One of the most disruptive aspects of Twitter as a communications medium is that it allows anyone to instantly publish and distribute information to potentially thousands or even millions of people — something that seems inherently democratic, particularly when it is used by protesters and revolutionaries in Egypt and elsewhere. A new study looking at how information is propagated through Twitter, however, shows that a relatively small group of celebrities, organizations and media outlets are still responsible for most of the content on the network.
The study, which was published by a group of researchers at Yahoo (s yhoo) Labs and a colleague from Cornell University (PDF link), looked at the “follower graph” or connections between about 42 million users who were on Twitter as of July 2009. This is the same snapshot of the service that researcher Haewoon Kwak and his team used in research published last year, which came to the conclusion that Twitter is much more like an information-distribution network — that is, a media entity — than it is a social network, a point we have also tried to make here at GigaOM on a number of occasions.
The Yahoo researchers used Twitter lists to find those users who were most followed, and then classified those “elite” users into four groups: media, celebrities, organizations and bloggers. Looking at the flow of information from those individuals to other “ordinary” users of Twitter, the research team concluded that about 50 percent of all the information that gets passed around on the network comes from just 20,000 elite sources — that is, media outlets, organizations, celebrities and bloggers (coincidentally, Time magazine just published a list of top 140 Twitter feeds). As the report put it:
In spite of this fragmentation, it remains the case that 20K elite users, comprising less than 0.05% of the user population, attracts almost 50% of all attention within Twitter. Even if the media has lost attention relative to other elites, information flows have not become egalitarian by any means.
This conclusion isn’t really that surprising. Like any other network, Twitter is subject to the kind of “power law” distribution that others have described in social and information theory, in which those who have a high profile for some reason — because they are celebrities, because they are connected to those who wield some influence or power, or because they have a strong brand identity developed elsewhere — are more likely to have disproportionate amounts of influence. This is a kind of “rich get richer” principle.
The upshot of that for Twitter is that yes, Charlie Sheen and Ashton Kutcher are going to get lots of followers, and sources such as the New York Times (s nyt) and the Washington Post are going to have their links re-tweeted a lot. But the study also points out that Twitter reinforces the concept of a two-step information flow — in other words, content doesn’t just move from “elite” sources directly to thousands or hundreds of thousands of ordinary users. Instead, it relies on intermediaries to pass it along: namely, all of the non-elite but still influential users who re-tweet and link. Media blogger Anna Tarkov described this process as “pollinating” in a recent post, and that’s a good way of looking at it.
So while they may not have completely democratized the way that information is distributed, Twitter and other social media have expanded and amplified it in ground-breaking ways, as shown by the events in Tunisia and Egypt and Libya. Having a majority of the content come from just 20,000 elite users may not sound all that diversified, but it’s arguably a lot broader than the media sphere was before the web and publishing tools like blogs and Twitter came along.
For media outlets like the New York Times, the Yahoo study reinforces how important it is to consider these new kinds of information flows when implementing things like pay walls and subscription plans around content — sources like the Times may carry a lot of influence, but they still rely on thousands of non-elite followers to spread their content around, and they ignore that at their own peril.