世論 What Japan Thinks

Mixi is one of the most popular invite only Social Networking Service (SNS) in Japan, apparently, like Friendster et al. in the USA. They are not my cup of tea (just like I’m not really into all this blogrolling and incestuous linking here), but obviously other people like them. There was a report from a workshop at the Social Infomatics Fair 2005 at Kyoto University containing some interesting statistics on the social webs that had built up.

In February of this year, the number of users stood at around 360,000. Note it seems to be over a million now. The purpose of the research was to try to identify what sort of network of relationships had built up between users.

“My Mix” (in Japanese it’s pronounced closer to “My Mick”) is the slang term used to indicate the number of direct friends someone has registered; this would be roughly analogous to one’s blogroll in the blogging world.

According to a researcher named Kikuo YUTA from the Network Informatics Laboratories, 50.8% of the users have four or less people in their My Mix; 23.6% have but a solitary soul (perhaps just the person that invited them?) as a friend. On the other hand, the average length of a My Mix is just under 21.

For people with five or more names in their My Mix, the numbers stack up as follows:

5 to 11 friends	17.1%
12 to 25 friends	24.3%
26 to 40 friends	15.4%
41 to 87 friends	20.6%
88 to 197 friends	10.1%
198 to 1301 friends	2.9%

This adds up to 90.4% – what happened to the other 9.6%, you may very well ask. Well, here the story write-up is very vague (or I don’t understand it, of course), but what has happened is that the above table represents the share of the total number of links for each of the population sizes. This is explained further in the next paragraph.

A large minority of the My Mixers were very tightly intertwined, making them hubs for all the human relationships. Those with over 41 people in their My Mix list made up only 4.8% of the population, but accounted for 33.6% of the links. On the other hand, those with less than five in their My Mix (50.9%) had just 9.5% of the links.

Hmm, isn’t this just data simply derived from the population distribution, or is there a deeper meaning I am missing?

Kikuo YUTA also said that the high density of links in mixi was very rare – usually structure develops over time, but here it came into being almost right from the beginning. “It is the first time in the world such a unique structure has been found”, he said.

Maybe I’m missing something, but I would think that anything invitation-based would tend to have the sort of tight structure that they are describing? Being in Japanese limits the scope for spreading into disparate groups, I would think. Japanese just love being unique, however. Reading ahead I see more explanation… The rest of the article talked about other human aspects of the network, and whilst very interesting, is getting away from the main purpose of this blog, polls and surveys, so I’ll not translate. Also, the Japanese was getting rather hard!