As part of the Jeux du Commerce (JDC) 2027, TCCT is running the “Let’s Talk to Connect”
campaign for the benefit of Le Centre pour l’intelligence émotionnelle en ligne (Le CIEL), whose
mission is to raise young people’s awareness of digital issues and promote healthy, balanced
screen use. Le CIEL's mission is to raise young people's awareness of digital issues and
promote healthy, balanced screen use.
A platform has far more content than anyone can browse. A recommendation algorithm sorts it.
According to MediaSmarts, it filters out what a user is unlikely to care about and puts forward
what they are likely to enjoy so they don't feel overwhelmed and leave the platform.
To do this, the system relies on usage data: the content a person likes, comments on, or shares,
how long each visit lasts, and how soon they come back to the app. It also compares their
behavior with that of users who have similar habits, then suggests what those other users
enjoyed.
Platforms explain part of how this works. According to Instagram, the home feed mixes posts
from accounts a person follows, recommended content, and ads. It is ranked using the posts a
person has liked, shared, saved, or commented on, how popular each post is, and past
interactions with the poster. The Explore section shows only recommendations from accounts
the user doesn't follow, and most Reels come from those accounts too. YouTube says its
recommendations rely on watch and search history, subscriptions, likes and dislikes, "Not
interested" feedback, and satisfaction surveys.
Rabbit holes are a common effect of recommendation algorithms. The term describes a chain of
suggested videos or posts, one after another, that keeps a person on the platform longer than
planned. In a MediaSmarts study of young people aged 13 to 17, participants talked about
rabbit holes and mindless scrolling. A short break during a study session can easily stretch well
past what was planned.
It is also worth keeping in mind that a feed shows only part of other people's lives. People
mostly post what they choose to share, such as a job offer, a finished project, or a trip. This
selection bias means that comparing it with one's own path gives an incomplete picture.
A feed built from what a person has already watched tends to repeat the same content. After a
single video on a topic, the recommendations can suggest more on the same theme, and they
can be hard to get rid of once the interest has passed. It is also impossible to know what the
feed leaves out.
This also affects how people get their news. A feed reflects, in part, the habits of the person
using it, so what it shows of current events is never a full picture. The young people in the
MediaSmarts study were skeptical that algorithms present accurate information, especially when
searching for a school project. For university work, it can help to rely on sources chosen on
purpose rather than on whatever is recommended.
Several tools can help steer what shows up in a feed. The MediaSmarts tip sheet How to Train
Your Algorithm covers many of them, and platforms document their own.
● Send clear signals. MediaSmarts recommends using the tools that tell a platform a post
isn't of interest (Not interested, Dislike, Hide). Instagram also lets users snooze suggested
posts for 30 days, and YouTube lets them exclude a channel from recommendations.
● Start fresh. The tip sheet suggests resetting an app's preferences from time to time and
gives steps for TikTok, Instagram, and YouTube.
● Diversify sources. It also recommends exploring other topics and following creators from
different backgrounds.
● View followed accounts only. Instagram has a view limited to followed accounts,
showing their posts from the last 30 days in chronological order.
● See why something shows up. Instagram explains why a post or ad is being shown,
and YouTube lets users turn off and delete their watch history.
Options vary from one platform to another and change over time. Each service's help page has
the current steps.
● Le CIEL: an organization founded in 2019 whose prevention workshops cover algorithms
and digital self-defense, among other topics.
● MediaSmarts, Recommendation algorithms: explains how recommendation systems
work and two ways to respond, choosing sources and sending signals to the algorithm.
● MediaSmarts, How to Train Your Algorithm: a tip sheet of practical steps for shaping
recommendations, including steps for TikTok, Instagram, and YouTube.
● University of Ottawa, health and wellness: health and wellness services for students.
The Student Wellness line is 613-562-5777.