Reddit’s Globalization over Twenty Years: Inferring Community Time Zone from Activity Timestamps DOI: XXXXXXX.XXXXXXX CCS: Information systems Social networks CCS: Information systems Social networking sites CCS: Information systems Traffic analysis CCS: Human-centered computing Social network analysis CCS: Human-centered computing Empirical studies in collaborative and social computing

Franco Della Negra, Matteo Cinelli, Mattia Samory · 2026

Read Reddit’s Globalization over Twenty Years: Inferring Community Time Zone from Activity Timestamps DOI: XXXXXXX.XXXXXXX CCS: Information systems Social networks CCS: Information systems Social networking sites CCS: Information systems Traffic analysis CCS: Human-centered computing Social network analysis CCS: Human-centered computing Empirical studies in collaborative and social computing by Franco Della Negra, Matteo Cinelli, Mattia Samory (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-04371.txt.

Introduction

Franco Della Negra, Mattia Samory, and Matteo Cinelli estimate where Reddit communities are based using only the times at which they post. Online communities span the world, but measuring that spread at scale is hard, and hourly activity alone carries a daily rhythm tied to local time.

Their best method pins a community's time zone down to within about half an hour, and fewer than a thousand comments are enough for peak accuracy. A simple rule, that activity bottoms out near four in the morning local time, does almost as well. They validate the approach across sports, finance, and other topics, then apply it widely.

Catalog record

CREATORFranco Della Negra, Matteo Cinelli, Mattia Samory
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE73 KB
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2605.04371

