What's the most popular number in Hacker News titles?
3 hours ago
- #Hacker News
- #data analysis
- #popular numbers
- The most popular number in Hacker News titles is initially found to be 2 using a naive regex query on the HN dataset, which counts digit sequences regardless of context.
- A smarter regex pattern that captures whole comma-grouped numbers, treats decimals as single tokens, and filters years reveals that 1 becomes the most popular number, followed by 2 and 3.
- Small numbers dominate due to three main reasons: listicles (e.g., '3 ways to do X'), version numbers (e.g., software releases like Rust 1.0), and round numbers used as thresholds or rankings (e.g., 'Top 10').
- External events affect number popularity, such as '19' spiking in 2020 due to COVID-19 and '4' jumping in 2023 because of GPT-4's launch.
- The analysis concludes that the prominence of small numbers in titles is not due to Benford's Law but rather to common content patterns and specific historical events.