SNA Classics - Revisited: The Strengh of Weak Ties (Granovetter)
עודכן: 31 בדצמ׳ 2022
This paper, published 1973, is an SNA classic with more than 59,000 citations. It's author, M.S. Granovetter, is known for his research about how people find out about vacant jobs and the importance of weak social ties in this process. These groundbreaking findings were counter-intuitive because research intuition till then emphasized the importance of the individual's strong ties to his social surrounding. Intuitively, I guess most of us think about our strong ties as more important and influential than our weak ties.
I also guess many SNA analysts will content themselves with the gist of his research, but the article contains additional insights relevant to this day, even after 46 years:
Network distance: Granovetter shows how a information languishes beyond the first and second circle of ties in an Ego Network - a thing to keep in mind when performing a snowball sampling on a dataset. Interestingly, this issue relates to the NSA's hearing in congress 2013. If we add Robin Dunbar's famous research to to Granovetter findings, we get that people know their first circle very well, but they know only a small percent of their second circle and none of the people from the third circle. During the hearings, the NSA denied conducting surveillance of American citizens but noted that the "agency can perform 'three-hop queries' through Americans' data and records". So, "three-hop queries" conducted on terrorists, meaning investigating terrorist's third circle of acquaintances, is almost by definition an investigation of ordinary civilians, because the ego investigated doesn't know its third circle of ties.
The Micro/Macro Paradox of strong ties: An interesting theory (which Granovetter mentions as yet-to-be-proven) is that strong ties - on the micro level - create cohesion, while paradoxically, fragmenting the network on the macro level. In the early 2000's, network scientists were able to demonstrate this theory on communities (clusters) in networks, as in the case of Girvan-Newman Algorithm for community detection. The algorithm is based on removing weak ties in the network to expose clusters of strong ties. Also Dunbar, which we have mentioned earlier, suggested that a person can keep strong ties with up to 150~ people (the famous "Dunbar Number").
The article raises methodological points to consider while assessing a theory, especially when the theory rests upon the author's intuition or common sense, but lacks empirical proof. If anything, SNA taught me to throw my intuition/life experience/common sense out of the nearest window when it comes to analyzing complex systems. But, hey, isn't life itself a complex system? Well, I guess that's what makes it interesting :)
I guess most people aren't aware that this iconic paper was rejected when first introduced in 1969. Maybe its publishing later on has to do with the surge in unemployment that the US has seen in the early 1970's. In 1969 unemployment was about 3% and in later years more than doubled. So a paper about how to get a job might have sounded relevant. An even more important unknown fact about this paper is that Granovetter revisited the paper 10 years after its publication. At first, he found confirmation for the importance of ties in a study about a Canadian company, that out of over 2000 employees, 40% of them got the job through relationships, though it was against company's policy. But when he delved through more recent studies, he found that the idea behind "weak ties" as a stepping stone for a career didn't match the data: Blue-collar employees found most of their jobs via strong ties, and their weak ties (usually to other weakened communities) achieved less jobs with smaller pay.
Paradoxically, it means that the strong rely on their weak ties and the weak on their strong ties (family, for example) to get jobs. Not surprisingly, this paper got only tenth of the citations that the original paper got. No one wants to be the bearer of bad news... That's another reason why I admire Granovetter - he was big enough to correct himself. <Hat off>