An enormous body of literature argues that recommendation algorithms drive political polarization by creating “filter bubbles” and “rabbit holes.” Using four experiments with nearly 9,000 participants, we show that manipulating algorithmic recommendations to create these conditions has limited effects on opinions. Our experiments employ a custom-built video platform with a naturalistic, YouTube-like interface presenting real YouTube videos and recommendations. We experimentally manipulate YouTube’s actual recommendation algorithm to simulate filter bubbles and rabbit holes by presenting ideologically balanced and slanted choices. Our design allows us to intervene in a feedback loop that has confounded the study of algorithmic polarization—the complex interplay between supply of recommendations and user demand for content—to examine downstream effects on policy attitudes. We use over 130,000 experimentally manipulated recommendations and 31,000 platform interactions to estimate how recommendation algorithms alter users’ media consumption decisions and, indirectly, their political attitudes. Our results cast doubt on widely circulating theories of algorithmic polarization by showing that even heavy-handed (although short-term) perturbations of real-world recommendations have limited causal effects on policy attitudes. Given our inability to detect consistent evidence for algorithmic effects, we argue the burden of proof for claims about algorithm-induced polarization has shifted. Our methodology, which captures and modifies the output of real-world recommendation algorithms, offers a path forward for future investigations of black-box artificial intelligence systems. Our findings reveal practical limits to effect sizes that are feasibly detectable in academic experiments.

Download Full Paper

Read Press Release (February 2025)

How a team talks is a rich source of information about its collaboration process, revealing how its members resolve conflicts, coordinate activities, and monitor goal progress. However, the full potential of communication data often goes unrealized: analyzing communication requires a series of contingent decisions about which constructs to study and how to measure them. Consequently, researchers interested in team communication face a high barrier to entry, challenges with reproducibility, and constraints on their ability to capture phenomena that emerge from interactions between multiple conversational dimensions. To address these challenges, we present the Team Communication Toolkit (team_comm_tools on the Python Package Index), an open-source Python package that extracts more than 160 communication features from any corpus of text-based conversations. Our collection of measures allows the analyst to understand the effect of conversational features in the context of a broader family of features that may equivalently explain the phenomenon. Rather than forcing researchers to commit to a single measurement of complex phenomena such as constructiveness or assertiveness, the toolkit’s flexible output format makes it possible to systematically compare measures, isolate confounders, and identify higher-order interactions. We call our general analytical approach the “Communication Space,” reflecting the fact that researchers can now more easily explore the realm of possible ways to analyze conversational data. Finally, we demonstrate the utility of our tool and framework through a case study exploring conflict escalation in both real-world and synthetic data.

Download Preprint

Read Press Release (August 2024)

Read Penn Today Feature (August 2024)

Many studies of team performance involve teams working on one specific task or at most a handful of tasks; thus, it is natural to ask how the results of any such study generalize beyond the task(s) at hand. However, this question has been challenging to answer due to limited commensurability among tasks: although many taxonomies have been proposed, it is still difficult to quantify how similar or different any task is to any other. Consequently, determining whether a result acquired in connection with one task can hold on other tasks has remained challenging. This paper addresses the commensurability problem by introducing a multidimensional representation, the “Task Space,” that synthesizes 24 task dimensions (e.g., “degree of creativity required,” “amount of physical effort required,” “extent of demonstrable correctness”), making it possible to quantify a between-task distance measure. We then use the Task Space to label 102 laboratory tasks from an interdisciplinary social science literature. To demonstrate the utility of our approach, we analyze an empirical dataset from human subject experiments of 125 teams completing 10 different tasks. Our findings show that the Task Space outperforms existing taxonomies in predicting key team outcomes, including a team’s solution quality and attributes of its communication patterns. Our representation also reveals how common team composition factors, such as its size and level of cognitive reflection, have different impacts on a team’s performance depending on specific task characteristics. These results highlight the Task Space's potential for enhancing our understanding of team performance across varied contexts.

Download Preprint

When conflicts escalate, is it due to what is said or how it is said? In the conflict literature, two theoretical approaches take opposing views: one focuses on the content of the disagreement, while the other focuses on how it is expressed. This paper aims to integrate these two perspectives through a computational analysis of 191 communication features -- 128 related to expression and 63 to content. We analyze 1,200 GPT-4 simulated conversations and 12,630 real-world discussions from Reddit. We find that expression features more reliably predict destructive conflict outcomes across both settings, although the most important features differ. In the Reddit data, conversational dynamics such as turn-taking and conversational equality are highly predictive, but they are not predictive in simulated conversations. These results may suggest a possible limitation in simulating social interactions with language models, and we discuss the implications for our findings on building social computing systems.

Download Short Paper

By identifying the socio-technical conditions required for teams to work effectively remotely, the Distance Matters framework has been influential in CSCW since its introduction in 2000. Advances in collaboration technology and practices have since brought teams increasingly closer to achieving these conditions. This paper presents a ten-month ethnography in a remote organization, where we observed that despite exhibiting excellent remote collaboration, teams paradoxically struggled to collaborate across team boundaries. We extend the Distance Matters framework to account for inter-team collaboration, arguing that challenges analogous to those in the original intra-team framework --- common ground, collaboration readiness, collaboration technology readiness, and coupling of work --- persist but are actualized differently at the inter-team scale. Finally, we identify a fundamental tension between the intra- and inter-team layers: the collaboration technology and practices that help individual teams thrive (e.g., adopting customized collaboration software) can also prompt collaboration challenges in the inter-team layer, and conversely the technology and practices that facilitate inter-team collaboration (e.g., strong centralized IT organizations) can harm practices at the intra-team layer. The addition of the inter-team layer to the Distance Matters framework opens new opportunities for CSCW, where balancing the tension between team and organizational collaboration needs will be a critical technological, operational, and organizational challenge for remote work in the coming decades.

Download Full Paper

A jury of one’s peers is a prominent way to adjudicate disputes and is increasingly used in participatory governance online. The fairness of this approach rests on the assumption that juries are consistent: that the same jury would hand down similar judgments to similar cases. However, prior literature suggests that social influence would instead cause early interactions to cascade into different judgments for similar cases. In this paper, we report an online experiment that changes participants’ pseudonyms as they appear to collaborators, temporarily masking a jury’s awareness that they have deliberated together before. This technique allows us to measure consistency by reconvening the same jury on similar cases. Counter to expectation, juries are equally consistent as individuals, a result that is “good for democracy.” But this consistency arises in part due to group polarization, as consensus develops by hardening initial majority opinions. Furthermore, we find that aggregating groups’ perspectives without deliberation erodes consistency.

Download Full Paper

View Data/Materials Released via Open Science Framework

Gender bias has been a pervasive issue in U.S. political coverage since women gained the franchise in 1920. Women are routinely targeted for their appearance, likeability, and familial qualities—characteristics for which men are rarely scrutinized. As a way to help journalists and editors identify and correct gender-biased language in political reporting, we introduce “Disarming Loaded Words” (DLW) as a new solution to this issue. DLW is a computational tool, backed by both machine learning and human expert curation, that tags potentially biased words in a document and provides feedback and context for why a word may be problematic. DLW provides a minimally disruptive way to nudge journalists toward understanding unconscious biases. It was successfully prototyped as a standalone Google Docs add-on, and the code for its model, interface, and API are now open-source and available for use on other platforms.

Download Full Paper