Ms. Khan’s class is learning about social media algorithms. These are the systems social media companies use to select the material most likely to engage a user. Social media companies track what a user has previously engaged with and what material is most engaging to other, similar users. An algorithm uses that data to select the material to show the user next.

“I wish algorithms would stop recommending depressing videos that aren’t compatible with relaxing,” Finn complained. “I didn’t even ask for them. I just want to see videos about dancing and animals.”

“It’s not just about what you engage with,” Arya commented. “It’s also about what other users like you engage with. The algorithm wants to show you whatever is likely to keep you engaged, even if it makes you feel bad, even if it’s not true!”

“Okay, but what makes the algorithm think I would engage with this content,” Finn wondered. “How does it decide?”

Ms. Khan explained, “Social media companies don’t share their algorithms with the public, but we do know some things about how they work. Maybe you two should do some research and share the information you find with the class.”

Finn and Arya found several reliable sources of information on this topic. They created an example to demonstrate why an algorithm might choose a particular post more often for users’ feeds.

An algorithm is determining what to show Jonah next. 

• Post A: Short video showing how to do a popular dance move.

• Post B: Short video saying the mayor has decided to stop all school cafeterias from serving meat. [This is not true.]

A social media algorithm uses data that show people in Jonah’s same area engage with Post B five times as often as Post A by watching, commenting, or sharing it. 

Therefore, the algorithm puts Post B into Jonah’s feed, even though Jonah enjoys dance videos.

Imagine that Jonah watched Post B and left a comment saying the story isn’t true and that the source of the video is not trustworthy. Would the algorithm show more or fewer posts like Post B after?

Why might a social media algorithm prefer content a user engages with over content a user enjoys?

Discussion Question:

Algorithms expose users to content that is engaging to other people who are like them but not to content that is interesting to people who are different from them. Is that a good thing or a bad thing? Why?