Understanding collective regularity in human mobility as a familiar stranger phenomenon

Yan Leng, Dominiquo Santistevan, Alex PentlandScientific 2021 · DOI

Abstract

Beyond the physical structures that contain daily routines, urban city dwellers repeatedly encounter strangers that similarly shape their environments. Familiar strangers are neither formal acquaintances nor completely anonymous faces in daily urban life. Due to data limitations, there is a lack of research focused on uncovering the structure of the "Familiar Stranger" phenomenon at a large scale while simultaneously investigating the social relationships between such strangers. Using countrywide mobile phone records from Andorra, we empirically show the existence of such a phenomenon as well as details concerning these strangers' relative social relations. To understand the social and spatial components of familiar strangers more deeply, we study the temporal regularity and spatial structure of collective urban mobility to shed light on the mechanisms that guide these interactions. Furthermore, we explore the relationship between social distances and the number of encounters to show that more significant physical encounters correspond to a shorter social distance. Understanding these social and physical networks has essential implications for epidemics spreading, urban planning, and information diffusion.

Introduction

In 1972, psychologist Stanley Milgram introduced the concept of the "Familiar Stranger"—individuals who share physical spaces repeatedly, such as at a bus stop or a coffee shop, yet never formally interact. While these individuals form a latent social fabric within urban environments, they have historically been difficult to study at scale due to data constraints. This research addresses this gap by leveraging longitudinal mobile phone data to quantify the regularity of these encounters, their mathematical underpinnings, and their relationship to broader social networks.

Methods and Data

The study utilizes Call Detail Records (CDR) from Andorra, covering the period from April 1, 2016, to July 31, 2016. The initial dataset comprised approximately 2.78 million users, which was filtered to 54,758 regular residents to focus on stable urban patterns. This resulted in 51.8 million unique encounter pairs and 140.6 million total instances.

The methodology employed a dual-layered approach:

  1. Mobility Network: Encounters were defined as two users connecting to the same cell tower within the same 15-minute window.
  2. Social Network: Social ties were defined by direct call or SMS history. True "strangers" were isolated by filtering out pairs with existing communication history.

Analytical techniques included temporal analysis of inter-event times, gravity modeling to predict spatial interactions, and Point of Interest (POI) mapping via Voronoi tessellation to categorize encounter locations.

Results

Temporal Regularity and the 24-Hour Pulse

A key finding is the high degree of "collective regularity" in human mobility. Encounters are not random; they are driven by synchronized daily rhythms. Analysis of re-encounter intervals reveals sharp peaks at 24-hour and 7-day intervals, suggesting that physical encounters are highly predictable based on routine.

Percentage of re-encounters over time, showing strong 24-hour and weekly periodicities.

Spatial Structure and Gravity Modeling

The spatial distribution of these encounters follows a gravity model ($T_{ij} = C \frac{N_i^\alpha N_j^\beta}{D_{ij}^\gamma}$), which predicts the flow of re-encounters between locations. The model fit demonstrates that while physical distance ($D_{ij}$) discourages encounters, the popularity of locations ($N_i, N_j$) is a more significant driver of repeated sightings.

Actual vs. predicted encounter pairs using the gravity model, demonstrating high predictive accuracy.

Social Proximity

The research highlights a significant correlation between physical encounters and social distance. As the frequency of physical encounters between strangers increases, the average "graph distance" (degrees of separation) in the communication network decreases. This suggests that repeated physical proximity is a strong precursor to social proximity, even in the absence of direct interaction.

Discussion

While CDR data provides a powerful lens, it remains a proxy for social ties. The unique geography of Andorra may also influence the specific mathematical parameters observed. Future work should explore these dynamics in larger, more diverse metropolitan areas and investigate the transition points where familiar strangers become formal acquaintances. These insights have profound implications for urban planning and public health, particularly in modeling the spread of information or disease through latent social networks.

Summary updated

Abstract

Beyond the physical structures that contain daily routines, urban city dwellers repeatedly encounter strangers that similarly shape their environments. Familiar strangers are neither formal acquaintances nor completely anonymous faces in daily urban life. Due to data limitations, there is a lack of research focused on uncovering the structure of the "Familiar Stranger" phenomenon at a large scale while simultaneously investigating the social relationships between such strangers. Using countrywide mobile phone records from Andorra, we empirically show the existence of such a phenomenon as well as details concerning these strangers' relative social relations. To understand the social and spatial components of familiar strangers more deeply, we study the temporal regularity and spatial structure of collective urban mobility to shed light on the mechanisms that guide these interactions. Furthermore, we explore the relationship between social distances and the number of encounters to show that more significant physical encounters correspond to a shorter social distance. Understanding these social and physical networks has essential implications for epidemics spreading, urban planning, and information diffusion.
BibTeX
@inproceedings{leng2021understanding,
  author = {Yan Leng and Dominiquo Santistevan and Alex Pentland},
  title = {{Understanding collective regularity in human mobility as a familiar stranger phenomenon}},
  booktitle = {Scientific},
  year = {2021},
  doi = {10.1038/s41598-021-98475-x},
}
Figure