The short tail: How news stories move through Digg and Twitter

In a study how news stories become popular through two social networking sites, researchers compare trending stories to viewing a contagious disease spreading through the general population. 

Written in 2010, the study [pdf] by Kristina Lerman and Rumi Ghosh at the USC Information Sciences Institute, investigated Twitter and Digg, which both allows people to follow others, receiving their updates on a moving status bar. In Twitter, users can post 'updates' on anything (up to 140 characters), but the social network has become a great tool for the curation of news stories.

Digg is more centered on news stories, where users can vote for (or digg) stories and allow others users to view them. 

Like other media on the World Wide Web, like music, videos and wikipedia pages, the researchers argue news stories are heavily long-tailed: A very few items become extremely popular, a few a vaguely popular leaving the vast majority to go largely unnoticed. To me, this is the study's golden kernel: Proof that these social networks are important tools to forward popular media stories, but less perfect for illuminating lesser-known stories. 

The mean number of Digg stories runs about 50, while a few get 400 votes and a very small minority can reach as many as 1500. Compared to Twitter, stories move through Digg much faster -- largely because the network of friends is so large. Also, once a story receives many Diggs, it moves to the front page, allowing unsubscribed participants to vote it up. Only fast rising stories can remain on the front page because Digg received (in 2010) about 20,000 stories each day. 

Twitter, a much larger network than Digg, nevertheless allows stories to expand -- but not at the same speeds as Digg. Twitter does not have a like function, so the researchers counted whenever a story was retweeted by another user to be seen by all of his or her users. This adds the potential contagion to spread deeper into the network. The researchers also found that stories reach this level on Twitter, but do not remain so viral for long.

From the researchers:  

In spite of the similarities, there are quantitative differences in the structure and function of social networks on Digg and Twitter. Digg networks are dense and highly interconnected. A story posted on Digg initially spreads quickly through the network, with users who are following the submitter also likely to follow other voters. After the story is promoted to Digg's front page, however, it is exposed to a large number of unconnected users. The spread of the story on the network slows significantly, though the story may still generate a large response from Digg audience. The Twitter social network is less dense than Digg's, and stories spread through the network slower than Digg stories do initially, but they continue spreading at this rate as the story ages and generally penetrate the network father than Digg stories.

The study: 

Information Contagion: an Empirical Study of the Spread of News on Digg and
Twitter Social Networks

Kristina Lerman and Rumi Ghosh


How to win subscribers and advertisers

Websites earn revenue three basic ways: charge a subscription, charge advertising or do both. 

Higher subscription prices offer fewer viewers. However, those viewers may be the most attractive for advertisers because they have the income to pay for the high subscription. But they also may have the lowest tolerance for adverts. 

That's from a seminal 2003 paper [pdf] on the potential impacts of various advertising and subscription models for electronic media. 

The authors investigate the efficacy of four subscription and advertising strategies: 

  1. Limited access, which generally drive away low income (or minimally interested users) because they refuse to pay the subscription price. 
  2. Free access, which generally drive away high income users because advertisements are too common. 
  3. Pooling, where all customers accept the available advert price 
  4. Separating, where customers are segmented by self selecting their preferred subscription price.

The authors argue that a pure adverstiser-supported or pure pay-per-view models will only be successful under certain parameters: Mostly if the site only appeals to high- and low-income audience segments. 

Media sites will most likely pick up advertisers with higher end products -- those that produce a higher margin. But if these advertisers are fighting for the lower-income (or low interest) segment, there will be a lower probability of successful advertising. Advertising is a consumer annoyance, the authors write, but it can be used to market different products to different audience segments. "if different consumer segments are willing
to pay different prices for the opportunity to watch the program at different levels of advertising," they write. 
Thus, the authors lean heavily toward separating, allowing the site to segment the audience. For one, providing users with a choice to opt in or opt out of different amounts of advertising is easy to do on electronic sites. Secondly, advertisers will pay a higher price when they can better target different sets of users. 

They conclude by looking toward a future where content and advertising is married:  

It is not likely that the separating pricing strategies that we discuss here will become more pervasive with time since contemporary electronic media make it possible to implement it much more efficiently than was possible in the past. The implications are higher profits for media providers, more choices for customers, and more targeted advertising for advertisers. 

The paper: 

Advertising versus pay-per-view in electronic media
Intern. J. of Research in Marketing
20 (2003) 13 – 30
Ashutosh Prasada, Vijay Mahajanb and Bart Bronnenberg