Original title: Concentration Without Representation
Burak Ozturan, David Lazer · September 2026
A tiny group of people posts most of what we see on social media. Scientists wanted to know why past studies gave very different numbers for how concentrated this sharing is, and who these super-sharers actually are. Researchers studied 38 million news shares from 493,237 registered US voters on Twitter over 57 months. They found that how long you watch users changes the concentration number completely. Watching people for one week made the top one percent look like they posted 22 percent of all news. Watching them for 57 months made that top one percent jump to 52 percent. This happens because casual users stop posting over time, not because regular users post more. The study also showed that people who post the most and people who reach the most followers are not identical. Both elite groups are older and more Democratic than average, and do not represent regular voters.
Imagine counting how many times people in a town honk their car horns. If you count for one hour, a few drivers honk a lot and most honk rarely. If you count for a whole year, the same few drivers have honked many times, but now they look even more dominant because you caught more of their rare honks. The drivers themselves did not change—you just watched longer. In this paper, the observation window (how long you track shares) changes how concentrated sharing appears, and the same users stay near the same share rate regardless of window length.
If you think online sharing reflects what most people think or share, this paper shows you are wrong. The top 1% of users account for 22% of shares in a week but 52% over 57 months, meaning concentration is an artifact of how long you watch. Also, the people who share the most are not the same people whose shares reach the most people, and both groups skew older and more Democratic than the voting public. This matters for anyone trying to understand public opinion, political campaigns, or misinformation from social media data.
The abstract does not specify whether the findings generalize beyond US registered voters on Twitter, nor does it address causal mechanisms for why heavy sharers are older and more Democratic. The residual gap against misinformation estimates is attributed to content class but not further characterized. The study does not examine platform algorithms or their role in amplifying certain users.
Ask anything about this paper. Answers come from its abstract, and say so when the abstract doesn't cover your question.
Concentration Without Representation — Burak Ozturan, David Lazer — SocArXiv — doi:10.31235/osf.io/dhq73_v1
Abstract licence: CC BY
Explanation generated by Synthesis from the abstract — it is not the original paper. This is a preprint, so it hasn't necessarily been peer reviewed.