A Deep Dive into X's Community Notes: An Analysis of English and Spanish

Watch the launch briefing . 

 Download the Report PDF: 

 
 Download
 1. INTRODUCTION

 As social media companies face the need to curb the spread of rapidly evolving online harms on their platforms, “community moderation” has emerged as a popular experiment for addressing false, misleading, or “confusing” content. X, whose Community Notes system has invited users to collaboratively annotate posts for years, was at the helm of this model of moderation, with other tech giants now jumping in to implement their own, similar versions. Meta rolled out user-driven feedback tools based on X’s open-source software in April 2025. A few weeks after, TikTok announced Footnotes, its pilot to crowdsource context.

 Though Community Notes has become a flagship model for decentralized content moderation, the approach itself is hotly debated. Supporters call community-driven content moderation a democratic alternative to opaque, top-down decision-making. Critics argue this type of system shifts too much responsibility onto the users and is too slow and unbalanced to meet the scale and speed of today’s information crises.

 To offer a clear understanding of how Community Notes actually works, what could be improved, and the potential of this model for strengthening a healthier online ecosystem, the Digital Democracy Institute of the Americas (DDIA) set out to analyze the entire public dataset published by X between January 2021 and March 2025: more than 1.76 million notes written in 55 languages and available on the platform’s website.

 The first of multiple studies DDIA hopes to conduct on Community Notes, this analysis focuses on the structure, operational mechanisms, and distribution of notes in English and Spanish on the platform. As other tech companies consider whether and how to build their own community-driven moderation systems, the findings presented in this report shed light on the flagship model’s pitfalls, performance, and potential.

 2. KEY FINDINGS

 ABOUT X’S COMMUNITY NOTES PUBLIC DATA 

 What’s in the data: X received 1.76 million Community Notes across 55 languages between January 2021 and March 2025.

 What’s positive: The data is technically public and accessible for download, a rare move among social media platforms.

 What’s challenging: 

 The dataset is massive and tough to analyze effectively at scale without the use of cloud-based tools (e.g., BigQuery).

 Crucial metadata (including language tags, geographic information, and contributor demographics) is absent.

 Free access to X’s API has been closed since 2023 (to everyone), which means those studying the dataset can only see which Community Notes are connected with which X posts individually.

 Why it matters: These limitations make it hard for external researchers to fully evaluate the system’s reach, fairness, and impact.

 ABOUT NOTES SUBMITTED TO THE PROGRAM 

 What’s in the data: English and Spanish are the two most-used languages in Community Notes. 1.12 million notes were 

... (truncated, 30 KB total)