related articles and a dedicated “Explainer” tab encourage ongoing engagement with important topics.
2. Absorbable Headlines
My second proposal is Absorbable Headlines, short, engaging “fun facts” about topics users usually skip, shown within their personalized feed.
Young users often only read headlines and rarely open full articles, especially on topics they find boring. But since many headlines are more clickbait than informative, they don’t actually learn much. This idea puts key information directly on the main screen, disguised as quick, easy-to-read facts.
For example, a user who mostly reads about baking or celebrities might occasionally see a short, fact-style post about politics or the environment, helping them absorb important news effortlessly while scrolling.

3. Competency Tracker
My latest proposal is the Competency Tracker, a feature that shows each user’s skill level (or interest level) in different topics to motivate them through gamification.
I tested several ways to display these levels, but testers found it discouraging when their results were compared to average users. Most people preferred a simpler system with three levels: Rookie, Competent, and Expert. These levels are automatically determined based on factors like reading history and quiz results, and are displayed next to each topic.

