> If users change their opinions easily, they will be attracted to the extreme ends of the polarity spectrum.
This is an interesting result. In general, I think most would agree it's a good thing to change your opinion based on new information. Yet if their model is correct, doing so leads one away from what the author implies is the ideal neutral position. So, is it a bad thing to change your mind, or do thinking people naturally gravitate away from "ideal" neutrality?
What does it mean to be neutral when you have to make a decision? Take capital punishment, for instance. How can you be neutral about it? You say, well, we are only gonna execute some of them?
You can be in favor of status quo (doing nothing), but it is not neutrality.
What is interesting is that, if we turn this opinion change to the max, there are mathematical models showing that people should converge to the average (i.e. moderate) opinion of the system.
However our model, via the four parameters we can play with, allows for a narrow range of their values to avoid this consensus reaching. And it is in this narrow range that we can fit best observational data from Twitter.
I don't understand how the title follows from the article. The main finding seems to be that low tolerance increases polarization, which is what one would intuitively expect. (And it's actually known as the old folk wisdom, "do not feed the trolls".)
Since polarized system will be less in state of conflict, they seem to conclude that having high tolerance for different opinion means more conflict, and that is worse? Do I read this correctly?
I think the gist is something like "people and news institutions from opposite poles avoid contact with each other when there's polarization, creating partisan echo chambers".
Which seems to reverse the arrow of causation relative to what the headline suggests. Polarization seems to come first, then the news institutions become partisan so that they enter the cozy echo chamber of one side, rather than being exposed to fire from both sides in the no man's land in the middle.
I have a lot of experience with Twitter and I don't think conflict is minimized there. There's definitely a lot of "interaction" between opposing sides. Mostly of the form "look at this stupid and evil thing the other side said, let's reaffirm our correctness and righteousness by mocking and attacking it".
The reason there isn't meaningful interaction between opposing sides is because they both have caricatures of the other side that the other side rejects. A caricatures B, B rejects the caricature, A embraces the caricature all the more because it expresses the real truth that dishonest B refuses to admit, etc.
Let me try to clarify (one of the authors here). The focus of the title is on the news source. It is the news source that tries to avoid conflict in this model. Specifically, at each time step, they see they are getting flagged (attacked, receiving backlash, etc) and they try to minimize it for the next step.
Differently from what the sibling comment says, this is not bred out from a pre-existing polarization. The starting point for users and news sources is to have a normally distributed opinion, where most of the people (and news sources) are actual neutral. Yet, for a certain combination of parameters, this is not enough and will end up with polarization. And unfortunately that combination of parameters is the one compatible with observed data.
The reason why this happens links to a previous study [1]. Even absent polarization, it is the most central news organizations that receive most of the backlash because of some characteristics of the system (e.g. homophily: people naturally tend to be friend with people with similar opinion as them). This causes central neutral sources to spread more, and thus reach parts of the network that heavily disagrees with them.
I will concede that the initial framing of my post is indeed confusing, because it seems that I'm talking about users avoiding conflict, while the paper is about news sources avoiding conflicts. I'll try to do better in the future.
6 comments
[ 0.20 ms ] story [ 30.3 ms ] threadThis is an interesting result. In general, I think most would agree it's a good thing to change your opinion based on new information. Yet if their model is correct, doing so leads one away from what the author implies is the ideal neutral position. So, is it a bad thing to change your mind, or do thinking people naturally gravitate away from "ideal" neutrality?
You can be in favor of status quo (doing nothing), but it is not neutrality.
However our model, via the four parameters we can play with, allows for a narrow range of their values to avoid this consensus reaching. And it is in this narrow range that we can fit best observational data from Twitter.
Since polarized system will be less in state of conflict, they seem to conclude that having high tolerance for different opinion means more conflict, and that is worse? Do I read this correctly?
I think the gist is something like "people and news institutions from opposite poles avoid contact with each other when there's polarization, creating partisan echo chambers".
Which seems to reverse the arrow of causation relative to what the headline suggests. Polarization seems to come first, then the news institutions become partisan so that they enter the cozy echo chamber of one side, rather than being exposed to fire from both sides in the no man's land in the middle.
I have a lot of experience with Twitter and I don't think conflict is minimized there. There's definitely a lot of "interaction" between opposing sides. Mostly of the form "look at this stupid and evil thing the other side said, let's reaffirm our correctness and righteousness by mocking and attacking it".
The reason there isn't meaningful interaction between opposing sides is because they both have caricatures of the other side that the other side rejects. A caricatures B, B rejects the caricature, A embraces the caricature all the more because it expresses the real truth that dishonest B refuses to admit, etc.
Differently from what the sibling comment says, this is not bred out from a pre-existing polarization. The starting point for users and news sources is to have a normally distributed opinion, where most of the people (and news sources) are actual neutral. Yet, for a certain combination of parameters, this is not enough and will end up with polarization. And unfortunately that combination of parameters is the one compatible with observed data.
The reason why this happens links to a previous study [1]. Even absent polarization, it is the most central news organizations that receive most of the backlash because of some characteristics of the system (e.g. homophily: people naturally tend to be friend with people with similar opinion as them). This causes central neutral sources to spread more, and thus reach parts of the network that heavily disagrees with them.
I will concede that the initial framing of my post is indeed confusing, because it seems that I'm talking about users avoiding conflict, while the paper is about news sources avoiding conflicts. I'll try to do better in the future.
[1] https://royalsocietypublishing.org/doi/pdf/10.1098/rsif.2020...