YouTube has released its first quarterly Community Guidelines Enforcement Report and launched a Reporting Dashboard that lets users see the status of videos they’ve flagged for review. The inaugural report, which covers the last quarter of 2017, follows up on a promise YouTube made in December to give users more transparency into how it handles abuse and decides what videos will be removed.
“This regular update will help show the progress we’re making in removing violative content from our platform,” the company said in a post on its official blog. “By the end of the year, we plan to refine our reporting systems and add additional data, including data on comments, speed or removal and policy removal reasons.”
But the report is unlikely to quell complaints from people who believe YouTube’s rules are haphazardly applied in an effort to appease advertisers upset their commercials had played before videos with violent extremist content. The issue came to the forefront last year after a report by The Times, but many content creators say YouTube’s updated policies have made it very difficult to monetize on the platform, even though their videos don’t violate its rules.
YouTube, however, claims that its anti-abuse machine learning algorithm, which it relies on to monitor and handle potential violations at scale, is “paying off across high-risk, low-volume areas (like violent extremism) and in high-volume areas (like spam).”
Its report says that YouTube removed 8.2 million videos during the last quarter of 2017, most of which were spam or contained adult content. Of that number, 6.7 million were automatically flagged by its anti-abuse algorithms first.
Of the videos reported by a person, 1.1 million were flagged by a member of YouTube’s Trusted Flagger program, which includes individuals, government agencies and NGOs that have received training from the platform’s Trust & Safety and Public Policy teams.
YouTube’s report positions views a video received before being removed as a benchmark for the success of its anti-abuse measures. At the beginning of 2017, 8% of videos removed for violent extremist content were taken down before clocking 10 views. After YouTube started using its machine-learning algorithms in June 2017, however, it says that percentage increased to more than 50% (in a footnote, YouTube clarified that this data does not include videos that were automatically and flagged before they could be published and therefore received no views). From October to December, 75.9% of all automatically flagged videos on the platform were removed before they received any views.
During that same period, 9.3 million videos were flagged by people, with nearly 95% coming from YouTube users and the rest from its Trusted Flagger program and government agencies or NGOs. People can select a reason when they flag a video. Most were flagged for sexual content (30.1%) or spam (26.4%).
Last year, YouTube said it wanted to increase the number of people “working to address violative content” to 10,000 across Google by the end of 2018. Now it says it has almost reached that goal and also hired more full-time anti-abuse experts and expanded their regional teams. It also claims that the addition of machine-learning algorithms enables more people to review videos.
In its report, YouTube gave more information about how those algorithms work.
“With respect to the automated systems that detect extremist content, our teams have manually reviewed over two million videos to provide large volumes of training examples, which improve the machine learning flagging technology,” it said, adding that it has started applying that technology to other content violations as well.
— Jake Snow (@snowjake) April 23, 2018
YouTube’s report may not ameliorate the concerns of content creators who saw their revenue drop during what they refer to as the “Adpocalpyse” or help them figure out how to monetize successfully again. On the other hand, it is a victory for people, including free speech activists, who have called for social media platforms to be more transparent about how they handle flagged content and policy violations, and may put more pressure on Facebook and Twitter.