[pjc50]

Silent ignorance is fine. It's the loudly wrong people that are the problem. Wrong comments take more effort to rebut than they do to post; get enough of them and the SNR drops too low, because often the rebuttal is also "noise" - most people don't need a comment reminding them that global warming exists, for example.

[yanjuk]

My assumption is that commenting has the effect of reducing ignorance, even where ignorant comments are plentiful i.e. silent ignorance is more deadly than mostly-ignorant noise. This is because the very act of writing a comment, be it laden with misconceptions, causes its author to think, at least a little. Also reading different points of view helps people to identify common misconceptions and to make up their minds.

(This doesn't include disruptive comments like fluff, complaining and gratuitous humour.)

[nickpsecurity]

"My assumption is that commenting has the effect of reducing ignorance, even where ignorant comments are plentiful i.e. silent ignorance is more deadly than mostly-ignorant noise."

What little, empirical evidence we have on these things shows the opposite: misinformation spreads more rapidly with less likelihood of being countered than accurate information. Facebook and others have done studies to show that. It would probably take a bit of time to go into the components of that but people (a) make spot judgments more than they do solid research; (b) prefer stuff compatible with their existing beliefs; (c) get those beliefs from biased sources; and (d) often attack people more as accountability goes down. This combo, which oversimplifies things but captures quite a bit, will create and reinforce echo chambers quite naturally if it's both easy to comment and vote.

And we see it everywhere from HN to Reddit to political discussions about U.S. election. People presenting or voting in a balanced way giving disagreeing parties a chance are a tiny minority in overall discourse. Reputation and moderation systems should probably be designed with that as an assumption built-in.

[grzm]

"[M]isinformation spreads more rapidly with less likelihood of being countered than accurate information. Facebook and others have done studies to show that."

Ouch. As much as I'd like to imagine this isn't the case, my gut tells me it's true.

If you happen to have references to such studies at hand, would you mind providing them? If not, no big deal. I'm sure I can find them with a little searching.

"Reputation and moderation systems should probably be designed with that as an assumption built-in."

I agree. There are two pieces to this as well: (a) discouraging undesirable behavior and (b) encouraging desirable behavior. Reputation systems are harder to implement as well if you allow for anonymous behavior, which I think there is a place for. What does reputation mean if it's not something that sticks with an individual?

[nickpsecurity]

Here is Facebook's that I still have from earlier conversation:

http://www.iflscience.com/technology/facebook-echo-chambers-...

Far as your Googling, cross-reference the words study or empirical with "echo chamber," "filter bubble," and "misinformation." That should get you best results.

[grzm]

Thanks!

[yanjuk]

I don't think 'doing solid research' is what comments are about or should be about: that's what the articles are for. Comments are more like what goes on with coffee after the research presentation. They're about saying what you think, asking questions, correcting other comments, correcting the article, pointing to other articles. The mere act of writing causes ideas to be clarified.

The increasing majority of us don't expect any but a fraction of comments to be good. (This applies to research articles too. Sturgeon's Law.) They are like unrefined ore. Which is why we need as many as possible. Misconceptions are dangerous but they are less dangerous when brought into the light where they can be identified and corrected.

It's true that misinformation and conspiracy theories spread rapidly but, crucially, so does the correction of error, especially older and more parochial errors, which may even need to spread a bit before this can happen.

Look at places where people don't have access to the internet and you won't find an abundance of true information. Rather you'll find much less information together with cruder and more parochial misinformation such as ideas about witches and evil spirits.

So I don't think we need to worry too much about the quality of information in the comment section or any frontier of knowledge. It's mostly wrong. What is of more concern is how to guard and improve the conditions under which knowledge grows, beyond what we've already learnt (such as expelling trolls, discouraging politics, fluff).

[nickpsecurity]

" Comments are more like what goes on with coffee after the research presentation. They're about saying what you think, asking questions, correcting other comments, correcting the article, pointing to other articles. "

I would like that to be the default approach to how comments are handled. It's pretty much what we're doing here. On a lot of social media especially, it's done differently where they're spreading what they believe is the truth to get others to believe it. It's often reposted by many of those others. Further, I've seen polls showing most people on left and right are getting their news from social media and websites these days. These comments and shares reinforce the echo chambers they're in where they hit them before the corporate media does.

"It's true that misinformation and conspiracy theories spread rapidly but, crucially, so does the correction of error, especially older and more parochial errors, which may even need to spread a bit before this can happen."

Facebook's studies found the opposite: misinformation spreads far and wide while retractions happened for under half of it. I think it was in 30-40% range. Misinformation outperforms corrections on the Internet. That wasn't as much a problem in the early days but not people are trusting misinformation more because it often looks legit and comes from friends. Instead of lies, it's often just selective presentation of the truth where other information is filtered out. That's why it's more effective.