Analyzing the Top Trends on Hacker News
Blog post illustration

Analyzing the Top Trends on Hacker News


To uncover the most popular topics on Hacker News, we tracked the frequency of posts by topic over a three-month period. We originally built this because we were overwhelmed by the pace of AI news and papers being released, so we created an automated HN news monitoring service that delivers relevant news straight to our inbox or RSS feed.

The service is called HackSnack and uses Kadoa to extract, summarize, and classify the front page posts and analyzes the different perspectives in the comments. Every three hours, we extract the HTML content of each front-page post link and analyze it.

Here is how the workflow looks like:

Flow builder

And this is how an analyzed post looks like:

Hacksnack

Most Selected Topics by Users

The users preferences provide an interesting perspective on current interests and trends of the HN community. The most selected topics for monitoring reveal what AI is the currently the most popular topic on HN, reflecting the ongoing AI hype. HN has always been the place where people share OSS projects, so it's not surprising that 'Open Source' is the second most popular topic.

It's important to note that while these topics are pre-defined by us, users have the freedom to add their own custom topics. We actively add new popular topics to the list, making sure that the selection evolves with the community's interests. Additionally, synonymous tags such as 'Generative AI,' 'Artificial Intelligence,' 'AI,' and 'LLMs' are merged together.

Chart

Most Popular Front-Page Topics

Unsurprisingly, we see that the most discussed topics closely align with user preferences, likely because these are the posts that attract the most upvotes.

Chart

We plan to publish some more insights over a longer time period and we'll continue to refine the technology. Some things we have in mind:

  • Sentiment analysis
  • Insights into the comment perspectives
  • Highlight changes over time with an interactive historical data page

While the methodology is not yet perfect, we believe it presents an interesting way to observe trends and shifts in the interests of the HN community.