The Comparison Trap Has an Algorithm Now

In 1954, the psychologist Leon Festinger proposed something that seems almost too obvious to be a theory: people evaluate themselves by comparing themselves to others. He called it social comparison theory, and he wasn't thinking about phones — he was thinking about neighbours, colleagues, the people in the next town over. Comparison, in his framing, was a tool. In the absence of an objective yardstick for how well you're doing, you look sideways.

The tool still works the same way. What's changed is everything around it.

The sample size problem

Festinger's comparison targets were local and finite: the people you actually knew, doing things you could actually see. Your comparison set might have been a few dozen people, most of whom you had reasonably accurate information about, including their bad days.

Now the comparison set is, functionally, everyone. Not everyone you know — everyone an algorithm has decided is worth showing you, drawn from a pool of people optimising their most photogenic, most impressive, most enviable moments for an audience of strangers. You are no longer comparing yourself to your street. You're comparing yourself to the best three seconds of several thousand people's week.

You're not comparing yourself to other people's lives. You're comparing yourself to other people's highlight reels, selected by a system that rewards whichever highlight performs best.

This is the part that gets missed in the "everyone only posts their best moments" explanation, which by now most people have heard and quietly discounted because knowing it doesn't seem to help. The issue isn't just that people curate. It's that a recommendation system then re-curates on top of that curation, surfacing the most engaging fraction of an already-curated pool. You're not seeing a representative sample of even the polished version of other people's lives. You're seeing the outliers of the outliers.

Why upward comparison hooks harder than downward comparison

Social comparison researchers distinguish between upward comparison (looking at people who seem better off) and downward comparison (looking at people who seem worse off). Downward comparison tends to make people feel better, at least briefly. Upward comparison is more complicated — it can be motivating in small doses and corrosive in large ones, and the difference often comes down to how attainable the standard feels and how much control you feel you have over closing the gap.

Feeds are structurally biased toward upward comparison content, because upward comparison content — the new house, the relationship milestone, the body transformation, the promotion — generates more engagement than the ordinary Tuesday most people are actually having. An algorithm trained to maximise engagement will, without anyone deciding this explicitly, learn to serve you more of the content that makes comparison sharpest.

1. Open the app
2. See curated, algorithmically selected upward comparisons
3. Feel a small gap between self and standard
4. Engage longer, seeking resolution or reassurance
5. Algorithm reads engagement as interest, serves more of the same

What the evidence actually says

This is a place to be careful, because the research is more mixed than either the "social media is destroying self-esteem" camp or the "it's basically fine" camp likes to admit. Jean Twenge has argued for a meaningful link between social media use and declines in adolescent wellbeing, particularly among girls. Amy Orben and Andrew Przybylski, using the same large datasets, have found effect sizes closer to those of eating potatoes — real, but small, and dwarfed by other factors like sleep and family relationships. Both are looking at genuine data. They disagree about how much a small average effect matters and what's driving it.

What this does not mean

It does not mean comparison is new, or that people were psychologically fine before social media and are now uniformly worse off. Comparison anxiety existed around magazines, around neighbourhoods, around school. It does not mean everyone who uses social media experiences heightened comparison distress — plenty of people scroll happily and feel largely unaffected, and it's worth being honest about that rather than assuming universal harm. What seems most defensible is narrower: the mechanism has more reach, more selection pressure toward upward comparisons, and less friction than it used to, so for people already prone to comparison, it has more surface area to work with.

In real life

The comparison usually doesn't announce itself as comparison. It shows up as a mood that arrived from nowhere — a flatness after fifteen minutes of scrolling, a specific irritation at your own kitchen, your own body, your own Tuesday. Most people don't consciously think "I am comparing myself to this person." They just notice they feel worse and assume it's unrelated to what they were just looking at.

Noticing the link is most of the work. Comparison that stays unconscious runs on autopilot; comparison you can name is comparison you can question.

Try this

A thought to keep

Festinger thought comparison was how people worked out where they stood in the absence of better information. The uncomfortable modern twist is that the information is now abundant, constant, and selected for maximum contrast — which means the yardstick itself has stopped being neutral. The old advice was "stop comparing yourself to others." The more honest advice might be: notice who is choosing what you're comparing yourself to, and ask whether it's you.