Facebook and YouTube reveal their offensiveness offensive and intolerance of intolerance

Facebook and Google-owned YouTube have both attempted to offer some transparency when it comes to one of today’s haziest trends: abusive, offensive or ‘inappropriate’ online content.

While these are firms which have hardly helped themselves over the last couple of years considering the almost allergic reactions to transparency, you do have to have a bit of sympathy; being a mediator of appropriate content is a tight-rope walk. If you come down too tough it would be seen as being overly sensitive and limiting freedom of speech, while too far the other direction can lead to loss of advertisers, a PR disaster and more ammunition for self-righteous politicians.

Starting with Facebook, the team has decided to unveil its policies towards offensive content in its continued damage-limitation quest following the Cambridge Analytica scandal. Employing a content policy team, which Facebook describes as having subject matter experts in numerous fields including hate speech, child safety and terrorism, is a good starting point and sets the standards. These individuals will consult external experts to understand the development of social norms and language to adapt these policies in time, and hopefully find the sweet spot which keeps everyone happy.

This is the big issue which we see with the policy; there is no such thing as policy which will keep everyone happy. Some countries like guns, some don’t. Some treat women as equal, some are based in the stone-age when it comes to enlightenment. Some embrace national stereotypes, while others find nicknames offensive. The world is regionalised, as are the people who live on it. Facebook has been forced by rule makers and politicians into finding a solution, of which there isn’t one, unless you want social media to turn into a beige exchange of people discussing the weather.

But as the social media giant states, policy is only as good as the enforcement. Reports of offensive content will be reviewed by the Community Operations team, there are currently 7,500 content reviewers and the additional help of artificial intelligence, before action is taken. That said, should a post be removed, the user now has the right to appeal. This certainly sounds like an incredibly democratic way to approach the situation, but again, it leads to nuances which could send the team further down the rabbit hole.

The person who posted the deemed offensive content may not have been offended by it in the first place. This is where it becomes tricky. Obviously there are examples which are clearly offensive and inappropriate, but there will be examples where content is offensive to some and not others. Take guns for instance. In the US, posing with an instrument of death would not be deemed offensive, but in the UK, parents might be concerned that if their children see the image, weapons and violence become normalised in their minds. These are two seemingly similar countries, which have quite different opinions on such content. Who decides who is right and who is wrong?

While Facebook has tried to demonstrate its newly found principles of transparency, YouTube has been less nuanced, simply posted figures from the last quarter on how many videos have been removed from the platform due to policy violations. The report has been released at the same time as the Google financials, so we can expect this report once a quarter.

Over the last quarter, 8 million videos were removed from the platform, the majority of which were spam or people attempting to upload adult content. 6.7 million of these videos were first flagged for review by machines rather than humans, with 76% being removed before a single view. Removing the video before it has been viewed is proving to be a successful exercise for YouTube, as you can see below, which shows how videos depicting violent extremism were removed.

The promise of artificial intelligence in this space is probably the only way in which the problem is going to be dealt with, and it does seem YouTube is getting a good handle on the situation. You do have to bear in mind there are far less posts to be reviewed on YouTube compared to Facebook, though being able to lean on the expertise of the Google AI team will certainly help here.

While the machine learning technologies developed by YouTube are able to remove videos, it is worth noting the majority make their way to content reviewers here as well. Firstly the video is reviewed, but the team also look at the metadata such as the title, description or tags. The reviewer will also be able to identify the context, identifying whether the purpose of the content is educational, documentary, scientific or artistic.

This is where YouTube is making a stance against the PC-army. Should the video fall into one of these categories, it will generally be left up. People have the right to be offended, but it doesn’t mean what they are offended by is offensive. Some of these videos might be hidden behind an age restriction, or in some cases nothing is done. A good example would be those who do not believe in Darwin’s Theory of Evolution. Just because there are some schools who believe in creationism and are offended by these videos online, does not mean everyone would be. This element of YouTube’s policy is something we like; the team seem to realise there is no such thing as a perfect solution for everyone, but are taking appropriate steps to make sensible and logical decisions.

We should never give the internet giants too much credit, as most have shown themselves to be mistrusting and misleading in one area or another, but tackling offensive content online is a very difficult task. There is no such thing as a perfect policy which covers everyone around the world, and we wonder whether Facebook’s search for this silver bullet could land it in a worse position than it is in now.