Oct 10 2016

Modeling Misinformation

tweetdemicAs social media changes the nature of information and communication, it is interesting to think about how information and misinformation spread through these networks. A paper from 2015 seeks to model the spread of information through social networks to better understand the effect of specific variables.

The study: Fact-checking Effect on Viral Hoaxes: A Model of Misinformation Spread in Social Networks, is not a real-world examination but a mathematical model. As with all such models, it is overly simplistic in order to isolate a few key variables. For the purpose of this model they are treating misinformation as a virus. Someone exposed to the misinformation is potentially infected. At random, they will either be gullible, believe the misinformation and spread it on, or they will be skeptical, fact check the misinformation and then spread correcting information.

In the model they can vary the percentage of people who are believers vs fact checkers. They can also vary the time it takes on average for people to forget the information and therefore become susceptible to infection again. They used a stochastic model, which means the spread of the information and whether or not an individual is a believer or fact checker was random.

They found, not surprisingly, that the greater the percentage of fact checkers, the more likely it is that the hoax will essentially be wiped out. If the percentage of fact checkers is too low then belief in the hoax can become endemic – it can persist indefinitely in the population.

Interestingly, the rate at which the misinformation spread did not matter in this model.

Essentially what this model shows mathematically is that the presence of fact checkers in the network of information spread serves to slow the spread of misinformation, and if their activity is sufficient they can actually eradicate misinformation and keep it from becoming endemic. The critical factor is simply the percentage of fact checkers.

It is interesting to think about how this simplistic model differs from reality, or in other words, what is the structure of the networks through which information spreads in the real world?

My first thought is that the spread of information is not stochastic. People have networks of connections and places where they get their information. Further, social media outlets increasingly “personalize” the information that is fed to its users. If you look at a website catering to believers in conspiracy theories, you will be fed news items related to conspiracy theories.

In the extreme these effects result in “echo chambers” in which people are exposed extensively through their networks to information that caters to a specific ideology or belief. If you are an anti-vaxer you are fed a constant diet of anti-vaxer propaganda, to the point where you are living in an alternate anti-vaxer reality.

Mature echo chambers have their own experts, their own facts, and their own interpretation of reality that is constantly reinforced by the community.

Every community, including skeptics, creates an echo chamber to some degree. Obviously different communities have different relationships with reality – with transparency, legitimate expertise, and dedication to a valid process of evaluating information. That will get to the quality of the information more than the nature of its spread. I do think that the degree to which members look outside their community for information and perspective is a feature that varies among communities.

In reality I think people are exposed to various sources of general information: Echo chamber – information that reflects a specific ideology or community; curated news – information that has been filtered for their preferences to some degree, or that is biased toward their ideology (like a conservative or liberal news outlet); general news – information that is independent from their existing belief and that is just out there in the general public; and personal network – family and close friends that will expose them to information.

Ideological communities serve as an important source of information and fact checking, and then spread that information to the general public. There are obviously also nodes that are hyperconnected – popular writers or outlets that serve to significantly amplify and spread information more than a random person.

To some degree there is competition among believer and skeptical communities, acting through their nodes, to spread their perspective to the general public. Here numbers and activity level matter.

Another variable that the authors of the model discuss as a potential for future research is the notion that believers and fact checkers may not address an issue at the same time. In other words, a believer community may become active on a certain issue before the skeptical community gets wind of it. The belief therefore spreads unchecked for a time before fact checking then starts to pull it back.


This model is obviously very limited in terms of describing what is actually happening in the world, which the authors acknowledge. It does demonstrate mathematically what I think intuition tells us – fact checkers need to be out there correcting misinformation in order to limit and possibly stop the spread of misinformation, at least to the general public. There are likely to always be susceptible populations where the misinformation will become endemic because the hosts are especially receptive.

Many people have been lamenting recently that we appear to be living in a post-fact world. The echo chambers seem to be winning. It is amazing to witness the normalizing of just lying and blatantly making up your own facts to suit your ideological needs.

However, there is also a backlash against this behavior. Perhaps we are witnessing the formation of a new ideological divide – between those who still have some respect for objective facts and a valid process, and those who are more free to make up or believe whatever supports their ideology.

The model does reinforce the critical role for activist skeptics and also good journalism and science communication. We need to nurture networks of fact checkers and skeptics. We need to inoculate the public against misinformation by teaching critical thinking. We need to connect skeptical outlets to general outlets as much as possible.

We may also need to rethink the current model of heavily curating information to the individual. It may be too late for that, but there seems to be a role for high quality, transparent, fact-checked general news outlets that are not tailored to any particular ideology, and represent a source of shared valid information. There needs to be a common set of fact on which we can all agree. Without that as a starting point, constructive discussions are impossible.

8 responses so far