Twitter exhibits the “friendship paradox”.

a) Expected degree distributions for a 1.25% random sample of the Twitter network (black line), friends of this randomly chosen group (red line), and the same friends group with duplicates removed (blue line); b) Larger samples of friends show a smaller difference in degree distribution from the overall network (black = overall network, green = 25% sample, blue = 7.5% sample, red = 1.25%); c) and d) Respectively, In-degree (follower) and out-degree (followee) distribution of a random sample of 500,000 users, 1.25% of Twitters users (the “control” group, black line) and the theoretical (red line) and observed (blue line) in-degree and out-degree distributions of their friends (the “sensor” group) with duplicates from the friends group removed.