The need for website content moderation and how to do it
Social media is one of the realities of the modern world. Like flying and call centers and crowded cities. It is there. It is growing. It is used by millions, sorry billions of people around the world. And it is open. It thrives on what is known as user generated content (UGC), content that you and I, and billions of others, create or copy-paste and share on our accounts and with our networks. Content that can easily be further shared by the many people who get access to it from us. In fact, social media is seen as the ticket to instant stardom by many people who keep trying to create content that they hope will catch the fancy of people and will ‘go viral.’
Social media platforms are websites, or web properties.
Many other websites, that are not necessarily designed as platforms to encourage interaction between people, also invite participation from visitors and users and customers.
Many organizations leverage the power of social media platforms and make an effort to carve out spaces for themselves within the overall umbrella of the platform, where they can promote interaction on themes that may be of interest to their organization and brand.
The need for website content moderation
What this results in is creation of content in gargantuan, impossibly large, quantities. Imagine 4.6 billion people (the number of internet users in January 2021, according to Statista) around the world, almost 60% of the world’s population, busy creating content.
Someone is uploading pictures of her last vacation in Scotland, someone is being wished on his birthday by friends and acquaintances, someone is uploading pictures of a new line of dresses their business has introduced, someone is digitizing and uploading court judgments of the last many years, someone is sending out messages asking for references for a carpenter, and so on.
These are just four out of the 4.6 billion. The examples we have taken are probably innocuous transactions that keep happening all the time. We participate to the extent they connect to us, else mostly ignore them.
As we know, it takes all sorts to make the world. Over thousands of years of living together, we have developed a sort of structure for civil society and boundaries for acceptable as well as unacceptable behavior. Legal frameworks in many jurisdictions are also guided by these unwritten rules of civil society.
These norms of acceptable behavior, primarily developed through and for interaction, are also applicable to our conduct on the internet. Though it is not possible to cause physical harm through the internet, some of the content that exists can leave a deep, lasting and disturbing impression on the psyche of the consumer.
The reach for my few seconds of fame by creating or sharing content that will go viral, making me famous
The anonymity offered by the internet, where a user is sitting in a dark corner of the world, away from prying eyes (at least physical) and people, can affect different people in different ways. For some, it may provide comfort and solace, while for others it could be an encouragement for the mischief that they are not able to do in real life, believing they are secure, or for getting the (in)famy they believe they deserve.
Whether it is content about violence, or about child pornography, or about rape and murder, or about religious hatred, everything is possible. Once on a platform, it develops a life of its own and multiplies quickly, pushed by voyeurs as well as by people seeking to get their own one minute of fame by sharing ‘rare’ content.
Thankfully, the proportion of such content is very small, but it exists and needs to be discovered and eliminated. Leaving it loose on the internet is dangerous as it exposes not only adults, but even youngsters to objectionable material.
While the above may be examples of extreme content, there are many more commonplace examples of what might be called simple violations of website and community guidelines and practices. For example, a member of a Facebook group created for sharing the hand embroidery work they are doing, might put up some promotional post advertising the prices of her products. The owner may need to moderate that. An anti-religion message may be posted on a WhatsApp group created for sharing job opportunities and may need to be deleted by the administrator.
The need to manage such content gives rise to the need for website content moderation.
As a data serviced focused BPO company, oWorkers has been providing this service to its clients from around the world for over 8 years. With increase in social media usage, the need for moderation, as well as our support for clients, has grown. Our recognition as one of the top three data services providers amongst leading BPO outfits is the result of client support as much as it is a recognition of our staff.
What needs to be moderated?
That is the simple answer.
Once it has been established that objectionable content can be posted, it also needs to be understood that the objectionable content can lurk in corners or crevices of the world wide web. Hence, if user generated content (UGC) can be posted on the site, it needs to be moderated.
People and organizations are also creating sub-properties on the social media platforms to take advantage of the free, flowing interaction for the benefit of their brand. They create groups and communities with the objective of bringing together favorably disposed users, keep them interested, as well as reach out to fresher audiences and markets through them.
Content is being generated all the time in the four common formats: text, images, audio and video. The richer the content, the greater the opportunity for impact. Video, which contains visuals and audio, is the most impactful.
There is a huge variety to contend with. From articles and posts, to comments and reviews and descriptions. Language is the basic communication tool for humans and text is how it is represented. With each usage, the context also needs to be understood. In addition to keyworks, phrases and sentences also need to be evaluated.
The intelligence that computers have is based on what we might call ‘text.’ They have been taught to understand characters and sequences of characters in specific formats, that has come to be known as software code, based on which they are able to carry out their actions.
At this point, audio cannot be interpreted by machines. It needs to be converted to text through natural language processing (NLP) engines for machines to make sense of the. Hence, website content moderation becomes more complex for machines to handle. Of course, human beings can understand and interpret audio and moderate it.
