[ This is a comment I made in the running response to this post on Azeem Azhar's Weblog. Azeem, formerly a tech writer at The Guardian, has since worked as a high-tech consultant and startup CEO. ]

Bernardo Huberman thinks Web sites drive to power law distributions in part because of math. For whatever reason (slashdotting, celebrity), a site becomes popular. Since the Web is growing, at any time interval, hits increase for all 'live' Web sites above some threshold of activity.

Assume that the growth is the same per cent for all sites at each interval: then the leading sites will begin to pull away since they continually increase by a larger number of page reads or links relative to others, since their growth is a percentage of a larger base.

If growth is unequal (which I think is the actual case), and more popular sites grow faster, then the effect is amplified further. Positive feedback loop, predictable outcome. Since there are finite Web surfers during any interval, the big ones get bigger at the expense of the little ones.

Wealth works the same way: people who have only a little bit more than their neighbors can save and make investments and earn interest, where the poor neighbor can only just afford to pay the bills. Small sums, invested, compounded over time, can become quite large. Larger sums get even bigger even faster.

So, it's luck and small differences in initial conditions... I would wager there is a white paper in Complexity Theory circles somewhere that describes a similar mechanism. We humans of course, stand ever ready to attach racial or cultural signifigance to this, when it's probably just luck and math (read Jared Diamond's Guns, Germs and Steel about the roots of inequality, if you haven't already).

Interestingly, though, there can be lots of domains where this operates. A really popular Weblog doesn't get anything like the hits that a popular portal does: yet it's enough to make a few celebs... Dave Winer and Doc Searls come to mind.

So if you can't win on one battlefield, start a new one... BTW I will shortly start a 'Robot Noise' Web site where we'll try a few experiments to test some of the theories proposed here and elsewhere...
Comments 5:28:02 PM    

Chris Gulker's view from Silicon Valley - in words and pictures