Too many topics, too little time » power law

Ab-normal Distribution

Jeremy — Thu, 02 Oct 2008 14:34:45 +0000

The Universe of Discourse : The Lake Wobegon Distribution is a tight little explanation using baseball that also maps well into the way i think of most of the blather about power laws and related seeming web 2.0 statistic, which is that if you really analyze what is going, you find out that really it isn’t quite what you think. My critique of the power law people is to generally say, if you arrive at a power law, you are likely asking the wrong questions and that if you asked the same question in a different population, you’d likely find the same thing which would in the end, show your power law as insignificant. This article makes a different argument about how most people think normal curves are normal, but it shows pretty clearly that most places we’d expect a normal curve we find abnormal variation especially outside of the center bell.

power laws again

Jeremy — Mon, 26 Jan 2004 05:34:51 +0000

several people are coming here for the powerlaws abstract. it is here
http://www.cddc.vt.edu/jeremy/blog/2003/08/22.html

paper idea: it’s not the power law: how statistical generalizations are confusing social modeling on the internet

Jeremy — Fri, 22 Aug 2003 16:02:31 +0000

In this paper, i will argue why what many people argue are power law models of social networks on the internet are usually not really that at all. By critiquing the assumptions built into the data gathering and measurement, combined with the theoretical modeled used, it is easy to see that through considering time in different ways and by fragmenting the data set, that there are much more complex and interesting phenomena involved in these power law situations that is being masked and hidden by the generalization. In the end, separating out the individual phenomena that seem to map to power law situations usually illustrate the the phenomena as a whole is not related to the power law, and that by using the power law to describe the phenomena, we end up losing much of the unique understandings that could make or break the application of technologies in this arena for a variety of purposes.