| Drop me a line if you'd like to discuss this / share w/ reports. username at Protonmail. I'm sort of a Can Haz All The Tables sort of guy, and I'm largely processing via awk (and a few other shell tools). So pasting that here would get a bit tedious... It's also been interesting to look at how HN has, and hasn't, changed over the years. Your categorical analysis would be an interesting filter to look at over time, especially regarding accusations that HN is drifting in various directions. The other bit that stands out to me is how constrained a set the front page is (30 slots per day, 10,950 per year, 10,980 in a leap year), as well as how thin submission titles are for gleaning meaning and context (I'm ... somewhat frustrated by this). Though there is clearly signal that gets through. I don't have time-of-day granularity, but can look at day-of-week (and have) and month-of-year (not yet) looking for seasonality. DoW has been interesting (usually peaks Tue/Wed, starts trailing off on Fri, Sat & Sun are low points, based on votes/comments, but give higher odds of a given submission landing). You might want to look at Whaly's work as well (I'd edited it into my larger top-level comment above: <https://whaly.io/posts/hacker-news-2021-retrospective>). |