Log Log

I'm into AnalyzingWiki?. I just don't look a lot at the contents, but at the activities.

It seems the huge majority of the people who make an edit just make one edit. Finally, I've got all into a block_graphics. When I make both the x_axis and the y_axis use a logarithmic scale, I just see a straight line, going down, with a small hump at the end - something like this:

 |\ 
 | \ 
 |  \  y-axis = log(number of people)
 |   \  x-axis = log(number of pages)
 |    \ 
 |     \/\

(see also end of page, if luck is on)

So the graphics show how many people edited how many pages.

I think this graphics show a general rule. For instance in all kind of sports.

 Most people aren't involved at all.
 A lot of people are involved a little.
 Some people are involved a lot. 
 Only very few people are involved a real lot

I figure when you study how many people are involved in a sport, and you measure the time they spent in this sport and put these two values in a loglog graphic, it will generate a straight line.

This is contrary to a normal distribution, well maybe someone can explain this some day later. Why would you expect a normal distribution in the first place?

About the lump: The most active people also do a lot of small edits. The people in this lump are the real community. The lump causes the self-repairing activity.

Q How do you know that "the lump causes the self-repairing activity"? How do you know that "the people in this lump are the real community"?

A Well I do not "know" this, I think this, so this is my interpretation. If I look at the contents of the data it also seems this way. The data is all ChangesIn<Months> and RecentEdits, so you can check it out yourself too.

Another interpretation: Still, this huge mass of people who just pass by and aren't involved are needed, to generate a small community. So this huge mass of people is the backbone. If you look around to other WikiWikis you will find a lot with just occasional edits, and those WikiWikis are mere guestbooks. Those WikiWikis are not really alive.

"LogLog"

http://www.leescafe.freeservers.com/gif/loglog.gif

Note: remote embedding of this image is disallowed by freeservers.com. Copy and paste the URL into your browser.

A straight line on a log-log plot means log y = A log x + B, which means y = C x^k. In this case, it looke as if k is close to -1, which would suggest an instance of ZipfsLaw.(http://mathworld.wolfram.com/ZipfsLaw.html )

Well, the line hits the X_axis at about log(400) and it hits the Y_axis at about log(2000). ( about = a citation from my memory).