The Samovar

Freedom and inequality (and blogs)
November 8, 2008, 5:35 am
Filed under: Politics | Tags: , , , , ,

Readership of blogs follows a power law distribution, that means roughly that if you rank all the blogs in order of readership, then blog N will have a readership proportional to 1/N (actually it’s Na but a is close to -1). This means that almost all blog traffic goes to a small number of blogs, and the vast majority of blogs will have almost no readers (like this one, but that’s fine by me). The same rule appears to apply to many situations: readership of web pages generally, word frequencies in the English language, digital song sales, etc. There are some mathematical arguments that if you assume people are free to choose from a large number of options (of any sort), but that other peoples’ choices will affect them, even if only marginally, then power law distributions follow. In other words, this is a natural and unavoidable situation.

Well, I haven’t thought terribly deeply about this, but are there consequences for democracy? Presumably change happens slowly in these power law distributed readerships because they’re based on others’ preferences. Does this put a fundamental limit on the rate of change of opinions? Does it make propaganda inevitable? I don’t know. Anyway, for those of us who believe in equality, I think that there is a challenge to be met in these observations.



Isn’t the party system in most countries a direct and inevitable result of power law distributions? Of course we sometimes forget that changes in the criteria for power laws (even perhaps very small changes on occasion), lead to different power laws.

Comment by Colm

That’s an interesting question. I’ve always thought that the two party system was the result of the FPTP electoral system. But in a proportional representation system you might expect to see a power law distribution. Do you know if there’s any data on this? It’s an interesting possibility.

Comment by Dan | thesamovar

Comments are closed.

%d bloggers like this: