Political scientists have known for a while that we're not so good at knowing our friends' political views; we, instead, tend to just think our friends agree with us much more than they actually do.
Why that's the case is a bit of a mystery. One theory is that we hide disagreement from our friends (or even lie a little about our beliefs) when we're in mixed company (especially when we're in the minority) - so it's only natural that when we disagree with our friends they wouldn't know it. Another explanation is that, regardless of how much information or disinformation we have about our friends, we all suffer from "false consensus bias," the tendency to think that others think like us.
Sharad Goel, Winter Mason and Duncan Watts try to pick apart what may be behind our misperceptions in 2010 paper that surveys Facebook friends, but I'm not sure they solve the mystery. (Although I do have a cold so maybe I'm too foggy to see it.)
They do, however, have tons of cool observations and insights.
After looking at 900 pairs of friends, first off, they find that - as we'd expect - friends agree with each other about 75% of the time, which is 12 percentage points higher than random (since you'd expect a random pair in their sample to agree 63% of the time. They also find - again no surprises - that friends tend to overestimate how much they agree with their friends.
Things get interesting when the authors look at differences in those overestimations. They find two things. One, the more we actually disagree with our friends, the more we overestimate how much we agree. Two, the 'closer' we are to friends (ie the more friends we have in common), the less we overestimate our agreement.
The authors puzzle over why we might see those differences: is it because we actually know more about our close friends? Do we likewise share more information with those who we agree with, so we have a better sense of what they believe?
The puzzle, though, might be solved with two assumptions - one of which the authors point out, the other which they miss (although I may have missed their discussion of it). When it comes to overestimating our agreement with those we disagree with, we might just be dealing with a ceiling effect; if agree with someone on 90% of political issues, I can only overestimate our agreement by 10 percentage points, but with someone I agree with 60% of the time there's lots of room for overestimation.
The other explanation is one the authors point to as well: this could all be about false consensus. The authors find that when we agree in reality, we know it 90% of the time (that is, only 10% of the time do we get it wrong and think we disagree). But when we disagree with someone, we only know it 40% off the time! What that means is that if we consistently guess agreement correctly 90% of the time and disagreement 40% of the time, we're going to "correctly" guess agreement with those we agree with more often.
Thursday, November 3, 2016
Tuesday, November 1, 2016
The shape of virality
Tweets rarely go viral. Again, as Goel et al tell us in the paper I blogged about yesterday, almost all tweets we see are either one shot posts (95%) or first round retweets (3%). But that means 2% are viral-ish, being retweeted at least twice down the chain.
But what do those viral cascades look like? Are they like the spread of the flu, slowly working their way from person to person, infecting a few at at time but eventually hitting large swathes of the population? Or do they spread in bursts, propelled by super-tweeters? And is there something about the inherent tweet-worthiness of a post that makes it more likely to go viral?
Goel takes on these questions, again with Duncan Watts (and adding on Ashton Anderson and Jake Hofman), in an extremely impressive paper that tracks over a billion tweets and simulates diffusion on model networks with 25 million nodes. (Woah.)
First, they find that what we might call tweet cascades are even rarer than stated above. If you consider cascades that have at least 100 retweets, those make up only 0.025% of all initial tweets. (What's less clear is what percent of tweets we see on our wall are initial tweets or retweets. The fact that the authors track 600 million initial tweets and a total of 1.2 billion "adoptions" suggests that half of the tweets we see are re-tweets.)
When they do go viral, to get back to the questions above, they don't look like flu epidemics or like broadcasts - rather a mash of both kinds of cascades. Since pictures will save me using a thousand words:
But what do those viral cascades look like? Are they like the spread of the flu, slowly working their way from person to person, infecting a few at at time but eventually hitting large swathes of the population? Or do they spread in bursts, propelled by super-tweeters? And is there something about the inherent tweet-worthiness of a post that makes it more likely to go viral?
Goel takes on these questions, again with Duncan Watts (and adding on Ashton Anderson and Jake Hofman), in an extremely impressive paper that tracks over a billion tweets and simulates diffusion on model networks with 25 million nodes. (Woah.)
First, they find that what we might call tweet cascades are even rarer than stated above. If you consider cascades that have at least 100 retweets, those make up only 0.025% of all initial tweets. (What's less clear is what percent of tweets we see on our wall are initial tweets or retweets. The fact that the authors track 600 million initial tweets and a total of 1.2 billion "adoptions" suggests that half of the tweets we see are re-tweets.)
