In securities filings, Twitter says it believes fake accounts represent fewer than 5% of its 230 million active users. Independent researchers believe the number is higher.
Italian security researchers Andrea Stroppa and Carlo De Micheli say they found 20 million fake accounts for sale on Twitter this summer. That would amount to nearly 9% of Twitter’s monthly active users. The Italian researchers also found software for sale that allows spammers to create unlimited fake accounts. The researchers decoded robot-programming software to reveal how easy it is for spammers to control the convincing fakes…
Jason Ding, a researcher at Barracuda Labs who has studied fake Twitter followers for more than a year, also thinks Twitter underestimates the prevalence of fake accounts on the network. Mr. Ding says users don’t understand how active and realistic the fakes can appear.
Read on for more details how the battle between the black market and Twitter’s use of algorithms to discover fake accounts is going. Even if the average user can’t quite figure out who is a real or fake user, the consequences are real:
The fake accounts remain a cloud over Twitter Inc. in the wake of its successful initial public offering. “Twitter is where many people get news,” says Sherry Turkle, director of the MIT Initiative on Technology and Self. “If what is trending on Twitter is being faked by robots, people need to know that. This will and should undermine trust.”
According to this article and others, it appears that fake accounts are most commonly used for promotional purposes, whether for Washington politicians or entertainment stars. How harmful are these fake accounts which might be used to boost the number of followers or retweet material?
On the other hand, Turkle suggests these fake accounts could easily mask what is really happening on Twitter. Perhaps they are pushing certain Twitter trends, which then influences other users. Or, perhaps these fake Twitter accounts could push false news reports, which could have some different consequences depending on the situation. It could be worse if a large number of users find out they were interacting with or trying to engage with fake accounts.
While I agree with Turkle that this does present an important trust issue, I wonder if it would take some high profile case before this becomes a real issue. Imagine someone is able to use a set of fake accounts to pull off a terrorist act or throw off the government.