Power Law distributions emerge in systems when a large number of people freely choose between a large number of options. The basic idea, often referred to as the "80/20 rule", is that a very small number of options are chosen by the large majority of choosers. The Nth most popular option will be chosen 1/Nth as many times as the most popular. For example, if the most popular blog receives 1000 views, the second most popular will receive 1/2 that, or 500, the 3rd will receive 1/3 or 333, and so on. Following this pattern, the 10th most popular will receive 100 views, the 100th will receive 10, and the 1000th most popular receives 1 view. Graphically this looks like a very steep slope representing only the most popular, followed by a very long tail of the way below-average. This creates a large, non-intuitive imbalance in a system that is believed to provide equal opportunities, such as blogging, or the internet as a whole, for that matter. In reality, because the huge success of the very top blogs weights the average in their favor, more than 2/3's of blogs are below average in the amount of traffic it receives. The pattern emerges not as a result of a huge imbalance in the quality of available choices, but because people tend to follow each other's choices. Later users are given a field of options that have already been chosen by those that came before them, and popularity greatly affects their perception of these options. When there is this tendency toward agreement in large, open systems, power laws emerge, and once they emerge they tend towards homeostais.

As new readers enter the blogosphere they increase the readership of the top blogs, and as new blogs enter, they increase the "long tail" of blogs with (way) below average readership. As the blogosphere exands, the power law distribution will change the nature of the blogs at the two extremes of the distribution so that they hardly resemble each other anymore. Eventually, due their popularity, the top blogs stop being able to respond to their readers, and become more like traditional, one-way broadcasters. The blogs at the lower end of the distribution, with only a few readers, will be much more conversational. Only the blogs in the middle of the distribution will still resemble what Shirky terms "Blog Classic": blogs written by one or a few people with a moderate audience that enjoy an engaged relationship with the writers. Blog Classic's, however, will be greatly outnumbered by the long tail of small conversational blogs, and their audience will be dwarfed by the mega-blogs. Shirky concludes by remarking that this is a natural evolution of open systems and not by design. Weblogs in the early days were egalitarian not because of their intrinsic nature, but because there weren't enough of them yet, nor enough readers, for the inequality of a power law distribution to emerge.

Along with this article, www.shirky.com has many articles on social media, systems, economics, and open source, and is worth a look.