Twitter followers are under scrutiny lately, due in large part to StatusPeople’s Faker App. The app samples a Twitter account and returns “fake” and “inactive” estimates based upon a certain criteria. Unfortunately, some articles have overlooked the limitations of the app and made bold, incorrect assertions about certain (large) Twitter accounts.
The press reports on the President’s fake Twitter followers
It was widely reported last week that 70% of President Obama’s Twitter followers are fake.
The genesis of these reports appears to be the New York Times. In the article, “Buying their way to Twitter fame,” the NYT uses StatusPeople’s suddenly trendy Faker App to determine this number. The problem in this statement is that they don’t acknowledge (or even explore) the constraints of the tool they’re using. It’s particularly problematic because the accuracy of the tool decreases as follower size increases, and Obama has a pretty big following (to the tune of 18.8 million).
StatusPeople’s methodology has improved…. but not enough for the big dogs
Rob Waller of StatusPeople posted a couple of days ago that their Faker App has increased the population of followers that they draw from. Here is a quick snapshot of their methodology:
Before: assessed a random 10% of the last 5,000 followers
Now: assesses a random 1% sample of the last 100,000 followers
This may not seem like a big difference, but it is. Where the previous methodology was only accurate for the most recent followers, this takes a more representative sampling from most accounts.
Consider the case of President Obama, though. With 18.8 million followers, assessing 1000 followers represents .005% of his followers. Sampling only the most recent 100K makes this an unrepresentative sample of his followers, especially given the fact that bots target and follow the biggest accounts on Twitter to mask their behavior from filters. Since the likelihood for bots to be filtered increases through time and the likelihood to be followed by bots increases with follower size, it stands to reckon that there would be a lower percentage of fake followers in the 18.7 million followers that aren’t assessed.
What is the relevance of fake Twitter followers for the President?
Due to the relative newness of social media’s incorporation into political consciousness and how immediately politicians need to gain traction on a national stage, I suspect that nearly every politician (Barack Obama included) has purchased followers. I don’t see this as controversial or even relevant.
The relevance of Twitter and other social networks in politics is how real followers are leveraged to take action. Hopefully Obama and Romney have more informative metrics than follower count to measure that.