“This post may contain misleading information” – an analysis of the use of warning labels on social media posts

-- By EstherStefanini - 22 Nov 2020

Introduction

Whilst I have been recently able to pry myself from using Instagram (I had abandoned Facebook several years ago), I have failed to reduce my Twitter usage. It acts as a messaging platform, source of entertainment for me and most importantly, my source for news and global affairs.

What do you think it would take to help you rebalance the media diet by getting actual news from actual news organizations? What tools or forms of presentation would reduce the dependence on Twitter for what highly sophisticated social structures have evolved to do better?

During the recent election frenzy, I found myself on the app a lot more than usual. However, I noticed contents warnings on political posts, particularly those posted by President Trump. The warning message: “this may contain misleading information” followed by a link to more sites with supposedly more impartial information appeared under quite a few of his posts. I was intrigued – since taking this course I had spent some time thinking about how social media has detrimentally changed society and its prowess to spread ‘false news’ is one such issue. Yet, it seems as if a solution had been found. Unfortunately, this is not the case. I will analyze how social media platforms have recently utilized such warning labels and demonstrate that they are merely a band-aid and not a cure.

Why Warning Labels are Needed

The success of Trump’s 2016 campaign has been partly attributed to its warming embrace of Facebook and Twitter as an advertising tool.

Mostly by them. I don't think the persistence of large TV advertising expenditures dwarfing the ad spending on social media suggests that the professionals who run that and other campaigns believe any very strong form of the proposition.

Most notably, thousands of these ads were affiliated with Russians, spreading exaggerated, and often false information. It is not just the US who has seen Facebook being used to erode democracy; the now defunct Cambridge Analytica used social media data to influence the elections in dozens of other nations including Kenya, the UK, Trinidad and Tobago and even triggered a genocide in Myanmar. Whilst political ad campaigning is normal and expected, the business model of social media (relying on shares and likes), the almost-addictive nature and the propensity of users to believe everything at face value, creates a platform in which users are quite easily influenced, even without realizing it. Facebook, Twitter and Instagram were pressured to acknowledge their role in circumventing electoral democracy in the 2016 elections and have adopted similar methods to address the spread of misinformation on their platforms.

When Did They Start?

In December 2016, Facebook announced the introduction of ‘disputed flags’ – red badges which appeared under articles after being checked by third-party fact checkers. This was later replaced a year later and instead of red flags, Facebook provided links under questionable posts to more reputable sites. The company argued that related articles were more effective than disputed flags in discouraging users to share false news. (although they failed to explain how they met this conclusion. I am inclined to think that the new initiative was more palatable to the right who complain that flagging censors a lot of their promotional media). As of September 2019, Facebook continues to address misinformed posts in this way, as well as removing fake accounts (bots) and utilising AI systems to recognise such content. However, politicians are exempt from their fact-checking program.

Instagram, a subsidiary of Facebook, marks false information in a similar way. Anyone can report a post that sounds suspicious, which will cause it to be checked by an independent party. Instagram states they “work with 45 third-party fact-checkers across the globe who are certified through the non-partisan International Fact-Checking Network to help identify, review and label false information”. Instagram’s labelling goes a step further than Facebook – marked posts are blurred out until users click ‘view post’ after they understand that it may contain false information. They also make it harder to find the post by hashtag search and it will not show up in the ‘Explore’ page.

Twitter caught up with the other major platforms this year, introducing labels and warning messages and affiliated links for users to find out more. Tweets mentioning COVID-19 particularly came with links to global and health information. Unlike Facebook, Twitter does flag tweets made by politicians and public officials which sparked my interest in the topic – dozens of tweets published by Trump have been marked as ‘disputed’ or ‘false’, including his recent claim that he “won the election”.

How Effective Are the Warnings?

A study conducted in January 2019 found that “the “Disputed” tag [on Facebook posts] reduced the mean proportion of respondents who accept a headline as “Somewhat accurate” or “Very accurate” when no general warning was provided from 29% in the baseline condition to 19%, a ten-percentage point decline”. A later study in July 2019 also concluded that the warning labels on Facebook posts had the effect of reducing the likelihood of fake news being shared – after being presented with a fabricated Facebook post accompanied with a warning label, 23% of respondents said they were generally likely to share the fabricated post. However, the comments sections below articles that discussed the introduction and efficacy of misinformation labels revealed quite a different sentiment. One Guardian reader believes that “fake news is a left wing boogie man scapegoat for their failures”. Another said, it is “pointless, no one reads the original article nevermind the related ones. The battle is won and lost with memes”.

Despite the attempts, it is clearly difficult to assess if the warning labels will effectively work and protect easily influenced minds and democracy generally. Instagram and Facebook have admitted that they cannot find and label all suspicious posts, many will fall under the radar due to the sheer number of content that is uploaded every day. Furthermore, the readers who are most likely to align themselves with the exaggerated right-wing propaganda are not likely to be deterred by a warning sign, as the comment above suggests.

Conclusion

Fake news existed before social media and it will continue to exist. In my opinion, the warning labels are an admirable addition but will make little to no difference to their spread and does not deal with any of the other problems of social media platforms. I think the real issue is that people (including me) are over-reliant on social media for keeping up with the news.

Hence my question at the top. If that's the real issue, let's see what we can learn about it by studying you.

I think the best route to the improvement of the present draft is to take a lawyer's view of the labeling behavior. What are the companies trying to achieve by way of reducing their legal and/or social liabilities? What legal measures are they trying to avoid? How can legal and technical measures be devised that accord with the political effort to reduce their distorting effects on the epistemic confidence that we are conducting democratic self-governance with due regard to the establishment of shared social "facts"?

---

You are entitled to restrict access to your paper if you want to. But we all derive immense benefit from reading one another's work, and I hope you won't feel the need unless the subject matter is personal and its disclosure would be harmful or undesirable. To restrict access to your paper simply delete the "#" character on the next two lines:

• #Set ALLOWTOPICVIEW = TWikiAdminGroup, EstherStefanini
• #Set DENYTOPICVIEW = TWikiGuest

Note: TWiki has strict formatting rules for preference declarations. Make sure you preserve the three spaces, asterisk, and extra space at the beginning of these lines. If you wish to give access to any other users simply add them to the comma separated ALLOWTOPICVIEW list.