Show HN: I analyzed 6 years of Hacker News data and here's what I found
app.hex.techI built this as part of a hackathon after analyzing Hacker News data from 2020–2025. While digging through the data, I noticed a few patterns around front-page posts that I didn’t expect, so I thought I’d share them here!
One thing that stood out is how misleading title-based predictions can be.
Most attempts to predict HN success rely heavily on the title. The problem is that we’re missing a crucial top-of-the-funnel signal: clicks. Two submissions can both have 3 upvotes, but one might have been seen by 30 people and the other by only 5. On HN, these look identical.
In reality, the first likely had a more clickable title, while the second had much higher engagement. If the second gets more exposure, it often ends up taking off. But without impression or click data, we can’t tell the difference. All we see are votes.
That missing signal makes it impossible to reliably decide what a “good” HN title is based on outcomes alone, and it explains why many posts feel unpredictable even in hindsight.
No comments yet.