What's the difference between "good" bots and "bad" bots? And why doesn't Snapchat have a fake news problem? Here's a quick guide to some of the discussions that might come up during the hearings this week.
More than half of internet traffic is bots. Bots have always played a major role in our internet ecosystem, although not all bots are bad. (Some, for example, are used to make our search experiences more accurate.) But the bots used to spread fake news are usually bad, and bad bots make up roughly 29% of internet traffic.
Accessibility attracts bots and fake accounts: Google, Facebook and Twitter want to make it easy for users all over the world to get on their platforms, because they believe in free speech and open access. But this level of openness means the barrier to entry on these platforms isn't just low for users, but for bots and bad actors as well.
- You'll notice that there aren't many fake accounts or bots on Snapchat, because tight networks of friends interact ephemerally, rather than a broadcast network built on friend discoverability and links that last longer than 24 hours. According to a Snap spokesman, this means the economic payoff is low for spamming, phishing, or circulating images or links via bots compared to the effort required.
Bots are programmed to perform simple internet tasks repeatedly: You can program a bot to like, share, or comment on something. Fake news perpetrators create fake stories that are often amplified by a network of bots that automatically like, share or comment on the content. Algorithms elevate content that is popular, further amplifying the effect.
The Internet Research Agency is the source of many Russian bots: It employs a large staff to spread fake news and disinformation and has been using bots to spread Russian propaganda for years.
Bots don't just spread fake news — they can create it. Distil Networks, a cybersecurity company that focuses on bot detection and mitigation, says it's continually warning its digital publishing clients about ways bad bots are used to skew online polls.
- "Online polls have almost no verification to an individual," says Chief Product and Strategy Officer Rami Essaid. "If you go in an online survey, there's not an individualized SSN, or license. They look at people based on cookie's or IP addresses that can easily be mimicked through bots by bad actors."