The big problem I see here is that longer comments are much more likely to include quotes and other contextual information that makes them seem 'better' compared to shorter comments. The author mentions this, but underestimates its importance.
For example, using HN karma as an approximate value, I left a one-line comment [1] that seems uninteresting by itself, but added significant information to the discussion, compared to another comment [2] that is very long. The thing is, it's just a gigantic quote. I think that these two cases are the norm, not outliers.
On the other hand, I think it would be very interesting (and more accurate) to compare comment length and karma value on HN.
Yes, using ratings that are made in the context the comment originally appears would probably be better. /. is another site which could be used as source material for such an analysis.
Whenever I write a long comment here I just know most of the people lose interest after the first paragraph. And it's because that's what I do when I try to tackle a long comment.
I have the feeling that they think I am a jackass, or douchebag, writing a bunch of mangled together crap of a response/rant because it's usually the most controversial comments that provoke such a response from me. And the response usually doesn't make much sense or have a point when I read it over. It just ends up being long, repetitive, and I forget what I was talking about by the end of it.
As great as long comments are, I always get the feeling no one reads them. maybe HN could implement an "I read this commment" button.
Interestingly that's actually the only reason why I write long comment responses. It's just that after the fact, presumably no one considered the comment and that I got through to no one.
So what about demanding a minimum amount of characters? In reference to a certain well-known internet service i'd propose to accept only comments with more than 140 characters.
I have nothing against short and witty comments but I too have found that when a person takes the time to write a long comment, he or she usually has something intelligent or interesting to say. It seems unlikely that an unintelligent individual will write a very long comment because he or she wont posses the knowledge to captivate the audience or have the ability to think of something long and interesting to say. Nevertheless, I occasionally get surprised by some of the long comments I see.
Well there are comments that are long and interesting and comments that are just long. And from my experience I've seen about equal amounts of both. However to be honest it's the short and witty comments that are always fun to read or funny.
Nice article. It seems a sad reflection of the baseline statistical knowledge that he had to write an entire paragraph distancing his readers from inferring a causal relationship though.
This is actually a very interesting analysis. He examines a whole bunch of cofactors that I wouldn't have even thought of. Comment length, word length, common adverbs, common pronouns, punctuation, all caps, swearing (since when is "douchebag" a swear word).
I wonder what would be revealed by this kind of analysis on the entire hacker news respository. Now, that's a large data set. (No, I'm not volunteering :-)
Yeah, I got some complaints about calling that section "swear words." Some people also complained about "cock" and "suck" and some of the other words. I guess "profanity" would have been a better term to use.
Now that's hilarious!!
a "thanks!" getting a -1 :-)
I was working on a similar thing, i.e. co-relation between the size of comments and their usefulness, 'coz I wanted to limit the chars on the comments in my next app and also the level of nesting. This analysis helped me and so I said - "thanks!"
Firstly, your going to disappoint some poor guy doing research on the size of his penis with your title.
As for the content of your article. Really liked it.
Some of the numbers in tables were a little hard to follow at times.
Other than that, I love these sort of articles which can point out some interesting tip bits of information which you otherwise wouldn't find. ( just assumed maybe )
Not that I can tell. My guess is that 'deleting' a comment actually sets a 'deleted' flag somewhere in the data base rather than actually removing it; but I haven't looked at the code so I'm just speculating.
> Is the idea of negatively voting down a comment ... To encourage the user to delete it?
No. The purpose is to encourage commentary that follows the site's discussion guidelines:
For example, using HN karma as an approximate value, I left a one-line comment [1] that seems uninteresting by itself, but added significant information to the discussion, compared to another comment [2] that is very long. The thing is, it's just a gigantic quote. I think that these two cases are the norm, not outliers.
On the other hand, I think it would be very interesting (and more accurate) to compare comment length and karma value on HN.
[1] http://news.ycombinator.com/item?id=830887
[2] http://news.ycombinator.com/item?id=814591