Facebook has a team of 60 engineers working on a brain-computer interface that will someday scan your brain a hundred times per second and translate your inner voice into text. Its Building 8 team is researching a way for humans to hear through their skin. And it’s reportedly developing eyeglasses and contact lenses that will bring augmented reality to the masses.
But they still can’t figure out how to prevent users from abusing the platform by uploading and sharing violent videos, such as Steve Stephens’s horrific murder of an elderly man in Cleveland last weekend (it took Facebook three hours to take down the video). And judging by the complications involved in such an effort, and similar attempts to tackle that problem by other social networks, it will take some time, as CEO Mark Zuckerberg acknowledged at F8 this week.
Facebook’s content moderation system relies on users who flag offending posts, which are then passed on to an army of moderators who make the determination of whether or not they’re suitable for the site. Though it has thousands of human moderators, that’s hardly enough to handle the swarm of content uploaded by close to 2 billion monthly active users. That is supplemented by artificial intelligence “to understand more quickly and accurately what is happening across our community,” but that could take years to work effectively.
Other Platforms, Other Solutions
Other social-media platforms, though much smaller than Facebook, have had varying degrees of success tackling the problem. YouNow, a social live-streaming platform that’s primarily targeted at teens, has a fairly robust content moderation system. The site has been around for quite a few years, and given its user demographic it needs to be extra careful with everything it uploads. One thing that YouNow has, which Facebook lacks, is a cohesive user culture and set of values that makes the site a “safe space.” Those who use YouNow are part of a positive community that they don’t want sullied. This was created, in part, thanks to strict content moderation that is quick to take down any abusive content. The founders used a mixture of community participation and technology to try and ensure that only kosher content was presented to the teens’ eyes from the get-go.
Obviously Facebook can’t adopt this solution since it lacks a unifying culture due to its sheer enormity as a global media juggernaut. (For context, last summer YouNow said that it had 150,000 live streams uploaded per day; Facebook announced in 2015 that it hit 8 billion daily video views per day.)
Still, YouNow does employ a very thorough system to make sure that inappropriate content isn’t seen by minors. Like Facebook, users can flag offensive videos. The site also employs proprietary technology to filter out such content–both videos flagged by users as well as those detected via its algorithmic system. It does that partly by analyzing comments in real time, which helps detect content that should be flagged.
YouTube has been dabbling with a similar community-based moderation system. Called “YouTube Heroes,” it empowers an elite group of users to police the site and help weed out the flood of horrific videos that you can imagine gets uploaded every second to the popular platform. Similarly, Reddit has been working to curb its well-known abuse and troll problem by rewriting offending posts and having moderators enforce a new content policy.
Instagram too is known for its heavy-handed content moderation, via such methods as blocking the hashtags for groups and communities that share potentially harmful content. Through this method, it has successfully blocked content that promotes eating disorders. How Instagram decides which hashtags to block remains a mystery. This is something many social media researchers have looked into, but given that hashtags are a prime way for users to connect, these blocks help reduce the flood of potentially abusive content.
On the other end of the spectrum there’s Twitter, which has been plagued with abusive users for years. True, the company has been doubling down on the problem, with a slew of new tools aimed at giving users more power to combat offending content and trolls. But when it comes to live-streaming video, users might have a harder time. According to Periscope’s terms of service:
You understand that by using the Services, you may be exposed to Content that might be offensive, harmful, inaccurate or otherwise inappropriate, or in some cases, postings that have been mislabeled or are otherwise deceptive. All Content is the sole responsibility of the person who originated such Content. We may not monitor or control the Content posted via the Services and, we cannot take responsibility for such Content.
Waiting For AI To Get Smarter
The real question that stands out in this debate is how long it will take artificial intelligence to advance to the point that it’s able to distinguish between a remake of Pulp Fiction filmed by some college freshmen and a real-life murder. YouNow is able to filter out such content through a blend of human moderation and technology, but that becomes much more difficult at the scale of Facebook. But the technology has advanced by leaps and bounds in recent years. Facebook told Wired earlier this week that half of the content it flags comes from its AI program. The Cleveland video fell through the cracks since it wasn’t flagged by humans immediately. Once it was flagged, says Facebook, it took a little over 20 minutes for it to be taken down.
Since such technological advancements could take years to implement, Facebook will have to adopt more human moderation and other methods—including the idea of making it more difficult to download videos to prevent users from quickly sharing them on third-party sites like Live Leak. But maybe it can adopt some of the solutions that have been used by other platforms or even partnering with them to collaborate on technologies that tackle the problem. If Zuckerberg is all about “connecting the world,” maybe one place to start is by partnering with allies in a shared campaign to keep out violent and hateful content.
For now, Facebook will be faced with a choice that cuts to the heart of its purpose: Does it want to create a safer space for users by implementing more rigorous safeguards, or does it favor a free space where there are only limited controls so that users can still upload their content in just seconds? Whatever the company says, it’s going to be years until it can have both.