Don't Judge a Hacker News Post by its Comments | Hacker News Reader

Don't Judge a Hacker News Post by its Comments

alexmarchant.com

15 pointsalex_marchant17 comments

I don't get the conclusion here. After determining that the majority in the poll result and the majority in the comments headed in opposite directions:

> I guess I was surprised because I assumed that the comments were more or less representative of the community (probably a naive thought), but they weren’t.

Apparently you also assumed that the poll result is representative of the community. Thus the conclusion that, since the comments look different, they must not represent the community.

But I don't see any good reason to assume that. Perhaps the comments represent the community, and the poll results do not. Or (most likely, I think) neither represents the community. I, for one, care about what shows up on HN, but I tend to ignore polls as silly wastes of time. What if most HNers are like me?

> Apparently you also assumed that the poll result is representative of the community.

Fair enough. The poll results may not be representative either... but the sample size is larger and the effort required to participate is lower, so it's more likely to be representative than comments.

A good point.

But the more I think about this issue, the murkier it gets. You say "more likely to be representative", but I must ask: representative of what? What is the "community", anyway? Is it anyone who takes 30 seconds to make an account and 10 seconds to read & vote? Or is it only the commenters? The former seems awfully broad. The latter is probably weighted toward people with a lot of time on their hands and a serious net addiction. It might also have a hint of elitism.

Another issue: comments can be deleted, but votes cannot be taken back. I thought a couple of the comments were quite convincing -- in both directions. I imagine reading and then commenting is the most common order. What if someone wants to change a vote?

I'm not sure a poll like this is an adequate sample to draw any conclusions from. Regardless of what the HN community at large would prefer as the result of that poll, it doesn't surprise me in the least that the people who are afraid of change, and thus have the most to lose if the Yessers win, are those who speak up. They already feel invested in the current way HN works, and so they're willing to spend the extra bit of effort to try to preserve that. However, people who would like to see reddit-style AMA aren't as invested in that idea, as it's a new one for HN, and thus have less incentive to spend the extra effort and comment about it.

I think you are begging the question here: there is an unstated assumption by your use of the word "sample" that everyone's opinion has equivalent weight as everyone else's. Suppose that each persons vote didn't count for just +1 to any option, but rather added their total karma to an option. Would yes still have won?

What I would be interested in is a demographic analysis of who made comments.

1. Whether the comments are mostly made by people who are frequently adding to discussions.

2. Whether the comments are mostly made by people who have a high level of karma.

Likewise analyses for the votes. OP makes a lot of assumptions that these analyses would answer.

That's not what begging the question means.

I (and likely others) often don't bother to comment when I agree with an article. In cases like that, I have nothing to add.

I was about to jokily reply, "same", but then I vaguely remembered that maybe the community standards discourage low-content comments. As they should. But the emergent result is that people add content by adding disagreement.

That's an interesting theory. Doesn't it stand to reason that if that were the case, then in a situation where low-content comments were acceptable, the same people would post low-content comments which disagreed (e.g. a simple "No.")?

I think it's an inherent property of human society that disagreement is easier to write, more compelling word-for-word and idea-for-idea. War is more interesting than peace, and so on. So I do think that natural tendency will bias online discussions where social pressure against discord are lesser than in person.

I'm not sure it necessarily follows to the extremes, but I do think one could find places where absurdly short disagreements are the norm. 4chan, say.

I spent some time at Reddit in years past, in fact I did two AMAs myself. I think I know the reason for the discrepancy between the votes and the comments.

One, people who post comments belong to a different demographic than those who simply vote -- the difference is in the level of effort, the level of aggressiveness and the willingness to get involved in a subsequent conversation. I think there's a gender difference (women are more likely to vote without commenting, men more likely to comment and not vote) -- bu I emphasize that, with regard to HN, this is a supposition with some observational evidence, but not a solid statistic.

Two, as to AMA threads themselves, when I was the subject of an AMA, the voting and commenting patterns were completely different than an ordinary conversation. After my second AMA with the same result, I finally figured it out -- many of those who posted questions and comments in AMA forums were women interested in meeting someone noteworthy, and who commented and voted in a way that set them apart from other kinds of forums. At Reddit there really is a large difference between the population of commenters at AMA events versus ordinary conversations, and the primary difference is gender.

For those who think this gender distinction is just my overactive imagination, read this AMA forum and judge for yourself:

http://www.reddit.com/r/IAmA/comments/q9qzn/iama_former_nasa...

What this would mean at HN is that an AMA would very likely be very different than one at Reddit, because there are fewer women who post here regularly compared to Reddit. That's too bad, but I am only reporting a fact.

I think if you compare silent votes versus comments anywhere, you will see a difference in views, not all of it is based on gender, but gender is a factor -- many women are more willing to cast a silent vote than post a comment and risk a hostile reply from someone accustomed to talking to men. Note that I didn't say "all women" and I don't know the reasons or how important the difference is, either quantitatively or qualitatively. But it is certainly real.

One deficiency of the voting system is that you can vote for multiple items (try voting for every item on the page) so I would take the voting numbers with a grain of salt.

Always great to see the quantitative bent to my anecdotal observations.

"Interesting idea, but I don’t think it’s that simple. I think the reasons for the AMA style posts are pretty obvious..."

Good point, and not one I had considered - that the reason for more 'negative' comments was that there was more to comment on if you were in the negative.

Part of me not seeing that is that I'm actually not sure the 'positive' reasons are that obvious - it's not self-evident to me that better quality and more technical Q/A would be created by AMAs on HN. But don't take that away from the good point about why the difference may have been created.

As a footnote, you creeped out my co-founder who found himself talking to me across the table one moment, and reading me quoted online the next. A good moment!

> As a footnote, you creeped out my co-founder who found himself talking to me across the table one moment, and reading me quoted online the next. A good moment!

HAHAHA.

> Part of me not seeing that is that I'm actually not sure the 'positive' reasons are that obvious

Definitely subjective. I guess my real idea behind the statement was that people who agree might think they're all agreeing for the same reasons.

I always click on the comments of a post before I go to the article. Primary, the comments give me a good idea of what to expect but also because Hacker News comments load a heck a lot faster than most of the websites that are linked to, so I get the content faster.

The comments are usually more informative than the article anyway.

I wonder if this could be related to Negativity Bias (http://en.wikipedia.org/wiki/Negativity_bias).

I did notice the same thing and found it interesting.

Reply on news.ycombinator.com