New study shows X's breaking-news speed depends on how you measure it
Researchers find opposite conclusions on X's news latency depending on sampling method.
Deep Dive
A new arXiv paper by Bazyari, Liu, and Moran shows that breaking-news timeliness conclusions depend heavily on how events are sampled. When using Wikipedia Current Events Portal events, news leads X by 21.6 minutes. But with Polymarket trading-volume spikes, the race is essentially tied. Meanwhile, Bluesky, Facebook public, and YouTube now account for 24–32% of earliest reports, and commercial data providers miss coverage for 24% of random events—exposing structural gaps in social-listening data.
Key Points
- X leads newswire by 21.6 min when sampling from Wikipedia, but ties at -0.02 min when sampling from Polymarket.
- Bluesky, Facebook public, and YouTube together account for 24–32% of earliest breaking-news wins across all events.
- 24% of randomly sampled Wikipedia Current Events Portal events have no on-topic coverage from the commercial provider used.
Why It Matters
Choosing the wrong sample can flip conclusions on X's speed; caution needed for news latency studies using commercial data.