Characterization of Strong and Long-Range Ties

A recent paper by Park, Blumenstock and Macy [1] provided evidence on the existence of strong long-range ties in population-scale networks from several countries and network types (Twitter mention network and phone call networks). Strength of the tie is measured in terms of interaction frequency and tie range is the length of the shortest path between the two nodes, if their direct link was to be removed. In this sense, long ties are those that span across diverse portions of the network that would otherwise be disconnected and share no common contact.

Long-range ties are often assumed to be weak [2]. However, there is a confusion between the range of a tie and its strength. The range of the tie is correlated with its strength, but the long-range ties do not necessarily have to be weak. In fact, the paper by Park et. al [1] provides compelling evidence on the existence of strong long-range ties, though a small fraction of all network edges. These long-range ties do not exhibit the structural characteristics of typical strong ties such as clustering or redundancy, so the nature of these ties and exchanges that happen across them is not well understood. In this project, we follow up on Park et al. study and aim to answer the following questions:

1. Does the Facebook social network exhibit the same pattern as observed in Park et al. [1]? Facebook maintains a single account for each individual. In contrast, multiple nodes on Twitter and phone communication networks could correspond to the same person (e.g., use of multiple phone numbers and twitter handles)? How frequent are strong long-range ties in Facebook? Are they less or more frequent than what’s reported in the paper?
2. Only 0.46% of all ties are strong long-range ties, hence it is very likely that not all individuals in the network have a strong long-range tie. In fact, maintaining such ties tends to require certain skills and commitment. What are the demographic characteristics of individuals that maintain strong long-range ties? What about their structural characteristics? For example, do strong long-range ties connect two periphery nodes, a periphery with a central node or two central nodes? Similarly, what is the degree distribution of nodes connected by a strong long-range tie?
3. What is the nature of strong long-range ties in Facebook? Are they work related? Or are they concerned more with interpersonal contact? Are they old ties (e.g. from high school) that remain strong as the individuals become structurally different over time? Or individuals acquire them as new ties by finding contacts that are structurally diverse? What is the distribution of geographic distance on strong long ties? Do these ties occur over long or short geographic distances?
4. The study of network resilience makes the distribution of tie ranges important. What’s the distribution of tie ranges in Facebook? How are long-range ties distributed among nodes? Do some nodes have a disproportionate level of long-range ties or are they distributed uniformly among all the nodes? The distribution of tie-ranges along with the distribution of these ties on the nodes could act as a measure of network fragility. They determine how vulnerable the network is to removal of long-range ties or nodes that hold them before the network becomes fragmented.

References

[1] Park, Patrick S., Joshua E. Blumenstock, and Michael W. Macy. “The strength of long-range ties in population-scale social networks.” Science 362.6421: 1410-1413, 2018.
[2] Granovetter, Mark S. “The strength of weak ties.” Social networks. Academic Press, 347-367, 1977.