Connectivity via Instant Messaging

The Washington Post today has an article by Peter Whoriskey based largely on a recent study done by two scientists at Microsoft Research. Here are key excerpts:

With records of 30 billion electronic conversations among 180 million people from around the world, researchers have concluded that any two people on average are distanced by just 6.6 degrees of separation, meaning that they could be linked by a string of seven or fewer acquaintances.

The database covered all of the Microsoft Messenger instant-messaging network in June 2006, or roughly half the world's instant-messaging traffic at that time, researchers said......

For the purposes of their experiment, two people were considered to be acquaintances if they had sent one another a text message. The researchers looked at the minimum chain lengths it would take to connect 180 billion different pairs of users in the database. They found that the average length was 6.6 steps and that 78 percent of the pairs could be connected in seven hops or less.

Some pairs, however, were separated by as many as 29 hops.

The study in question is "Planetary-Scale Views on a Large Instant-Messaging Network" by Jure Leskovec and Eric Horvitz. I would note that the 180 million people who used microsoft instant messaging in that one month period were self selected from a world population of more than 6 billion people. They were very disproportionately drawn from people between 15 and 35 years of age, and from relatively high income communities. They are thus likely to be the most wired people on the earth. I would assume that the average degree of separation among the six billion people not included in the study would be much higher. (Indeed the Washington Post had a recent article about a tribe in the Amazon basin that was newly discovered and which was portrayed as one of several that are totally unconnected from modern society.)

I am not criticizing the original report of the research study, and indeed the study itself is amazingly data and computation intensive, and its report provides visual aids which beautifully summarize the relations discovered in the mass of data. I learned a lot about the people who use Microsoft Instant Messaging and how they use it.

I would criticize the report in The Washington Post, which left me with an impression that is not substantiated by the research it purports to summarize.
Some of the interesting things that were not noted might include:

• Of some 240 million IM accounts about 180 million sent messages in a one month period.
• Those 180 million people had about 30 billion conversations with about seven messages per conversation on the average. That is an average of more than 165 conversations per person, or 5.5 per day.
• The distributions of number of conversation partners and number of conversations were very "long tailed", with a very few people have very large numbers of partners and conversations.
• Younger people has more but shorter conversations; older people had fewer but longer conversations on average.
• Women messaged other women about as much as men messaged other men, but cross gender messaging was much more common than either woman to woman or man to man messaging.
• The United States appears relatively strongly connected by instant messaging to other nations; conversations betweenSaudi Arabia and other Arab nations tend to be longer than those among other nations.