Facebook and Instagram Censorship
How their community “fact-check” system works
I read an article by Deborah L. Armstrong earlier today, and she made the following statement:
“I’ve experienced censorship. I’ve been banned repeatedly from Facebook for posting articles, memes, photographs and other content deemed “false” by the so-called “fact-checkers” (the very existence of which ought to make the hair stand up on the necks of anyone who claims to care about free speech) […].”
First, let us address two significant points about censorship on these platforms. First, like Twitter, Meta has a backdoor for the US government to request blocking or shadow banning of specific posts and users. There are testimonies both for and against the government’s involvement in Facebook. However, it is a fact that the government can and does make requests of Twitter, Facebook, and the rest of the social media giants to remove articles and users. These actions are information warfare – removing what the government does not want people to see—and a not-so-clear violation of free speech on the part of the US government.
With that said, let’s look specifically at Facebook’s “fact-check” system. Facebook and Instagram use the same “community monitoring” services and hire outside companies to do “fact-checking.” These services do two things for Meta: Facebook and Instagram have plausible deniability in any lawsuit about removing certain content. In principle, it is a good measure of what is acceptable in a community.
“Fact-check” is a misnomer. The better term is “community acceptability monitoring.” The decisions on social media platforms are made by “social media evaluators” from companies such as Telus International and Appen. Another company that does “fact-checking” is Science Feedback, which claims to do fact-checking using knowledgeable scientists. Let us look at these two types of information checkers individually.
We’ll discuss this article on Episode 20 of Scenes from the Evolution.
“Science Feedback is paving the way for a new kind of fact-checking, crowdsourced directly from scientists with relevant expertise, providing in-depth analysis on a whole article, as well as “fact-checking” a number of claims at once.”
Science Feedback takes applications from published scientists with PhDs in climate sciences or health sciences. Because of this, one would think their recommendations would be unbiased and accurate. However, some of the most significant contributors to this company are people like Eric Michelman, a climate change activist, and the Reis Foundation. If you go down their list of contributors, you will find that they tend to be leftist supporters of the US Democratic party.
After assessing Science Feedback’s latest claim reviews, questions about their selection process for scientists and the impartial nature of their evaluations come up. The “scientific evaluations” by this company, and others, are then used by the social media evaluators of the first-named companies when evaluating the “truth” in social media posts.
“I asked all Science Feedback’s reviewers about their “Misleading” label. Two agreed to on-camera interviews. When I asked what was misleading about my video, they surprised me by saying that they hadn’t even watched my video! They offered no defense for posting words in quotation marks that I’d never said.” - John Stossel
Social media evaluators are people selected from Western-friendly countries to “rate the relevancy and accuracy of ads on social media.” You can find the full job description on Telus or see this article about working with Appen.
Who do these companies hire to evaluate social media posts, what are their qualifications, and what guidance are they given on evaluating social media posts?
As shown in the image above, the requirements are that a social media evaluator must have lived in the United States for the past three years, can work 1 hour a day for a minimum of five days a week, and have a cultural awareness of issues in the United States. Unfortunately, many evaluators are not located in the US but use a VPN while claiming residence and cultural knowledge to earn what is good money for their location.
The guidance they give evaluators is as follows:
“[…] you will be asked to search for evidence about the claim using an internet search engine […].”
Then, “[…] provide some information about the evidence you found.”
“You will have up to 15 minutes of research time for each item.”
If after 15 minutes you did not find satisfactory evidence, “you should still answer the task’s questions […].”
However, on average, the evaluator has only two minutes (one hour divided by 30 tasks) to complete each task. Spending more time on a job means the evaluator either earns less or the company will remove them from the project for “not performing to standards.”
The companies pay a small daily fee for any day the evaluator does at least one evaluation, plus a per-task fee. For example, if the evaluator finishes all thirty tasks daily within an hour, they earn USD 9 for the day (or USD 63 for a seven-day week).
The evaluator must decide if something is an opinion or a statement of fact and evaluate the context. Then, research the item, and include a link to evidence supporting their decision “from a source you know is trustworthy.” Each task asks seven mandatory and three optional questions.
The companies also have an “accuracy metric” that the evaluators must meet to remain on the project. The social media platform defines that metric, and no published guidelines exist. Past and current evaluators say they quickly learn the “right answers” to controversial content.
As you can see, we can question the evaluators’ location, cultural awareness, knowledge of what evidence may be “reliable,” and the time given to research and understand the supporting evidence properly.
Try it. Go to your favorite search engine and research the claim “Climate change causes forest fires.” Then, answer the ten questions in the guidance and specify a link to your evidence from a reliable source, all in under two minutes.