Facebook Will Lose 80 Percent of Users by 2017, According to an Infectious Disease Model

The rise and fall of social media can be modeled after infectious diseases, which means that we'll probably be cured of Facebook soon.

via Niall Kennedy/Flickr

We'll all be cured of Facebook soon. At least, that's the conclusion of a new Princeton paper published this month on the Arxiv server that modeled the social network's popularity over time as if it were an infectious illness. The conclusions may sound hard to believe—the scientists predict Facebook will lose 80 percent of its user base by the end of the year—but they sound a lot more probable if you know how they calculated that number.

Epidemiological models have been used for other applications than just maladies. In 2005, a group of American scientists published a paper (PDF) on the way ideas spread, and proposed that ideas spread through a population in a similar fashion as contagious diseases. It turned out they were right: a model used for tracking diseases accurately predicted the way Feynman diagrams had spread through the scientific community.

Both of the aforementioned studies rely on the the SIR model, which has even been used to model zombie outbreaks. Because it's impossible to take into account all of the individual factors in a human population, the model splits up a group of humans into three categories: Susceptible (S), Infectious (I), and Recovered (R), the latter of which can mean cured or deceased. Let's stick to cured in this case, as hopefully people don't die when they leave Facebook. 

Anyway, as the susceptible people are infected, the population of infected individuals increases, enabling them to infect more people. But because no infection is 100 percent efficient, the infected population eventually recovers (either by being cured or dying). Over time, the number of susceptible people falls to zero, with the infected population following. In a graph this process would like this:

Blue is susceptible, green infected, and red recovered

In our Facebook case, it's clear what kind of users fall into what category: 1) the susceptible people who do not have Facebook, such as your aunts and uncles, who would use it if your mom tells them it's great; 2) the infected people like you and I, assuming you clicked on this article on Facebook; and 3) the trendsetting individuals who left Facebook because it took up too much of their time, or because all their aunts were suddenly commenting on their pictures, or simply because Facebook just isn't cool anymore.

Because these scientists are scientists, they couldn't just assume that a model made to predict the spread of the mumps could predict the downfall of a billion dollar company. So they tested their theory on the long deceased MySpace. To do this, they took publicly available Google data covering MySpace's collapse and plotted that against the SIR model. The two graphs turned out surprisingly similar, provided that they made a minor adjustment.

Left the traditional SIR model, on the right the modified version based on MySpace data

The traditional SIR model is built on the assumption that people recover after a certain period, but people who join a social network usually don't expect to leave it after a set number of weeks or months. Instead, user leave when their friends leave, so the scientists assumed that a user who joined the service early stayed for a longer period of time than someone who only recently joined.

The adjustment also suggests that the process of leaving a social network happens in a similar way to people joining it. Someone quits, convinces his friends to quit because of privacy or whatever, and they in their turn do the same. The amount of quitters rises almost exponentially in this way, which means that the downfall is relatively swift. Just like the rise, in fact.

Bye, Facebook

After the scientists verified their hypothesis and modified SIR model for MySpace, they tried it out on Facebook data they collected off Google. And guess what? The model fit the data eerily well, which means it looks like Facebook is on the brink of a massive drop in users. The authors write that, based on an extrapolation of their best fit model into the future, "Facebook will undergo a rapid decline in the coming years, losing 80 percent of its peak user base between 2015 and 2017." That might sound crazy, but if you look at the writing on the wall, it might not be.

Teens are leaving Facebook for alternatives like Instagram and Tumblr. And while once recent high-profile prediction of Facebook's death turned out to be a misreported bunch of nonsense, one of the conclusions of the research—that teens don't think Facebook is cool—has been increasingly supported elsewhere.

Add to that the fact that privacy concerns are driving people away, and the groundswell of an exodus may be forming. And even if users don't actually go through the steps of deactivating their accounts, a social network's utility is based on how many friends you've got using it, and if some friends stop using it daily, more will follow.

Of course, a model that worked for MySpace may not apply to Facebook—for one, MySpace was never as well integrated into the fabric of the web as Facebook is. This all might be little more than scientific pessimism. But it's hard to argue with the theory; if users start leaving in droves, why will the rest stay? If the model holds true, Facebook could be a ghost town in a year or so, populated by aunts and uncles who caught on too late, and others who are incurably ill with the sickness.