Fake News: Why is it spreading?
We’ve heard a lot lately about fake news recently, with claims that it’s misleading the public and even compromising elections. And despite many people being aware of the problem, it’s not going away. Filippo Menczer from Indiana University has made it his mission to find out. Tom Crawford heard how...
Filippo - Here what we wanted to look at was, even if we assumed people can do a good job recognising high quality information from low quality information, and further, did they preferred to select and share high quality information: how does this then play in the broader dynamics of social media? In other words, is good quality information likely to go viral more than junk and fake news? So we built a model to study that particular question.
Tom - What did you find out then? Was the high quality information more likely to go viral?
Filippo - Well, the short answer is no. We looked at two particular factors in our model. One is how much information is produced which determines the information load that people experience. If a lot of information, if a lot of things are posted then people get a lot of stuff in their feeds and they can’t possibly pay attention to all of it. On the other side, on the consumption side, we modeled how many things people are capable of paying attention to. We have finite attention and all of this information is competing for our limited attention.
When we make realistic assumptions about these quantities, these limits, then what the model shows is that low quality information is just as likely to go viral as high quality information. Statistically we cannot distinguish between the popularity signatures of low and high quality information.
Tom - What about things such as bots? I remember reading something - I think it was Katy Perry had reached 100 million followers on Twitter. But then, actually reading the article in detail, it was saying that up to half of these are believed to be fake accounts or bots. Are they also playing a role in spreading fake news?
Filippo - Yes. People who are running fake news websites are also using bots to amplify the visibility of their posts. Whether to monetise it and make a financial gain through ads or simply to manipulate public opinion. Creating bots, even pretty sophisticated ones that are hard to tell from real humans is relatively simple. We have found that they are extremely effective at pushing fake news to go viral. So they create the impression that many people are paying attention to it, which makes people curious to see what it’s about and so this then creates a loop by which they are able to generate a sort of cascade.
Tom - How can we then try and stop this, or how can we tell if some account is a bot?
Filippo - It is a hard task and it is getting harder because bots are becoming more sophisticated. We’ve been working on machine learning algorithms to detect social bots for a few years now and in the early stages bots were pretty simple. They just automatically tweeted or retweeted, and you could see their patterns, and you could recognise them from long names and numbers and so on.
But now, we find much mores sophisticated bots that are driven by humans. Very often the content itself is generated by a human but instead of being posted on one account it is posted on 1,000 or 10,000 accounts. Then these accounts follow each other and follow other people and respond and reply to humans, so that they create networks that make it quite difficult to detect them.