An image is just an image for a computer. A set of pixels of different colors placed randomly in a frame. It does not mean much. Artificial Intelligence (AI) engines have been developed and trained to understand images with the help of millions of training data fed to them. Unlike audio, images cannot even be converted to text through an NLP process making them at least partially intelligible to machines to facilitate automated moderation.
Once again, with their intelligence, humans are able to look at an image, place it in context, and take a decision on its relevance and suitability.
Video is the richest form of content posted by users. It contains image sequences as well as audio, both unintelligible to machines in their original form. Videos could also have subtitles, making them even more complex. Machines have to be trained through AI using Machine Learning (ML) to enable them to get a handle on video content.
As with other forms of content, users can understand and analyze video content and determine its suitability for a particular platform. Even for humans, it is a difficult job to analyse video content as they may need to track the visual sequence in addition to the audio and subtitles, if present.
oWorkers’ ability to attract walk-in talent, which emanates from our positioning in our communities as a preferred employer, gives us, and our clients, access to the choicest talent. This talent is then polished with the help of our dedicated training team to deliver a resource ready to ‘hit the ground running’ whether for a website content moderation project or any other. An additional advantage this consistent flow walk-in candidates provides is that of comfortably handling short-term volume ramps. You can never guess when a particular topic or website goes viral can you? We can hire almost a hundred additional resources in 48 hours.
How should website content moderation be done?
Being part of civil society, we ought to be generally aware of the rules and regulations that one needs to abide by. However, each of us, and each website and community, operates within a specific context. While participants of such communities and websites ought to be aware, it is never a bad idea to set the context for the sake of clarity. Posting the image of a woman’s face in Germany may not even be noticed but could be considered offensive in a society in the Middle East.
Hence, laying down the rules and regulations for participation on a website or community can be considered to be the first step for moderation. People who participate in the community do so in the knowledge that such guidelines exist and need to be abided by. These could relate to language, acceptable subjects, external links, advertising, anything.
There should be consequences. If there are no consequences, setting up rules often becomes a meaningless exercise. While owners are not judicial authorities pronouncing judgment and sentences on offenders, they are free to take action within the context of website content moderation. These typically lead to censure, content editing, content removal, limiting access and, when all else fails, blocking user access completely from the community. These may be taken in sequence or may be based on the severity of the infraction.
It is also a good idea to set expectations upfront. If you have decided that you will follow a policy of moderation such that UGC will only be visible after it has been reviewed, let the users know. It will prevent needless angst while they wait for their content to show up. It is also a good reminder for users that not everything is acceptable.
In addition to trained human resources, oWorkers is able to access the latest technology tools suitable for this activity, thanks to its enduring partnership with leading providers of technology. As they are used for client projects, they also stand to benefit from these technologies and tools. Add to this our super secure facilities, GDPR compliant, ISO (27001:2013 & 9001:2015) certification, and you have a complete solution. In view of the pandemic, we are also optimized to work completely from home or office, as dictated by the evolving situation.
Section 230 of the Communications Decency Act (CDA) is what makes the internet a vibrant marketplace for interactions and exchange of ideas and information. It provides intermediary liability protection to websites both for content posted by their users that could be vile, as well as for removing content that could be vile in order to make their website safer for a larger set of people. In a way, it promotes freedom of speech and creates an opportunity for vile content to be created, at the same time allowing for content to be reviewed and judged, whether it is terrorism related or deep fakes or child pornography or graphic violence of drug trafficking.
With billions using the internet and astronomical amounts of content being created every second, human moderation, still considered to be the safest method, will never be enough, considering the timeliness of moderation as well. Thousands of moderators are reviewing content this minute trying to decide on the suitability of each piece they come across.
The human moderators, that we believe are still the best option for moderation, are living in their own private hell while reviewing content. Firstly, they have a fine balancing act to perform, between the commercial interest of their employer, it is after all a for-profit business, the demands of free speech and ensuring safety and security of the masses of people who will be exposed to the content. In doing so, they are exposing themselves to text after text, audio followed by audio, image upon image and video after video of horrific rapes and murders and child abuse and hate mongering. All of it is leaving an impression on their minds, little by little. One can never say what unintended consequences such exposure may lead to.
Eventually, it is a question of taking collective responsibility, though that may not solve the immediate issue of posting of vile content. But it does help us realize that what is happening on the internet is perhaps a reflection of what we really are. We cannot run away from it. There may not be any easy solutions but it is our problem and we have to handle it.
oWorkers understand the challenges and is equipped to handle them. Operating with employed staff, as opposed to contractors and freelancers employed by many competitors, we regularly monitor each individual’s performance as part of a larger career management framework. For staff needing training, or being affected by the nature of work, steps like training programs and job rotation are in place.
With three centers in three of the most sought after delivery locations in the world, oWorkers employs a multicultural team which enables it to offer website content moderation services in 22 languages.