In 1967, the social psychologist Stanley Milgram conducted a seminal experiment to test the hypothesis that members of any large social network (in his case, the population of the United States) would be connected to each other through short chains of intermediate acquaintances. In order to test this contention, Milgram introduced a novel technique of sending passport-like packets to a few hundred randomly-selected individuals in Nebraska and Kansas, with the aim of sending the packets to one of two ``target'' in the Boston area. The task Milgram set for his subjects had the additional constraint that each person could send the packet (after recording certain demographic details about themselves) only to someone whom they knew on a first-name basis, and who they thought was more likely to know the target than they were themselves. To inform their decisions, Milgram provided some information about the target, including their name, address, and occupation. He then tracked each of the packets, by requesting that participants tear off a card and mail it directly to him at Harvard.

His famous result, now enshrined in popular culture, and sociology dogma, was that the average lengths of the resulting acquaintance chains was roughly six, where the final member of the chain was the target itself. This result led to the phrase ``six degrees of separation'' later popularized by John Guare's 1990 play of the same name and numerous parlor games. In the era of electronic mail, and the Internet, many people, from social scientists, to mathematicians, to lay people, assume that the hypothesis has been demonstrated and that the world is, in this sense at least, ``small''.

But is it really true? A careful reading of Milgram's own findings, suggests that the small world phenomenon, as commonly conceived, rests on extremely tenuous empirical foundations. The evidence that Milgram presents in support of his hypothesis leads to a considerably more restricted claim than is usual attributed to his work (only data from a single target is used and only a few dozen chains were ever completed, yet the small world phenomenon is frequently cited as universally valid). Furthermore, according to unpublished research by Judith Kleinfeld, based on her survey of Milgram's original notes in the Yale archives, data that Milgram did not publish (on the Kansas study) did not support his hypothesis. Given the apparently tenuous nature of the results, it is perhaps surprising that no large-scale follow up studies were ever completed. Certainly subsequent studies were conducted, but these can be characterized as either as equally small or smaller (in terms of number of participants), or else as highly restricted contextually (such as within a single university).

In this project, we intend to perform the first large scale, global verification of the small world hypothesis, using the modern Email equivalent of Milgram's passport innovation. We hope to test not only average properties of lengths of acquaintance chains, but also the distribution of lengths, along with the effect of race, class, nationality, occupation, and education. We intend to quantify the impact of additional target information upon search success and chain length, and also to investigate the importance of "centers" individuals who are thought to exist who are disproportionately responsible for directing messages to the targets.

In utilizing the power and convenience of electronic mail, we understand that our protocol will exclude a large fraction of potential participants who do not have access to the relevant technology, and so our test also will be of a restricted version of the small world hypothesis. Nevertheless, the population of Email users globally is estimated to be of the order of 100 million people, which is a significant social network by any measure, and one that is surely large enough to yield statistically reliable results.