UC Berkeley juniors Ash Bhat and Rohan Phadte publicly launched a Google Chrome extension Tuesday that allows Twitter users to identify accounts run by automated programs, or bots.
The extension, called Botcheck.me, monitors more than 100 characteristics of each Twitter account, using factors such as when the user joined Twitter, the time between each tweet and the topics the user tweets about. Downloading the Chrome extension places a button next to each Twitter account, which users can click to see Botcheck.me’s prediction of whether an account is a bot or a human.
Bhat and Phadte run RoBhat Labs out of their shared apartment in Berkeley, where they originally developed technology to identify the spread of “fake news” on Facebook. When they tried to apply this same algorithm to Twitter accounts, however, it yielded strange results: Many accounts frequently tweeted political posts that seemed “non-human,” according to Phadte. These accounts, often created less than a month before, had thousands of tweets and followers, tweeting as often as every minute.
“As a society, (we) use social networks like Twitter to engage and connect with others,” Phadte said in an email. “We wanted our tool to help people on Twitter to be more confident that you’re engaging with real, live people with important opinions on political topics, as opposed to an automated bot that amplifies polarizing opinions and falsified information.”
Recently, Twitter has received backlash for its lack of protections against bot accounts, with an investigation into Russian interference in the 2016 presidential election revealing that many accounts run by Russian bots were spreading both far-left and far-right propaganda. Bhat and Phadte said they do not know why Twitter has yet to develop bot-detecting technology.
“They said something along the lines of that they don’t have a good way of doing this, which definitely raised some eyebrows on our side, given that we have way less data than they did and we have created a pretty good model,” Bhat said.
Ken Goldberg, campus professor of engineering, said he believes Twitter has not made a strong effort to prevent bots because fake accounts benefit the company.
“They’re being judged by how many people are online and how many tweets are going out, so you can see why there’s no incentive for them to stop the bots,” Goldberg said.
In the 24 hours since its launch, about 1,700 users have already applied the extension to more than 3,200 Twitter accounts, according to Bhat. He said the next step for RoBhat Labs is getting in touch with someone from Twitter about the extension. Phadte added that he’s interested in applying the model to other social media platforms.
“Something we’ve been trying to encourage in our students is being socially aware and helping broader causes,” Goldberg said. “These two are really exemplary because they’ve put it into practice.”