Rumour is an unofficial story or piece of news that might be true or invented, and quickly spreads from person to person like fire fanned by the wind. Social media posts on Facebook,
and WhatsApp are quicker than traditional ways of communicating, and are capable of creating chaos, disorder and riots at the speed of lightning.
The riots in Britain in August 2011 originated from a few tweets. In recent years, we have heard many allegations on social media for the purpose of creating chaos, lynching, and instigating riots in different parts of India also. During the outbreak of the devastating Ebola virus in 2014, the rumour, “Texas Town Quarantined After Family Of Five Test Positive For The Ebola Virus”, was shared more than 330,000 times on Facebook.
Incorrect social media posts continuously claiming that Ebola can spread through the air, water or food, created an atmosphere of fear. However, sometimes social media is the only way to obtain information. For example, during Hurricane Sandy on the East Coast of the United States in 2012, many New Yorkers had to rely on mobile phones for information owing to power outages.
A mathematical model called SIR (Susceptible-Infected-Recovered) is the most popular method of studying the spread of epidemics. It is also widely used to describe the spread of rumours, because gossip travels at the speed of an infectious disease. According to the model, every person in a population is in one of three conditions: “Susceptible” to becoming infected, “infected”, or “recovered”. A susceptible person may become infected through contact with someone who is already infected (i.e. by becoming exposed to the rumour).
The number of “susceptible people in a specific time unit” is equal to “the number who were susceptible in the previous time unit” minus “the number who become infected within this time unit”. The number of “susceptible” reduces over time.
“The number of infected people at some time unit” equals “the number of those infected in the previous time unit” plus “the number of susceptible people who became infected at this time unit”, minus “the number of infected from the previous time unit who were recovered in between”. The number of infected persons is usually represented by a bell-shaped curve — it increases until it reaches a peak, and more people recover thereafter.
“The number of recovered people in any time unit” is equal to “the previous number of recovered”, plus “the number of infected people from the previous time unit who recovered in this time unit”. The number of recovered people usually grows.
The model assumes that if someone recovers, they get immunised and cannot be infected again. A statement such as “the infection rate is 10”, say, indicates that the (average) number of susceptible people who get infected by one infected person per unit of time (say, one minute) is 10. Also, the reciprocal of recovery rate — i.e. the expected recovery time — is an important parameter. Suppose the expected recovery time for an infected individual is an hour. Then, for a population of 10 lakh, with just one infected person initially (which means that somebody tweeting a rumour or spreading a rumour through a WhatsApp message), almost all the people can get infected within just three hours — although it’s hard to believe the speed with which this can happen. But it takes days to debunk the rumour.
Among the two parameters, the infection rate should be reduced and the recovery rate should be increased in order to slow down the rumour epidemic. WhatsApp has limited the number of chats where a message can be forwarded at a time to five in India, whereas the global limit is 20. Certainly, the possible idea is to reduce the value of the infection rate.
Increasing the value of the recovery rate is certainly not easy — debunking a rumour is admittedly a slow process. The recovery may be expedited through repeated public announcements as well as through appropriate anti-rumour processes. It is important to identify rumours at the earliest. Studies from a group at Warwick University and another at Indiana University have revealed that, on average, it takes more than 12 hours for a false claim to be debunked online. And severe damage might be caused in that time.
Admittedly, real life is definitely much more complicated than the mechanism of such a simple SIR model. Consequently, several modifications of this model have been studied and are still being studied by different researchers. However, this simple model might still help us visualise the dynamics of how rumours and propaganda are spread through social media.
I am sure, though, rumour or no rumour, events like the Arab Spring are uncontrollable.
The writer is professor of statistics, Indian Statistical Institute, Kolkata