Live data from Hacker News

Hacker News Data Analysis

blog.rjmetrics.com

11–20 of 79 posts

Re: Hacker News Data Analysis

#11
I once worked out there were 100:1 visitors to voters for a link.

Most of the people I know who peruse HN regularly are not registered users. They are happy to let others do the commenting (which they read).

http://williamedwardscoder.tumblr.com/post/18839832580/reddi...

It was super-surprising to see my own blog getting an average of 55pts on HN; I hadn't wondered about that before.

Re: Hacker News Data Analysis

#12
Very useful analysis. After running Hacker Newsletter for the past 2+ years I have seen basically this. However, the analysis seems to miss looking at things on a smaller scale like the day and time you post it which has proven to be a big factor [1]. I know even on a weekly basis (which is what I do for the newsletter), it seems some weeks have an abundance of high quality articles compared to others.

[1]: http://news.ycombinator.com/item?id=3251877

Re: Hacker News Data Analysis

#13
The retention rate actually seems relatively low as an absolute percentage, though the way it plateaus is interesting. I did an analysis of the retention of the oldest Slashdot users (http://www.kmjn.org/notes/early_slashdot_users.html), and it was much higher: about 70% after 2 years, rather than 30%. Took about 10 years to drop to 30%. Granted, that's for the earliest users, so retention rates are probably (much?) lower among later signups.

Re: Hacker News Data Analysis

#15
post #4
post #2

My takeaway --- from the fact that Matt Might's domain is second only to pg's --- is that you should write up easy to understand lecture notes on deep PL-related topics.

The fact that this story went straight to the top suggests that a meta-post about HN is the way to go.

Data-based posts about how to get to the front page, sure.

But otherwise, I'm not sure it really pays off.

Re: Hacker News Data Analysis

#16
#1 Lesson from all of this: Instead of talking about your product to your prospect, talk about something your prospect cares deeply about to your prospect.

I had no idea what you did and didn't really care until you used it in context of something I did care about: Hacker News. Now I know what you do, understand how it applies to me, and best of all, I'm starting to visualize how else I could use it.

We should all approach our prospects like you just did here. Nice job!

Re: Hacker News Data Analysis

#17

you say the two possible reasons you are not making the front page are: your content is weak, or people's taste's have changed. The fact that the number of submissions has not changed suggest to me a third and more plausible option: The quality of submissions, and therefore the competition for the "front page" has increased.

Doesn't "My content is weak" contain the implicit qualifier "compared to the average within the domain we're discussing"?

Re: Hacker News Data Analysis

#18
post #17

you say the two possible reasons you are not making the front page are: your content is weak, or people's taste's have changed. The fact that the number of submissions has not changed suggest to me a third and more plausible option: The quality of submissions, and therefore the competition for the "front page" has increased.

Doesn't "My content is weak" contain the implicit qualifier "compared to the average within the domain we're discussing"?

No. You can have a domain of discourse where ALL content is weak. In fact, you can have content in this domain that is stronger than most of the content, but it might still be weak.

Re: Hacker News Data Analysis

#19
I've noticed that some submissions drop off the news feed like a rock, while other submissions of the same story posted just a few hours later can gather considerable discussion, with submission time being the only apparent variable.

This leads me to speculate that there may be an optimal submission time or times throughout the day. I'd like to see analytics that look at the variation in the average number of comments/upvotes for submissions (or some other metric) to see if this theory holds any weight.

Post reply on HN