How To Get Upvoted On Hacker News
Hacker News regularly publishes a dataset to Kaggle. As I looked at my posts go dying quickly down the “new” section, I wondered what was the relationship between score and how long the story was remaining in the visible portion of the “new” section, raising the probability of upvote.
Let’s look at the distribution of scores. Scores start at 1 (you upvote yourself automatically when posting your story, that’s how far I got :)).
Taking a look at the correlation matrix between score and couple of computed properties, the top influencer is the day of the week.
Also during the week-end, less stories are posted than during the week. The average time between stories on Saturday is around 3mn, about twice as much as during the week such as Monday.
Not shockingly, when plotting the average score against the average time between stories, you get a pretty nice trend. What should be of equal influence is the number of viewers which we don’t know, but from making the simplistic assumption that avg score = k * nreader * time between stories, then there should be around 25% less readers over the week-end on HN.
