Whether about blogs, songs or news stories, when people must make decisions among many different alternatives, those making them later are often greatly influenced by those making them earlier.
That creates a cascading effect that results in a "popular-get-more-popular" sort of phenomenon along many different dimensions.
This is one reason so-called power law distributions are so common in social situations.
In particular, such power law dynamics partially explain the fact that a few blogs are visited by millions, while the vast majority are lucky if they attract the bloggers' close friends and relatives.
An informal explanation involves the huge and rapidly increasing number of blogs.
Given all these choices, a surfer will be more likely to visit sites that are already popular just as a diner walking in an area with many restaurants will be more likely to choose one that has already attracted other diners than one that is empty.
The blogs that attract the first visitors and links are marginally and then overwhelmingly more likely to attract later ones and hence ultimately more likely to become one of the few must-visit sites.
Being a pioneer helps in the blogosphere as elsewhere.
As Clive Thompson observed in a recent issue of New York magazine, most of the critical nodes in the blogosphere have been around for ages in Internet time.
Boing Boing, a tech blog online for five years, is the most-popular site according to Technorati, which ranks and catalogs sites.
Among gossip sites, Gawker is the leader, while Daily Kos and Instapundit are the most-popular liberal and conservative political sites, respectively. By contrast the vast majority of tech, gossip, political and other blogs are empty restaurants.
The same sort of disproportion of popularity characterizes Web sites generally as well as many other disparate phenomena. (For a slightly more mathematical exposition of power laws, see http://www.guardian.co.uk/life/lastword/story/0,,1265949,00.html.)
Another recent instance of the "monkey-see, monkey-do" power laws is a February study in the journal Science by sociologist Matthew Salganik and colleagues.
The study asked 14,000 teenagers to rate a number of songs on a 1 to 5 scale. The teens were randomly assigned to either an independent group or to a group subjected to social influence.
Both groups were exposed to unknown songs by unknown bands.
The independent group made its choices of which ones to listen to and download based on the songs alone, while the socially influenced group could also see how many times each song they listened to had been downloaded by others.
The same songs tended to be popular and unpopular in each group, but the popular songs in the socially influenced group tended to be much more popular.
Presumably the independent group's choices led to something like a normal distribution, whereas the socially influenced group's choices led to a power law distribution, a few very popular hits and the rest of the songs also-rans.
What constitutes a popular news story is, of course, also affected by these considerations. (A nice combination of news story/blog is the Web site reddit.com.)