Focus on Technology
1:02 pm
Wed March 5, 2014

How millions of tweets boil down to six conversations

Many people struggle to make sense of Twitter. The constant stream of information containing up to 140 characters can often overwhelm. But believe it or not all the tweets in the world appear to fit into just six patterns.

The Pew Research Center and the Social Media Research Foundation studied thousands of tweets over a four-year period and came up with these six different conversational archetypes.

  1. Divided-Polarized Crowds who do not interact with groups that disagree with them. (politics)
  2. Unified-Tight Crowds where participants strongly connect. (hobbies, professional topics)
  3. Fragmented-Brand Clusters formed around products and celebrities. (public events, popular subjects)
  4. Clusters-Community Clusters created around global news events and popular topics.
  5. In-Hub & Spoke- News media outlets whose followers re-tweet them.
  6. Out-Hub & Spoke-Companies and government agencies responding to complaints and requests.

Director of the Social Media Research Foundation Marc Smith, a sociologist, says social media leave behind a lot of data about who talks to who and when, and with this study he brings a new set of tools to "take pictures of this new landscape."

How did the researchers come up with these six patterns?

  • Using software they designed (now free and available to the public), they looked at pairs of people.
  • Every time somebody replied they considered that an event.
  • They collected messages every day for four years on a variety of different topics.
  • Researchers made network maps out of the more than a million tweets they collected.
  • After a while the maps started to resemble one another.
  • Every time somebody decided who they would reply to and who they would mention, a pattern formed.

Here is a link to the WVXU twitter pattern (In-Hub & Spoke) generated March 3rd.