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Reddit’s Globalization over Twenty Years: Inferring Community Time Zone from Activity Timestamps DOI: XXXXXXX.XXXXXXX CCS: Information systems Social networks CCS: Information systems Social networking sites CCS: Information systems Traffic analysis CCS: Human-centered computing Social network analysis CCS: Human-centered computing Empirical studies in collaborative and social computing Franco Della Negra, Matteo Cinelli, Mattia Samory Abstract . Online communities are a global phenomenon, but assessing their actual geographical spread requires accurate and scalable measurement. We propose and evaluate methods that infer the time zone of online communities solely from their temporal activity patterns, requiring nothing beyond hourly activity counts. Grounding our approach in the well-established finding that posting rhythms encode circadian structure, we compare time-domain and frequency-domain methods against a parsimonious heuristic: that activity reaches its minimum around 4 a.m. local time. On Reddit, we show that the best-performing method is accurate to a sub-30-minute resolution, and that fewer than a thousand comments are sufficient to reach peak performance. Similarly, our heuristic almost matches the accuracy of more complex methods, recovering the correct time zone within a one-hour margin on average. This simple method correlates significantly with the actual distribution of Reddit’s geographical spread; we validate its generalizability across communities organized around diverse cultural phenomena, from sports to finance, and apply it at scale to characterize the geographic evolution of Reddit from its founding to the present. Our method is portable across platforms and requires no user disclosure, making it a practical baseline for any study that must account for the geographic structure of online behavior. 1. Introduction A growing body of research has leveraged community affiliation and interaction patterns as proxies for user attributes, enabling the characterization of psychological states ( Chancellor and Choudhury 2020 ) , political orientations ( Waller and Anderson 2021 ) , and sociodemographic profiles at scale ( Tigunova et al. 2020 ) . These approaches have proven valuable both for understanding individual users and for profiling platforms as sociotechnical systems ( Balsamo et al. 2019 ; Eisenstein 2018 ; Monti et al. 2023 ; Piccardi et al. 2024 ) . Among the many dimensions that shape online behavior, geographic location occupies a privileged position: the so-called "big geographic sort", i.e., the tendency of people to cluster with others who are in the same location, manifests online just as it does offline, making geolocation a key lens through which to examine platform dynamics ( Bozarth et al. 2023 ) . Yet despite its importance, geographic location has received comparatively little attention in the computational social science literature on online communities. Where it has been studied, ground-truth labels are typically available for only a small subset of communities, leaving the broader validity of geographic inference methods largely unvalidated. This work addresses that gap by proposing and evaluating a method for inferring the geographic location of online communities (specifically, their time zone) directly from activity signals. This approach carries several methodological virtues: activity signals are difficult to counterfeit, inexpensive to compute, and, crucially, portable across platforms, as they do not depend on any platform-specific affordances or metadata. We organize our investigation around three research questions: (RQ1) How reliable is time zone inference based on time series data? (RQ2) How does inference quality vary with data availability? (RQ3) What is Reddit’s geographical distribution, and how did it evolve? Our contribution is threefold. First, we conduct a systematic comparison of multiple inference methods, evaluating both their accuracy and computational trade-offs. We demonstrate that the time zone of a subreddit can be inferred to within a one-hour margin, on average, using only activity counts, and we introduce a simple but effective heuristic that performs competitively with more computationally demanding methods: anchoring the inference on the assumption that activity is minimized around 4 a.m. local time. Importantly, observing only a few thousand comments, or less than one month of activity, suffices to recover 95% of the maximum achievable performance. Second, we provide extensive quantitative and qualitative evaluation of the method’s generalizability to unseen communities, showing that it successfully retrieves subreddits organized around salient cultural phenomena spanning sports, food, and finance. Third, we apply the method at scale to characterize the time zone composition of Reddit, showing that it correlates highly with independent measures of Reddit users’ locations (Pearson’s r\approx 0.6 ). We reveal how the platform has grown substantially less U.S.-centric and more global—albeit unevenly, with a secondary concentration around European time zones and markedly limited adoption in Asia and Oceania. These findings carry implications that extend beyond descriptive platform analysis. Time zone inference provides a principled bridge between online behavior and offline context, enabling researchers to account for—or deliberately focus on—location-specific social, cultural, and institutional factors ( Kates et al. 2021 ) . 2. Related Work We first discuss what is known about the relationship between geography and online behavior, and how researchers have tried to recover location from digital traces. We then turn to the specific signal we exploit: the temporal rhythm of online activity, which encodes where users are located. We discuss two complementary research lines: circadian analytics, which has characterized p(\text{timestamps}\mid\text{location}) and shown it to be richly structured, and the geolocation literature, which has recognized that temporal features are predictive of p(\text{location}\mid\text{activity}) yet has never evaluated them on their own terms. Our work addresses the methodological gap between circadian analytics and geolocation literature, and contributes to our understanding of the geography of online behavior by performing inference on the entire Reddit platform. 2.1. Geography Shapes Online Behavior Far from escaping their local contexts ( Turkle 1995 ; Rheingold 1993 ) , users bring their identity online: a fundamental premise of computational social science is that the digital traces left by users carry legible signals of who they are offline ( Bail et al. 2018 ; Choudhury et al. 2013 ) . Among the offline characteristics that shape online behavior, geographic location has proven structurally consequential. As an exemplary study of the “big geographic sort”, Bozarth et al. 2023 showed that the geographic composition of U.S. Reddit communities explains patterns of news circulation that platform-level algorithmic accounts had missed: interaction probability decays with physical distance even within a nominally global medium, and the community structure of the platform faithfully reproduces the cultural clustering of offline geography. Yet, despite this recognized importance, geographic location is under-studied in the literature on online communities. Where it has been examined, studies typically rely on one of two approaches, each with structural limitations. The first—and by far the most common—is to use explicit geographic signals: GPS coordinates attached to individual posts, or user-provided location fields in profiles. Geotags offer high spatial precision, but, for instance, fewer than one percent of Twitter posts carry them ( Pfeffer et al. 2023 ) , and platforms like Reddit do not support post-level geotagging at all. The second approach uses community affiliation or textual content as a proxy for location, either by identifying users who participate in explicitly localized subreddits ( Harrigian 2018 ; Alarfaj et al. 2025 ; Berragan et al. 2023 ) or by applying natural language processing to extract geographic references from post content ( Stillman and Kruspe 2024 ; Hoffmann and Heft 2020 ) . Both approaches operate at the level of individual users and

Read the complete text