Why Princeton researchers don’t know jack about Facebook

“Facebook is about to lose 80% of its users, study says” Time, January 21, 2014.

“Facebook will lose 80% of users by 2017, say Princeton researchers.” The Guardian, January 22, 2014

We should always be wary of exciting headlines; too often they mean that the journalist knows that there’s no story.

The study in question, ‘Epidemiological modeling of online social network dynamics‘ is interesting enough, and the idea of using epidemiological models to analyse social trends is extremely fertile.

But the data used by authors Canarella and Spechler (a pair of grad students, neither of whom is an epidemiologist) are bad. Here’s their description of what they did:

The proposed infectious recovery SIR model (irSIR model) is validated using publicly available Google search query data for “MySpace” as a case study of an OSN that has exhibited both adoption and abandonment phases. The irSIR model is then applied to search query data for “Facebook,” which is just beginning to show the onset of an abandonment phase.

They used Google Trends data. Here’s the trend for MySpace:

myspace

And here’s the trend for Facebook:

facebook

And here’s their re-based chart:

epidemiology

You can see why they might think they were onto something, right?

Wrong! They’re missing an important data point. According to October’s earnings call, 48% of Facebook’s Daily Active Users (DAUs) are accessing only from mobile.

Google – as one blogger points out – is still the luddite’s address bar. That is, people still search Google for Facebook rather than typing “facebook.com” into their browser. I think it’s unfair to call this audience “luddites” — they’re still the majority.

But when they’re on their smartphones, they just tap the home screen icon for Facebook. They don’t need to search. So roughly half of all Facebook’s DAUs have stopped searching Google for Facebook. That’s where those missing searches have gone.

Oh, and by the way, here’s the Google Trend for Twitter:

twitter

Update

Facebook responded with an hilarious post, ‘Debunking Princeton’. Go read it. I won’t spoil it here.

Comments

  1. says

    I would also add that this doesn't take into account that Facebook can now go in an entirely new direction – a complete innovation. If you have over half of all online citizens on your service then why keep it for just sharing cat videos? I doubt that we will see a repeat of Myspace just because of the network effect.

  2. Nigel Shardlow says

    It's not a serious paper is it? It certainly doesn't look like a serious model. The authors are PhD students. Just a bit of fun. The joke's on the journalists who picked it up, shame on them.

  3. says

    And doesn't this highlight the problem with using the same, imperfect measure (e.g. the number of times a term is searched is hardly representative of what you actually want to know) over a long period of time in a fast developing area like the internet – people's behaviour and habits are likely to change making these measures inconsistent over long time periods.

  4. says

    Alex Barr yes, if we considered Facebook as only a social network and analogous to myspace then we could use history to plan where it will go, but I think how we view services like this will change over the next few years.

  5. says

    Spechler's research areas are laser processing, energy storage, while Cannarella's is energy storage. I'd agree that it's not a serious paper; but I'd strongly suggest that they're trolling the press here.

    You're right, though, my headline is misleading. I'm sure Princeton knows plenty about Facebook.

Trackbacks

Please tell me what you think.