Most social networks exhibit a power law in some of their aspects; so it's not surprising that Flickr would also show power law behavior somewhere. In fact, Acme looked a power laws in tag use
, and Ambiguous made a reflection on a Pareto law for the visits on its Flickr stream
: a few pictures gather most visits, while the bulk of the pictures just gather a few visits.
So I started to look a it a bit systematically: first at my own stream
and then at few more
. Now, in English, I have data for 5 streams:
With data for 5 streams, a few patterns start to show. First, look at the line in the bottom: it's a stream with few visits, and few pictures; it does not even have 200, as the rest. It's a power law with exponent = 0.46, but there's crearly an exponential cutoff at the tail, which should be expected.
The other streams do not follow that exponential cutoff, buth they fall into two groups: exponent ~ 0.80 (red and green), and exponent ~ 0.45 (black and blue). Curiously enough, the two latest streams were created one day apart from each other; the other two a bit after (October 2004 and January 2005). I would say there's a certain correlation between age and power law exponent, but it's not clear how it changes.
For the time being, I'm doing this by hand; people email me a zip file with their 10 "Most popular-views" pages (and I'd be happy to get yours), but I'm looking at Fragant Disregard
, and I think it could be done automatically through the Flickr API... I'll have to check that.