The study used a sample of 1,183 people. They read an article about nanotechnology. A control group read a version of the article with comments that were uncivil and included insults such as "If you don't believe that nanotechnology is harmful, you're an idiot!" Others read the same article with comments that did not include rude, insulting and otherwise uncivil comments. The study's authors claim that the rude and uncivil comments made readers of the comments more "polarized," that is, made them "double down" in their views.
A quick terminological note: The media has reported this study as about online trolling (see here, here, here and here, e.g.). However, online trolling as it is commonly defined, isn't simply about using such rude and uncivil language. It is about gaining attention through insincere posts purporting to express some viewpoint but in fact, is meant to instigate an emotional reaction. You can use uncivil language without trolling. To the study's author's credit, they did not use the term "trolling." Science journalists, being what they are, reporting on this story are the one's guilty of such sloppy use of language.
Anyway, back to the substantive portion of the study. The study concluded with:
Online communication and discussion of new topics such as emerging technologies has the potential to enrich public deliberation. Nevertheless, this study’s ﬁndings show that online incivility may impede this democratic goal.This is to stretch their findings to an area that is not supported by their own data. What they actually found was that readers of those comments had stronger views than they did before after reading uncivil comments. That "polarization" (in the context of this study, polarization of subjective risk associated with nanotech) in itself does not show that it is bad for deliberation never mind the democratic process. There is nothing wrong with having strong opinions on some topic. In fact, having them, all else being equal, is a sign of a strong democracy. So they reasoned from the fact that readers' opinions were made stronger to the non sequitur that this may "impede" the "democratic goal."
Instead, what does impede the democratic goal is not strong opinions per se but intractability, i.e., stubborn, persistent opinions despite the presentation of overwhelming counter evidence. Now it may be the case that having strong opinions will make one less likely to change one's opinions in light of such counter evidence but that wasn't what was studied in this study. Furthermore, there are decided counter examples. Scientists, philosophers, and mathematicians and some layman for example have very strong opinions yet to do science, philosophy, mathematics and many things competently means changing one's opinion's in accord with the evidence. So there are people out there who though have strong opinions are far more likely than most to update their views in so far as the evidence warrants it. So it's not at least contradictory that you can both have a strong view yet not be resistant to rational updating of belief. This is obviously not to say all scientists, philosophers, etc are this way. Many stubbornly hold on to outdated views but they are in general far better at conforming their views to the evidence than most.
But even assuming that most people will become more obstinate when their opinions are strengthened in just such a way (unlike scientists, et al), it doesn't follow that this is generally detrimental to public deliberation in all incidents. In fact, it may even be harmful in many cases not to be uncivil.
Why is this? Take the example that there is an issue, say, global warming. There are global warming denialists and those who affirm the existence of anthropocentric global warming. If both sides have strong but obstinate opinions, it is far better for both truth and the world than the alternative that only the denialists have strong opinions. This is because the denialists are wrong and furthermore wrong about something that gravely affect our and our children's well-being. Incendiary language may be the motivating factor to strengthen the views of both sides so that there is more balance. When those who do know better don't have as strong as an opinion as those who don't know better, this creates an imbalance that harms deliberation for those who don't know better will be more obstinate (again, assuming that it will make them obstinate to view change) and may dominate a discussion.
Now where the findings are relevant in the ways the authors suggest is when there is no definitive evidence on some controversial topic (such as nanotech perhaps). Notice that the example I used above of global warming is rather black and white when it comes to the evidence. There is overwhelming evidence for not only anthrocentric global warming that this very well will lead to disastrous consequences (in fact, it already has for large parts of the world) and little evidence contrary. But in some cases, we don't know too much either way. For example, in many unsettled scientific, philosophical, political, issues there are tentative evidence for many different but conflicting views. It would be prudent as the study suggest to be civil in discussions about the veracity of these issues so as to prevent intractability of viewpoints. That much is clearly true. But in much of public discourse, one side is clearly right and the other is clearly wrong. Evolution is true. The earth is not 5,000-6,000 years old. Iraq does not have WMD. Vaccines do not cause autism. Smoking is dangerous to your health. Etc, etc. So in cases where a person's view is so strongly at odds with reality, it may be good for others engaged in deliberation to call a spade a spade; polarization may be what is called for especially when the obstinate, irrational side is overconfident while the side of reason is acting like a (to put it in mildly PC terms) wussy but in less definitive matters and among more reasonable people, it may be far more prudent to remain open minded, skeptical etc and to facilitate this kind of atmosphere, it may, as the study suggests, mean refraining from uncivil behavior.
Furthermore, the study did not study if "uncivil" behavior made the interlocutors more likely to change/update their views in light of new evidence. That would have been far more interesting because it is directly related to reasoned deliberation, the kind of deliberation necessary for a healthy democracy. Instead, the study focused on non participants of the discussion (3rd party "lurkers"). Being called out a fool may or may not make one more tractable to rational debate. Here's what I suspect. I think being insulted online will not make one more or less likely to change one's viewpoint if only one person does the name calling but if there are more than two people doing the name calling at the interlocutor, he or she will be far more likely than not to change their viewpoints. Sometimes it takes a little community effort to get ignorant and intractable people to be more reasonable. Just listening or being more open minded often requires a little peer push which may involve a little incivility to be truly effective. This suspicion is subject of course to empirical evidence and I hope it will be tested someday (maybe it already has but I don't know where the study(ies) is to be found).
I worry that the spread of this message may even have a net detrimental effect on reasoned deliberation. Here's my reasoning. Who likely reads articles about this study? It is more likely that those who are interested in this study are the more educated and a little more reasonable than the average person who tend to have opinions that tend to be stronger than the available evidence warrants. But if those who are more reasonable are made to think that this kind of behavior is detrimental, they are the ones that will curb their future behavior by being more "civil" online while those who are not so reasonable (likely less educated and those not likely to read these kinds of articles) will remain their obstinate and uncivil selves. This creates in imbalance that doesn't seem very beneficial to reasoned deliberation. Sadly, it seems that some the study has already influenced some actions detrimental to public discourse. For example on the respected Popular Science online magazine's comments section, the comment function has been shut off (looks like permanently) and the moderator's reasoning is partly based on the results of this study. It is sad to see that you'd throw the baby out with the bath water. Despite the fact that many comments are low quality, there are sometimes informative posts and they are worth having despite the bad apples. Bad apples often do not spoil the whole bunch. You deal with poor quality comments by doing your job as a moderator, not by complete censorship.