🧪 This platform is in early beta. Features may change and you might encounter bugs. We appreciate your patience!
friendship_paradox
The Friendship Paradox states that, on average, your friends have more friends than you do. This occurs because people with many connections appear in disproportionately many friend lists, skewing the average upward. It is a mathematical property of networks with unequal degree distributions, not a matter of perception.
On social media, most users find that their followers have more followers than they do. A user with 200 followers checks their friends' follower counts and finds the average is 800 — not because the user is unpopular, but because high-follower accounts appear in many people's friend lists.
At a new job, an employee feels socially behind because every colleague he meets seems to know more people in the office than he does. In reality, he is simply more likely to be introduced to well-connected employees first — the quiet workers with few work friends are statistically less likely to cross his path early on.
A college freshman attends her first few parties and notices that everyone around her seems to have a larger social circle and busier schedule than she does, fueling self-doubt. What she doesn't realize is that popular, outgoing students are overrepresented at social events, making the average attendee appear far more connected than the typical student actually is.
Binary (yes/no) questions an LLM must answer to identify this aspect:
Does the claim involve comparing an individual's connections to the average connections of their contacts?
Type: binaryAre highly connected nodes disproportionately represented in the sample due to their many links?
Type: binaryIs the network degree distribution skewed rather than uniform?
Type: binaryDoes the argument treat a biased network sample as representative of the overall population?
Type: binaryThe Friendship Paradox states that, on average, your friends have more friends than you do. This occurs because people with many connections appear in disproportionately many friend lists, skewing the average upward. It is a mathematical property of networks with unequal degree distributions, not a matter of perception.
High-degree nodes (people with many connections) are over-sampled when you look at friends-of-friends, because they appear in more friend lists. This sampling bias shifts the average upward, making most individuals appear below average compared to their own contacts.
Recognize that friend-of-friend samples are inherently biased toward highly connected individuals. Compare against the true population median rather than the mean of your contacts. Understand that this is a structural property of networks, not evidence of personal deficiency.
Social media platforms exploit this paradox — users feel they have fewer friends, followers, or engagement than their peers, driving increased usage. It also appears in epidemiology, where vaccinating friends of random individuals is more effective than vaccinating random individuals themselves.
A spurious correlation appears between two independent variables when the sample is conditioned on a common effect (collider). For example, among hospitalized patients, two unrelated diseases may appear negatively correlated because admission is the collider.
Ignoring general statistical base rates in favor of specific individual-case info.
Random observation of a process is more likely to catch long-duration events than short ones.
Use these tools to detect, analyze, or train this aspect.