When they do go viral, to get back to the questions above, they don't look like flu epidemics or like broadcasts - rather a mash of both kinds of cascades. Since pictures will save me using a thousand words:
The images are in order of their "structural virality" - the most "virally" cascade being in the bottom right corner - but all of them show a combination of both central tweeters broadcasting a tweet and lots of little tweeters passing it along.
A more interesting finding, though, is that there doesn't have to be anything particularly "sticky" about a tweet to see super cascades like the ones above. The authors do some impressive modeling on "scale free" networks (ones that look like Twitter) and find that even if you choose a fixed "stickiness" of tweets (ie the probability that they'll be retweeted) you'll find a similar array of cascades running simulations on those models as you find in reality on Twitter. In other words, whether a tweet turns out to be a dud or be a super-virus could just be a function of randomness. Cool stuff.
Monday, October 31, 2016
Viral social media stories?
When we think of the many ways social media has transformed our information worlds, one specter that comes to mind is that of the viral story - a news event that may be ignored by the lame-stream, but that seeps its way into Twitter or FB, slowly catches on (or flares immediately) and eventually saturates our online social networks.
But, while those events may or may not exist, they are probably exceedingly rare, according to a 2012 paper by Sharad Goel, Duncan Watts and Daniel Goldstein. Those researchers tracked 80,000 Twitter stories to see what the typical "cascade" (wave of retweets started by a single tweet) looked like. Perhaps unsurprisingly, 95% of cascades can't really be called such - they are made up of a single tweet that never gets retweeted.
But, the authors ask, even though most cascades never really happen, might there be enough huge cascades that they end up making up most (or much) of our social media news-stream. Not so. When we do retweet news stories we do so from the source; 60% of retweets aren't branching off from long information cascades, but are simply retweeting the story from its origin.
Of the 80,000 stories the researchers tracked, only about 0.001% make it out past 5 waves of retweets, suggesting that viral news stories are - if anything - extremely rare. The authors suggest instead that what may seem like viral social media stories may, in fact, be spread by the traditional media - and that they saturate our social media walls because everyone is picking up the story from, say, CNN or the Washington Post.
It could be, though, that the researchers' sample was too small to pick up the "mega-cascades" that we imagine are a unique feature of social media. The "Trayvon Martin" stories, which circulate for weeks on social media before being picked up by traditional media may be, truly, 1 in a million - or billion - rather than 1 in 80,000.
Still, the authors' findings make it hard to deny that the vast majority of news stories we see on Twitter are either initial tweets or one-off retweets - and that super cascades are the tiniest sliver of the news we see. Of the stories swirling around social media, almost all are stories users find from outside Twitter and bring into the network for a single exposure. It's show and tell rather than a game of telephone.
That would mean that what really matters for determining what we see in social media is not what people re-post, but rather what they find from the outside media and decide is important enough to post themselves.
But, while those events may or may not exist, they are probably exceedingly rare, according to a 2012 paper by Sharad Goel, Duncan Watts and Daniel Goldstein. Those researchers tracked 80,000 Twitter stories to see what the typical "cascade" (wave of retweets started by a single tweet) looked like. Perhaps unsurprisingly, 95% of cascades can't really be called such - they are made up of a single tweet that never gets retweeted.
But, the authors ask, even though most cascades never really happen, might there be enough huge cascades that they end up making up most (or much) of our social media news-stream. Not so. When we do retweet news stories we do so from the source; 60% of retweets aren't branching off from long information cascades, but are simply retweeting the story from its origin.
Of the 80,000 stories the researchers tracked, only about 0.001% make it out past 5 waves of retweets, suggesting that viral news stories are - if anything - extremely rare. The authors suggest instead that what may seem like viral social media stories may, in fact, be spread by the traditional media - and that they saturate our social media walls because everyone is picking up the story from, say, CNN or the Washington Post.
It could be, though, that the researchers' sample was too small to pick up the "mega-cascades" that we imagine are a unique feature of social media. The "Trayvon Martin" stories, which circulate for weeks on social media before being picked up by traditional media may be, truly, 1 in a million - or billion - rather than 1 in 80,000.
Still, the authors' findings make it hard to deny that the vast majority of news stories we see on Twitter are either initial tweets or one-off retweets - and that super cascades are the tiniest sliver of the news we see. Of the stories swirling around social media, almost all are stories users find from outside Twitter and bring into the network for a single exposure. It's show and tell rather than a game of telephone.
That would mean that what really matters for determining what we see in social media is not what people re-post, but rather what they find from the outside media and decide is important enough to post themselves.
Subscribe to:
Posts (Atom)