Analyzing the Most Frequent Numbers in Hacker News Headlines

By Max Glenister | Published July 15, 2026

It started with a trivial observation: two consecutive headlines on the Hacker News (HN) front page both contained the digit 6. While seemingly insignificant, it was the kind of minor detail that persists in one's mind until it is resolved. This led to a central question: Which number appears most frequently in HN titles?

To find the answer, I utilized the public playground provided by ClickHouse, which hosts the complete HN dataset—a resource that is inherently thrilling for someone with my inclinations.

Phase 1: The Naive Approach

Initially, I ran a straightforward query to extract every digit sequence found in story titles.

SELECT num, count() AS c 
FROM ( 
    SELECT arrayJoin(extractAll(title, '\d+')) AS num 
    FROM hackernews_history 
    WHERE type = 'story' AND title != '' 
) 
GROUP BY num 
ORDER BY c DESC 
LIMIT 10

Initial Results

At first glance, the dominance of smaller digits suggested Benford's Law. This mathematical observation posits that in many real-world datasets, the leading digit is more likely to be small. For example, the probability of a leading digit $d$ is:

$P(d) = \log_{10}\left(1 + \frac{1}{d}\right)$

This phenomenon is often seen in stock prices, river lengths, or election data. However, Benford's Law applies to the leading digit of numbers spanning several orders of magnitude—not single digits plucked from prose.

Phase 2: Refining the Methodology

The initial results were accurate misleading because they answered the wrong question. The regex \d+ is too blunt; it treats "2.0" as a 2 and a 0, splits "1,000" into 1 and 000, and extracts the 3 from "S3". Consequently, the massive lead for the number 2 was largely an artifact of "Web 2.0" and various software version numbers.

To fix this, I implemented a more sophisticated pattern that:

• Captures comma-separated numbers as single units.
• Treats decimals as single tokens.
• Respects word boundaries (\b).
• Filters out years (1900–2030) to avoid the obvious bias of a news site.

Technical Note: In ClickHouse, I had to use double backslashes (\\b) because the system interprets a single \b as a literal backspace character before the regex engine processes it.

SELECT num, count() AS c 
FROM ( 
    SELECT replaceAll(arrayJoin(extractAll(title, '\\b\\d{1,3}(?:,\\d{3})+\\b|\\b\\d+(?:\\.\\b'))), ',', '') AS num 
    FROM hackernews_history 
    WHERE type = 'story' AND title != '' 
) 
WHERE toUInt64OrZero(num) NOT BETWEEN 1900 AND 2030 
GROUP BY num 
ORDER BY c DESC 
LIMIT 10

Refined Results

Even with this, tokenization remains a "tar pit." For instance, "Python 3.11.2" is still split into "3.11" and "2", and the filter cannot distinguish the year 1984 from the Orwellian novel.

Analysis: Why These Numbers?

The distribution of numbers in HN titles is driven by three primary factors:

• The Listicle Effect: Titles like "5 things I wish I'd known" or "10 tools I use" are standard. These counts are usually small and round. You rarely see "17 things" unless the author truly has exactly 17.
• Software Versioning: Major releases (e.g., Rust 1.0, Llama 2, Gemini 3) start at 1 and increment slowly. This creates a natural accumulation of low numbers.
• Psychological Emphasis: Round numbers are used for impact. This is why 10 ranks significantly higher than 9 or 11.

The Process Flow

Temporal Anomalies

When plotting these numbers from 2007 to 2025, two external events disrupt the baseline:

1. The Number 19: While its all-time rank is modest, it exploded in 2020 with 10,439 appearances—five times its usual annual rate—due to COVID-19.
2. The Number 4: In 2023, the number 4 spiked from ~1,400 to 2,298, briefly rivaling 1 and 2. This coincided with the launch and dominance of GPT-4.

Final Thoughts

"The two 6s I saw on the front page in a row were entirely unremarkable. I have spent an evening establishing exactly that, and I regret nothing."

---

About the Author

Max Glenister is a Senior Full-stack Developer and Interface Designer based in Oxfordshire.

• Professional focus: Designing and implementing user interfaces.
• Personal interests: 3D printing, VR, embedded systems, and game dev.

Other Recent Posts:

• The progress of progress bars (11 Jul 2026)
• How I'm using CSS View Transitions (07 Jul 2026)
• Why I'm using a low-profile keyboard (03 Jul 2026)