Experiments in rethinking how algorithms rank the content we see

I love this experiment: the Prosocial Ranking Challenge by Mediators Foundation asked independent research teams to design social media content ranking algorithms. In other words, algorithms that decide which content appears on our social media feeds. Normally, such algorithms maximize addictiveness and ad views, but they could also do something else and arguably better.

Instead of maximizing clicks + ad views, these algorithms aim to:

• increase exposure to ideologically diverse perspectives,

• introduce disagreement thoughtfully, in the ways that won’t be instantly rejected by the reader (like showing content you might disagree with but from sources you usually trust),

• and increase how much factual public affairs info people see

They will test the winning submissions experimentally on 15,000 social media users and measure changes in what people consumed and how they experienced the feed, using a custom browser wrapper that lets them swap in these new algorithms.

Feeds don’t have to be addictive echo chambers. We can, in principle, design algorithms for dialogue, depolarization, or civic learning, and it’s important to give them a try.

👉 More here: https://www.mediatorsfoundation.org/current-projects/prosocial-ranking-challenge