Yesterday while pondering just that, I spoke with a friend and we decided to pull some data and make some graphs.
First order of business was to get a DB of posts, and it seemed like crawling was the only option.
But then luck strikes, and someone posts 6 months of HN data* !
Since what really interests me is the amount of traffic a post generates. I decided to get the average rating. With the assumption being:
- People read -> People vote -> High rank
I have ignored the relative placement on the main page, since a high placement during low-hours does not drive higher traffic, making the placement a poor indicator.
So, a bit of python and here it is:
The average Points and Comments for each post, based on its submission time (UTC).
The average Points and Comments for each post, based on its submission day-of-the-week.
With timing you can get your post almost 50% more.
These charts show the relative percentage of submission of front page articles based on time of submission (Hour(UTC) and Day-of-week):
* Please note that the DB used, only provides information on posts that made it to the main page. “Failed” submissions are not counted and might affect the results.
What do you think the chances are that showing your data/research will alter the optimal time?
It’s improbable. But, it depends on the community.
Kevin Vogelsang
8 Mar 10 at 16:15
@Kevin – I would be deeply honored that 10 minutes of effort managed to affect any sizable community. But there is no chance of that.
Most people post when they want, not everyone are after the “score” glory.
BorisDinkevich
8 Mar 10 at 16:33
interesting, good work dude
steve
8 Mar 10 at 16:40