It's Possible to Reduce Polarization without Reducing Engagement
The results of the Prosocial Ranking Challenge
The Prosocial Ranking Challenge
Updates on the Prosocial Ranking Challenge, a contest to build better social media algorithms with $60,000 in prizes. From the Center for Human-Compatible AI, UC Berkeley
Live Last read · last published · next check
Latest posts
Experiment Finished, Deep in Analysis
Results are promising but we aren't ready to release them
Experiment Launched! Project Midpoint Update
Contest winners, recruitment results, midline survey, and more
Announcing the Finalists of the Prosocial Ranking Challenge
Injecting credible news, bridging-based ranking, downranking toxicity, adding diversity, and more
Notes From Today's Info Session
You had great questions
Updated: What We Will Measure (DVs)
Some changes, much more detail.
Deadline Extended! Submissions due April 15
Everything moved back two weeks
Sample data is here!
Sample real data sets to generate test JSON for your ranker
What Might Work: A Guide to Previous Research
Try short-term outcomes, bridging-based ranking, and adding content
Updates: Add Your Own DVs, Info Session FAQs
Also updated DVs. Lots of new information for Challenge teams.