There's been a lot of talk about fake Twitter users lately. Everyone knows that there are fake users so "reports" and "studies" on the spread of the problem can at least sound plausible. But there are problems.
One recently popular way of discovering how many of a Twitter user's followers are fake, the Fake Follower Check, has been getting some attention, but its methods are sketchy at best.
For one, it only looks at a sample of the followers, 1,000 people at most. That's great for most people even for somewhat popular accounts. But it's not going to cut it for the ones with millions of followers, ironically, precisely the ones that people are "checking" the most.
For example, analyzing Obama's account with the tool reveals
that 28 percent of followers are "fake" and 39 percent are "inactive" leaving just 33 percent as legitimate followers. But the figures are almost certainly not accurate.
Some of the followers may indeed be fake, but not 28 percent, over 5 million of them. What's more, inactive users are a large part of the userbase, plenty of people use Twitter simply to follow what other people are saying, without tweeting anything themselves.