Electronic communications may have shrunk, rather than expanded, horizons

　　电子通讯可能使我们的视野变窄了，而非扩大了。

　　Babies' names and the internet

　　The rise of the internet was supposed to create a global village, in which people would be as likely to have friends in the antipodes as in their own street. Poppycock, of course. But the idea that it might instead have shrunk people’s horizons is truly counter-intuitive. Yet that is what Jacob Goldenberg and Moshe Levy of the Hebrew University in Jerusalem suspect. Their evidence is indirect, and from a strange source―the spread of babies’ names. But it does suggest that something worthy of investigation is going on.

　　The two researchers’ study of the spread of new names was prompted by their discovery that the relationship between the number of private e-mails sent in America and the distance between sender and recipient falls off far more steeply than they expected. People are overwhelmingly e-mailing others in the same city, rather than those far away.

　　That says something about human relations, but not how they have changed since e-mail became ubiquitous. So Dr Goldenberg and Dr Levy needed to find something pertinent that bridged the period in question and might thus shed more light on their result. In an inspired piece of lateral thinking, they decided to look at how babies’ names spread.

　　America’s Social Security Service keeps a register that contains the 100 most popular names in each state every year and the number of babies given them. Dr Goldenberg and Dr Levy speculated that when parents chose a name for a child, they were influenced by their interactions with other new parents, so the spread of the names the babies were given was a proxy for the pattern of those interactions.

　　The researchers took data from 1970 to 2005 and traced the propagation of names in and between states over that period. They did this by calculating what they called a “proximity-effect index” for each of the 100 most popular names in each state. For each name, they divided the country into two groups: one consisting of states in which the name was already popular and their immediate neighbours; and one consisting of states in which the name was relatively unknown.

　　They then counted how many babies had been born in each state, and the proportion given a certain name. For example, if an eighth of babies born in America were delivered in California then, all other things being equal―in particular, if geography made no difference―you might expect an eighth of all babies with any given name to be Californian.

　　What Dr Goldenberg and Dr Levy found was that the proportion of babies given a certain name in a state where that name was already popular or in a neighbouring state was 20% higher than would otherwise have been expected. This was true from the 1970s to the early 1990s.

　　From 1995 to 2005, however, the effect became even more pronounced. The proportion of newborns with common names in any given state and its immediate neighbours became 30% higher than would have been expected if there were no geographic effect. Dr Goldenberg and Dr Levy ascribe this rise to the internet. It certainly correlates with the emergence of the web, though whether the correlation reflects causation is unproven. But whatever the reason, it is a curious result.