One Goal: Maximise Watch Time

YouTube's recommendation algorithm has a single primary objective: maximise the time users spend on the platform. This objective is optimised using hundreds of signals — but it has no built-in concept of whether extended watching is good or bad for a 6-year-old.

How Recommendations Work

When your child finishes a video, YouTube's algorithm selects the next recommendation from a ranked list generated by two neural networks:

The system is very good at predicting what will keep a specific user watching — which is precisely the problem when that user is a child.

The Rabbit Hole Effect

Researchers at Mozilla and the Center for Humane Technology documented what they call the "rabbit hole" — a recommendation sequence where each video is slightly more extreme or emotionally intense than the last, because more intense content drives slightly higher engagement.

A child who starts on a Minecraft tutorial can end up watching conspiracy content or age-inappropriate videos within 5–6 autoplay steps — not because anyone intended it, but because the algorithm is following its engagement signal.

Why Children Are Particularly Vulnerable

Children under 12 have underdeveloped prefrontal cortex function — the area responsible for impulse control and future-thinking. They are significantly less able to choose to stop watching than adults, which makes them disproportionately susceptible to the design patterns YouTube uses to maximise engagement (autoplay, infinite scroll, notification hooks).

What You Can Do

Start monitoring your child's YouTube sessions with Kidoio →

Understand what your child is really watching

Kidoio analyses every YouTube video for emotional tone, age-appropriateness, and developmental impact — grounded in evidence-based child psychology.

Join waitlist →