What An Algorithm Does
A recommendation algorithm scores pieces of content against what it knows about a user and what the company has chosen to optimize. That goal may be watch time, clicks, purchases, shares, subscriptions, or something else.
The mechanism sounds technical. The effect is cultural. The system decides what appears, what repeats, what spreads, what disappears, and what emotional tone gets rewarded.
Institutions Form People
An institution forms people by creating patterns of attention and behavior. A school forms students through schedules, subjects, authority, peers, and standards. A church forms people through worship, teaching, calendar, and community. A family forms children through love, discipline, habit, and example.
A feed forms people too. It has a calendar, even if it is endless. It has rewards, even if they are likes and views. It has authorities, even if they are influencers and trending accounts. It has a curriculum, even if no one admits writing it.
Share of formative influence on children, 1900-2025
A calibrated model of relative formative influence by channel.
The Missing Accountability
The difference is accountability. A teacher can explain the lesson. A pastor can be challenged. A parent has duties. A local institution has a name and a place.
An algorithmic institution often has none of that legibility. The user sees outputs, not governing principles. Parents see behavior changes, not the exposure pattern. Citizens see polarization, not the ranking incentives that helped create it.
Make Formation Visible
If algorithms are formative institutions, then they should be treated as such. We should ask what they teach, what they reward, what they hide, and what kind of person they tend to produce.
That does not require banning recommendation. It requires moving recommendation out of the shadows and into public, parental, and user control.