A friend list is a biased sample
Imagine writing every person in a group on a card, then adding one card for each friendship they have. A person with many friends appears on many cards; someone with few appears only a few times. If you pick a card at random, you are more likely to land on the well-connected person.
That is the mechanism behind the friendship paradox: when people compare themselves with their friends, those friends often seem to have more friends on average. The comparison does not sample people evenly. It samples through friendships, giving socially connected people more chances to be selected.
Why the tilt grows in varied networks
If everyone had exactly the same number of friends, the effect would disappear. But real social networks include both people with small circles and people connected to many others. The more uneven that spread is, the stronger the sampling tilt can be.
In a simplified network, the average number of friends a randomly chosen person has is the network’s average degree. But a person reached by following a randomly chosen friendship tends to have a higher degree: each of their friendships is another route by which they could be reached. Mathematically, the difference depends on how much friend counts vary across the network.
A comparison, not a judgment
This is a feature of how the sample is chosen, not evidence that any one person is unpopular. It can also affect group-level observations: a survey conducted by asking people to name their friends may overrepresent highly connected people, even if everyone answers honestly.
So when a friend seems to know everyone, remember the route by which you encountered them. Their broad circle may be exactly what made them more likely to appear in your comparison.
