Bridging-Based Ranking Framework
by Keith Coleman & Jay Baxter • Product Lead (Keith) & Founding ML Engineer (Jay) for Community Notes at X (formerly Twitter)
Keith Coleman is a veteran product leader who previously led Consumer Product at Twitter before founding the Community Notes (formerly Birdwatch) team. Jay Baxter is the founding machine learning engineer who architected the bridging-based algorithm that powers the system.
🎙️ Episode Context
Keith Coleman and Jay Baxter reveal the inner workings of Community Notes, X's decentralized fact-checking system that has revolutionized online moderation. They discuss the "Bridging-based" algorithm that prioritizes consensus across polarized groups, the "Thermal" team structure that allows them to ship fast with a lean team, and the counter-intuitive product principles—like anonymity and open-sourcing algorithms—that built user trust.
Problem It Solves
Solving the problem of polarization and bias in crowdsourced moderation, where a simple majority vote would just reflect the dominant group's opinion.
Framework Overview
Instead of asking 'is this note popular?', the algorithm asks 'is this note helpful to people who usually disagree?'. It utilizes matrix factorization to identify user clusters and only surfaces content that bridges the divide between polarized groups.
🧠 Framework Structure
Identifies 'polarized' clusters based...
Requires a 'Bridging Signal': Positiv...
Prioritizes context over censorship: ...
Considers 'Helpfulness' as the North ...
When to Use
When building moderation systems, reputation systems, or recommendation engines in highly polarized or subjective environments.
Common Mistakes
Relying on simple majority voting (Mob Rule) or using engagement metrics (Likes) as a proxy for truth/quality.
Real World Example
During the Israel-Hamas conflict, despite extreme polarization, Community Notes successfully surfaced factual corrections because the algorithm filtered for objective facts that both sides implicitly acknowledged.
We actually look for agreement from people who have disagreed in the past... that's what makes the notes so neutral and accurate.
— Keith Coleman & Jay Baxter