Rahim Talibzade[1] and Bishoy Maher[2] - written evidence (IRN0015)

What role should users and online platforms play in establishing and maintaining online community standards for combating Fake News and malicious behaviour?

1.       Problem Definition:

The proliferation of fake news in everyday media outlets such as social media feeds, blogs, and online newspapers have made it challenging to identify trustworthy news sources. This global problem has exposed the vulnerability of individuals, institutions, and society to manipulations by malicious actors. “Fake news” has many definitions; however, The European Commission defines it as “intentional disinformation spread via online social platforms, broadcast news media or traditional print.”  It is fabricated information that mimics news media content in form but not intent.

Fake news can be classified into two main categories, these are:

1. Misinformation: Information that is false or misleading regardless of intent
2. Disinformation: Information that is purposely false, or misleading, spread with intent to deceive

Fake news takes on many forms including most prominently:

1.       Factually incorrect news articles or blog posts

2.       Parodies, Hoaxes, Fabricated Audio Visuals, and Memes

3.       Factually inaccurate statements or reports by public figures

Studies have shown the wide array of biases present within people, which makes them susceptible to these kinds of manipulations. Research has confirmed that people prefer information that validates their pre-existing attitudes – known as selective exposure. Furthermore, people view information that coincides with their ideologies and beliefs to be more persuasive than discordant information – known as confirmation bias. Lastly, people are inclined to believe more strongly in information that pleases them – known as the desirability bias. These online platforms’ business models are built around targeted-advertisements, therefore are purpose built to maximize user engagement. This often leads to social media platforms trapping their users in echo chambers, whereby the content they are served is tailored to their biases. This has in turn opened an opportunity for malicious, politically-motivated outlets to spread disinformation.

The fact that tech companies such as Facebook and Google have appropriated – and monopolised – the online advertising market has led to a pay-as-you-go business model, in which advertisers are only charged when a page is viewed or clicked on. This means that users are constantly exposed to an increasingly large collection of unregulated media content and ensures that social media companies have no incentive to play the role of “arbiters of truth.” Online platforms must take responsibility, if not of the information itself, then of informing their users and making efforts to discern between types of posted content. A new system of safeguards is clearly necessary.

2.       Current approaches

Current approaches include the use of state actors such as the Disinformation Review Office set up by the European Union; a network of experts, journalists, officials, NGOs, and others all collaborating to report disinformation content to EU officials. However, the review process is vague at best and further lacks clarity as to how actors are recruited to be a part of these “trusted” entities. This raises public concern on whether this approach may lead to infractions on the right to freedom of speech. Furthermore, the ability of a few hundred or even thousand members of this organization to classify millions of pieces of content posted daily remains a crux to this approach.

This has led to the use of Artificial Intelligence (AI) algorithms to automate the classification of content as fake or not; however, this approach remains a double-edged sword. On the one hand, algorithms are incredibly agile and can deal with the classification of enormous amounts of data with ease. However, the algorithm must have a clear objective function - a parameter-defined task that helps it evolve. Upon solving this task, the algorithm gains experience. It then uses that experience to make slight modifications to the previous iteration's parameters in order to improve the outcome of the following iteration. This self-improving function relies heavily on the input dataset to the algorithm. That is what defines the “guidelines” upon which the algorithm acts and ultimately dictates the output classification of the algorithm. Therefore, if the “guidelines” are ambiguous the output can be manipulated.

This has indeed happened already; whereby malicious actors are capable of introducing their own bots to online platforms to spread targeted disinformation that takes advantage of the aforementioned biases (Cambridge Analytica) – the outcome of which is the voluntary participation of people in further spreading disinformation and ultimately the promotion of spread by the algorithm as it perceives the content as legitimate.

3.       Proposal:

As seen above, the online content is difficult to regulate. This report submits that users should play a semi-formal role in maintaining certain standards and expectations of the news-labelled content. The online platforms should play an essential role in facilitating that. The suggestion of adding a small feature to classify something as News-related, an opportunity to put up the source and an ability to tag something as “Fake News” will be exemplified below.

Mechanism

An online platform implements an additional feature of content-tagging that is available once a user chooses to post something. This is aimed at users who tend to publish something that may come across as news. The feature would enable those who post to fill in a small pop-up form by ticking the boxes “News-related”, “Personal”, or create custom categories.

On the next line “Source”, the users are asked to indicate the source where they have the information from. This could have options of pasting a hyperlink, writing “self-reported”, referring to another reliable source, or mentioning a political expert who spoke of such an issue, and so on. This allows the users to make judgement of the reliability of the content even if it conforms with their biases.

When showing across newsfeed, it will clearly show something as “News-related” so to bring awareness that this user chose to classify their material as news. This special category would allow the separation of an opinion on a political situation and a descriptive account of the events.

To allow the online platform to retain user-friendly interface, such pop-up form will not be mandatory, and will be optional.  Users who aim to come across as “News” would voluntarily fill in that form to indicate the accuracy of the source.

The other users, who are reading it, will be able to classify something as “Fake News”, or add custom tags such as “Opinion” by clicking on the options next to the post. This wouldn’t mean the content is deleted, as other users will be able to see that this post has a lot of Fake News tags on it and who voted for it. The online platform itself is not involved in the process of deleting or regulating such content regardless of the downvotes.

4.       Existing analogy:

A popular online community Reddit has a similar score-based system[3], where people can upvote or downvote the posts. It then affects which content will be put at the top based on an amount of upvotes. In this suggestion however, the content will remain regardless of the amount of downvotes. This is to ensure that there is no self-built echo chamber[4]- a place where individuals are surrounded only by the content that people ideologically identical to them endorse.

5.       What it achieves:

A small feature to classify something as News-related, an opportunity to put up the source and an ability to tag something as “Fake News” helps achieve the following.

Online Communities

The online platform is not burdened by any potential liabilities. The suggestion is entirely user-centred. The online platforms include social media, forums, and other online communities.

The community regulation and additional tagging should curb the echo chambers, that currently amplify the fake news on social media, by promoting responsible content-sharing.

This helps complement current attempts to root out the fake news by the algorithms that are currently developed for social platforms.

The interface won’t be heavily affected, as ticking a box and a few more lines on sources will be a good reminder for people to be adequate when posting news-related content.

Fake bot farms’ influence will be significantly reduced, as now they have to publish a source. The accountability of content sharers and creators will be subject to a user-friendly standard.

End-users

People will be more aware of the content they read, and form judgement as to how reliable a source is, where it comes from, etc. For instance, if someone writes a post and may include sources to support that, people would be able to see that this person has read sites like BBC.com, or RussiaToday. This helps raise informal, self-governing standards that are set by the community through the online platform’s efforts.

This would complement the previously written[5] Anti-Fake News strategies that people could use to think about the source itself, now that they are able to check it. Upon seeing something tagged as “Fake News”, users will be careful when reading such content. They are less likely to be swayed, if they see the source as lacking validity.

An absence of any particular source could suggest a person is spreading disinformation or is writing a personal opinion post. An online community is able to discern and respond by classifying it themselves. A user will now be aware that he has to come across as reliable.

8 May 2018

1