Random sample of 400,000 Twitter accounts, with metadata

In 2013 I wrote an article called "Tweets Loud and Quiet" (https://www.oreilly.com/ideas/tweets-loud-and-quiet) based on a surprising realization: the median Twitter account has one follower, and the median active account has 61 followers.

In order to write the article, I compiled metadata from a random sample of roughly 400,000 Twitter accounts. The entire sample is now available in this GitHub repo.

Although it's almost four years old now, I get requests for access to the dataset on an almost weekly basis. It's no longer a contemporary snapshot of Twitter usage patterns, but it supports interesting research on social dynamics.