Is that data only from the front page? In that case, it doesn't correct for the likelihood of making it to the front page.

Yep, it is: if you view the csv with the original scraped data, you will find the rank of submissions is at most 30. (Also, you would never get average point scores upwards of 20 and 30 if all submissions were considered. And you were probably posing a rhetorical question, but heck, I'll answer it :)).

It is possible that more interesting articles are posted in the afternoon due to the work habits of the people who actually create the content that is being linked, and posts linking these articles happening in a short time span.

It may be possible to correct for this by only including links that are for content that was already "old" at the time of posting?

Interesting that you didn't follow your own advice by picking a time in the afternoon.

I made something that was of interest to me, so I thought to share it with like minded people..

That's afternoon UTC, so the submission time was near the higher point time period.

While there's no doubt that some times are better than others to submit, I think it's best not to abuse this to get karma. It's nice having at least a couple hour window where self-submitted blog posts from members of the HN community have a chance to make the front page, but this goes away if people are spamming the system with random stuff.

I posted this chart in the other thread, based on the same data.

http://tinyurl.com/hnrank

Shows the fraction of #1 stories, by submission time.

It seems HN never sleeps.

And that it's an international site.

Lots of hackers also have irregular sleep hours :)

The title doesn't match the article.

The article is average points and comments based on submission time.

Appearance on the front page is a relative thing depending on the scores of the other articles. So if 20 articles all go through at "peak" submission time, nobody goes anywhere.

What I would love to see is correspondence between points/page-rank and actual impressions the site gets.

I am assuming points have a better correlation with impressions (since readers are the ones who vote) than rank.

Anyone has any idea how to get better stats ?

Interesting idea.

Here's a small set of data:

http://breckyunits.com/hackernews_data_visits_as_a_function_...

I would be interested to see the data splitted out by day also. I can imagine, HN'ers are more active on fridays than on mondays.

This is interesting. The thing that jumps out the most to me is that there's definitely not an order of magnitude difference. Thus, I don't think when you submit your story should be a concern. Quality and relevance have much, much bigger impacts.

Can you do a days-of-the-week version of this? I bet Sunday doesn't look like Thursday.

Added, and you are right. Fri to Sun is ~30% difference.

What time zone is this?

It is shown on top of the bar chart (UTC)