I analyzed 1,000 Hacker News front-page stories, long titles win and AI eats a quarter of the attention

Everyone has a theory about what hits the Hacker News front page: punchy titles, Show HN, post early. I stopped guessing and pulled 1,000 stories from the last 30 days through the public Algolia API. The data killed two of my own assumptions.

What

I fetched every HN story from the last 30 days that cleared 100 points, 1,000 of them, with their points, comment counts, post time, domain and title. Then I ran the statistics.

The headline numbers:

• 1 in 5 front-page stories is about AI (21.3%), and they pull 25.2% of all upvotes. AI punches above its weight on attention.
• Long titles win. Stories with 15+ word titles had a median of 415 points, versus ~274-288 for shorter ones. The "keep it punchy" advice is wrong here.
• GitHub is the single biggest source (57 stories), ahead of Twitter/X (19), TechCrunch (15) and anthropic.com (10).
• Show HN slightly beats the field, median 295 points vs 281 for regular links, 248 for Ask HN.
• Median front-page story: 281 points. Only 2% were text-only posts.

Why it matters

If you ship developer products, write technical content, or launch on HN, you're optimizing against folklore. The data says two counterintuitive things: a specific, longer title outperforms a clever short one, and AI is not "a trend", it's a quarter of the room. If you're not in tech-adjacent AI, you're competing for the remaining 75% of attention.

Who it's for

Indie hackers and founders planning a Show HN, devrel and technical writers picking headlines, and anyone who keeps hearing "just post early and keep the title short" without evidence.

When & where

The points-by-hour data peaks at 11:00 and 21:00 UTC (early morning + evening US time), but timing was a weaker signal than title and topic. What you post and how you frame it mattered more than when.

How

No scraping, no key, the HN Algolia API is open:

import urllib.request, urllib.parse, json, time
cutoff = int(time.time()) - 30*24*3600
qs = urllib.parse.urlencode({
    "tags": "story",
    "numericFilters": f"created_at_i>{cutoff},points>100",
    "hitsPerPage": 1000,
})
hits = json.load(urllib.request.urlopen(
    "https://hn.algolia.com/api/v1/search?" + qs)).get("hits", [])
# then: group by AI-keyword regex, title word-count buckets, domain, hour → median points

One honest caveat: this is survivorship data, only stories that already cleared 100 points. It tells you what front-page winners look like, not your odds of getting there. The long-title effect is partly that detailed titles often wrap big launches and deep write-ups, not that padding a title helps.

The takeaway

Stop optimizing your HN title for brevity. Be specific, be substantive, and accept that you're sharing the room with AI, which now owns a quarter of it.

Data: Hacker News via the Algolia API, last 30 days, stories above 100 points. Chart and method are reproducible, the snippet above is the whole fetch.