Very interesting data set. Appears the average number of people who reply to my comments is 1.03, which suggests a style of writing/thinking can contribute consistent net-growth.
Of course I would promote that idea, but really I'd be willing to bet it looks something like a pareto distribution, and it would be interesting to see what puts a user on one side of that curve or the the other.
patio11's 1.39 average reply rate would be an outlier, and there are people whose comments have the authority to simply close discussions that would need upvote stats to distinguish them might get lost viewed in that narrow dimension, but for all the time people spend on this site, it would be interesting to get a sense of what that might mean.
Not sure I agree with your conclusions. A high amount of response could just mean a controversial/inflammatory comment or a comment of inquisitive nature that people want to respond to, instead of a high-value comment.
I apparently have an avg reply rate of 1.8 and would not say that my comments are somehow better than patios.
Popular claptrap might not be valuable. Engagement can be value, and a many-to-many thread is typically good, except when it becomes a 1:1 drill down match. I think a sincere controversy is exceptionally valuable because it frames something essential about the topic, at least when it isn't just the iteration of talking points. A bold provocation can also be useful because the quality of the responses may reveal unexamined assumptions.
Better? Meh, but indicative of a certain quality factor, I'd say no doubt.
Cool idea. Just to explore it further, I think it’d be interesting to create a model that included a couple other major factors, like when the comment was posted relative to the front-paging of the article and where the parent comment fell on the page (this comment is responding to the top comment, which will help it a lot).
I guess I’m not aware of an analysis with this goal (who and what types of comments engender more conversation) anywhere else, and actually doing it would probably be pretty fascinating. Again, cool idea.
About a year ago, I saw a Hacker News comment by someone claiming to sleep on average 0-2 hours a day. There were 20 replies to that comment but (he) never responded to any questions, and continued posting replies to other threads in the coming days.
I pulled all of (his) comments and their post times via the Hacker News API and plotted it, which revealed an interesting lack of posts from midnight-6AM EST across ~150 comments. I'm calling BS on the 0-2 hours of sleep :)
I only sleep about 4 hours a night on average (though I wish I slept more), but if you looked at my posts and comments they would fall within a 4-5 hour window (most posts happen when I check HN between 7-9am).
Ah, is that over the life-time of my being on HN? I've lived in 3 different timezones, US East, US West, East Australia, as well as spending a few months in Europe.
So, though it looks like I'm spending time on HN all through the day, this is not actually reflective of my behavior without taking that into consideration :)
Hope you didn't spend too much time on that, it's cool to see. Thanks.
Nice job! I am very skeptical of people who claim to not need much sleep.
We know there have been some who really needed much less sleep but they are widely known to be geniuses or hyper-productive (Nikola Tesla, Elon Musk etc.)
Most people who claim to sleep only 2 hours a day or whatever are probably being a little dishonest or are miscounting.
I had a teacher in high school that was like this. He got a job as a night checker at the grocery store to help pass the time and make a little extra money. He felt great when getting about 4 hours of sleep. He had researched it quite a bit and after talking to a doctor decided to embrace the extra time. I was always jealous.
But yeah I don't buy the 2 hour sleep thing, because according to this teacher it's almost always around 4 hours for people in this boat.
Nice work. I tested with user pg and noticed that the Y-axis on submissions starts at 31/12/1969 and the comments starts prior to 04/09/1982, compressing the actual data into the right of the graph.
Second, the username on your app is case sensitive, not sure if it is on HN (haven't tested).
Funny website. Shows a cookie notice banner but sets no cookies (except if you click on the cookie notice button - then they set a cookie accepted cookie)
Also: if you are located in Berlin you should create a Privacy Policy and Imprint. That's more important than a cookie banner ...
Nice!
for my username it shows a single green dot in the center of the chart, the infamous "introverted person" distribution (also known as "insecure commenter" distribution.
EDIT: obviously, this comment broke the "curve". This distribution is known for being unstable.
Hm not sure. Basically, I think Submissions graph is cool because the Y axis is points and it's fun to see your most popular posts. I'd like to see that same visualization applied to the comments graph (so replacing replies with points as the Y axis). Sometimes you can have a really popular comment that doesn't get many replies because people agree with it. Replies feels more like an indicator of controversial-ness, which are posts I'm probably not going to look as fondly back on.
Genuine question: where does this fall with regards to GDPR? If you aggregate data about a European, don’t you technically need to ask them for permission first?
Of course I would promote that idea, but really I'd be willing to bet it looks something like a pareto distribution, and it would be interesting to see what puts a user on one side of that curve or the the other.
patio11's 1.39 average reply rate would be an outlier, and there are people whose comments have the authority to simply close discussions that would need upvote stats to distinguish them might get lost viewed in that narrow dimension, but for all the time people spend on this site, it would be interesting to get a sense of what that might mean.